Renaming global spec vars. Replacing L2Banks with total_l2_chan (#251)

Signed-off-by: colramos-amd <colramos@amd.com>


[ROCm/rocprofiler-compute commit: 86b6c8e848]

projects/rocprofiler-compute/src/omniperf_analyze/analysis_base.py

@@ -135,7 +135,7 @@ class OmniAnalyze_Base:
         # load required configs
         for d in self.__args.path:
             sys_info = file_io.load_sys_info(Path(d[0], "sysinfo.csv"))
-            arch = sys_info.iloc[0]["arch"]
+            arch = sys_info.iloc[0]["gpu_arch"]
             args = self.__args
             self.generate_configs(
                 arch,
@@ -155,7 +155,7 @@ class OmniAnalyze_Base:
                     w.sys_info, self.__args.specs_correction
                 )
             w.avail_ips = w.sys_info["ip_blocks"].item().split("|")
-            arch = w.sys_info.iloc[0]["arch"]
+            arch = w.sys_info.iloc[0]["gpu_arch"]
             w.dfs = copy.deepcopy(self._arch_configs[arch].dfs)
             w.dfs_type = self._arch_configs[arch].dfs_type
             self._runs[d[0]] = w

projects/rocprofiler-compute/src/omniperf_analyze/analysis_cli.py

@@ -70,13 +70,13 @@ class cli_analysis(OmniAnalyze_Base):
             tty.show_kernel_stats(
                 self.get_args(),
                 self._runs,
-                self._arch_configs[self._runs[self.get_args().path[0][0]].sys_info.iloc[0]["arch"]],
+                self._arch_configs[self._runs[self.get_args().path[0][0]].sys_info.iloc[0]["gpu_arch"]],
                 self._output
             )
         else:
             tty.show_all(
                 self.get_args(),
                 self._runs,
-                self._arch_configs[self._runs[self.get_args().path[0][0]].sys_info.iloc[0]["arch"]],
+                self._arch_configs[self._runs[self.get_args().path[0][0]].sys_info.iloc[0]["gpu_arch"]],
                 self._output
             )

projects/rocprofiler-compute/src/omniperf_base.py

@@ -159,10 +159,12 @@ class Omniperf:
     def load_soc_specs(self, sysinfo=None):
         """Load OmniSoC instance for Omniperf run
         """
-        if not sysinfo is None:
-            self.__mspec = MachineSpecs(self.__args, sysinfo)
+        self.__mspec = MachineSpecs(self.__args, sysinfo)
+        if self.__args.specs:
+            print(self.__mspec)
+            sys.exit(0)
 
-        arch = self.__mspec.arch
+        arch = self.__mspec.gpu_arch
 
         # NB: This checker is a bit redundent. We already check this in specs module
         if arch not in self.__supported_archs.keys():
@@ -188,10 +190,10 @@ class Omniperf:
         )
         self.__args = parser.parse_args()
 
-        if self.__args.specs:
-            print(MachineSpecs(self.__args))
-            sys.exit(0)
         if self.__args.mode == None:
+            if self.__args.specs:
+                print(MachineSpecs(self.__args))
+                sys.exit(0)
             parser.print_help(sys.stderr)
             error("Omniperf requires a valid mode.")
 
@@ -204,20 +206,20 @@ class Omniperf:
 
         # Update default path
         if self.__args.path == os.path.join(os.getcwd(), "workloads"):
-            self.__args.path = os.path.join(self.__args.path, self.__args.name, self.__mspec.GPU)
+            self.__args.path = os.path.join(self.__args.path, self.__args.name, self.__mspec.gpu_model)
 
         logging.info("Profiler choice = %s" % self.__profiler_mode)
 
         # instantiate desired profiler
         if self.__profiler_mode == "rocprofv1":
             from omniperf_profile.profiler_rocprof_v1 import rocprof_v1_profiler
-            profiler = rocprof_v1_profiler(self.__args, self.__profiler_mode, self.__soc[self.__mspec.arch])
+            profiler = rocprof_v1_profiler(self.__args, self.__profiler_mode, self.__soc[self.__mspec.gpu_arch])
         elif self.__profiler_mode == "rocprofv2":
             from omniperf_profile.profiler_rocprof_v2 import rocprof_v2_profiler
-            profiler = rocprof_v2_profiler(self.__args, self.__profiler_mode, self.__soc[self.__mspec.arch])        
+            profiler = rocprof_v2_profiler(self.__args, self.__profiler_mode, self.__soc[self.__mspec.gpu_arch])        
         elif self.__profiler_mode == "rocscope":
             from omniperf_profile.profiler_rocscope import rocscope_profiler
-            profiler = rocscope_profiler(self.__args, self.__profiler_mode, self.__soc[self.__mspec.arch])
+            profiler = rocscope_profiler(self.__args, self.__profiler_mode, self.__soc[self.__mspec.gpu_arch])
         else:
             logging.error("Unsupported profiler")
             sys.exit(1)
@@ -225,11 +227,11 @@ class Omniperf:
         # -----------------------
         # run profiling workflow
         #-----------------------
-        self.__soc[self.__mspec.arch].profiling_setup()
+        self.__soc[self.__mspec.gpu_arch].profiling_setup()
         profiler.pre_processing()
         profiler.run_profiling(self.__version["ver"], config.prog)
         profiler.post_processing()
-        self.__soc[self.__mspec.arch].post_profiling()
+        self.__soc[self.__mspec.gpu_arch].post_profiling()
 
         return
 

projects/rocprofiler-compute/src/omniperf_profile/profiler_base.py

@@ -312,7 +312,7 @@ class OmniProfiler_Base:
         # log basic info
         logging.info(str(prog) + " ver: " + str(version))
         logging.info("Path: " + str(os.path.abspath(self.__args.path)))
-        logging.info("Target: " + str(self._soc._mspec.GPU))
+        logging.info("Target: " + str(self._soc._mspec.gpu_model))
         logging.info("Command: " + str(self.__args.remaining))
         logging.info("Kernel Selection: " + str(self.__args.kernel))
         logging.info("Dispatch Selection: " + str(self.__args.dispatch))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/0200_system-speed-of-light.yaml

@@ -23,62 +23,62 @@ Panel Config:
           VALU FLOPs:
             value: None # No perf counter
             unit: GFLOPs
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: None # No perf counter
             tips:
           VALU IOPs:
             value: None # No perf counter
             unit: GIOPs
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (BF16):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 512) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 512) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (F16):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (F32):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (F64):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: None # No perf counter
             tips:
           MFMA IOPs (Int8):
             value: None # No perf counter
             unit: GOPs
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: None # No perf counter
             tips:
           Active CUs:
             value: $numActiveCUs
             unit: CUs
-            peak: $numCU
-            pop: ((100 * $numActiveCUs) / $numCU)
+            peak: $cu_per_gpu
+            pop: ((100 * $numActiveCUs) / $cu_per_gpu)
             tips: 
           SALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           VALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips:
           MFMA Utilization:
             value: None # No HW module
@@ -115,18 +115,18 @@ Panel Config:
           Wavefront Occupancy:
             value: AVG((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE))
             unit: Wavefronts
-            peak: ($maxWavesPerCU * $numCU)
-            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($maxWavesPerCU
-              * $numCU))))
+            peak: ($max_waves_per_cu * $cu_per_gpu)
+            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($max_waves_per_cu
+              * $cu_per_gpu))))
             coll_level: SQ_LEVEL_WAVES
             tips: 
           Theoretical LDS Bandwidth:
-            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: (($sclk * $numCU) * 0.128)
-            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            peak: (($max_sclk * $cu_per_gpu) * 0.128)
+            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             tips: 
           LDS Bank Conflicts/Access:
             value: AVG(((SQ_LDS_BANK_CONFLICT / (SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT))
@@ -151,9 +151,9 @@ Panel Config:
           vL1D Cache BW:
             value: AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numCU)
+            peak: ((($max_sclk / 1000) * 64) * $cu_per_gpu)
             pop: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             tips: 
           L2 Cache Hit Rate:
             value: AVG((((100 * TCC_HIT_sum) / (TCC_HIT_sum + TCC_MISS_sum)) if ((TCC_HIT_sum
@@ -166,25 +166,25 @@ Panel Config:
           L2 Cache BW:
             value: AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * TO_INT($L2Banks))
+            peak: ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan))
             pop: ((100 * AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * TO_INT($L2Banks)))
+              / ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan)))
             tips: 
           L2-Fabric Read BW:
             value: AVG((((TCC_EA_RDREQ_32B_sum * 32) + ((TCC_EA_RDREQ_sum - TCC_EA_RDREQ_32B_sum)
               * 64)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA_RDREQ_32B_sum * 32) + ((TCC_EA_RDREQ_sum - TCC_EA_RDREQ_32B_sum)
-              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Write BW:
             value: AVG((((TCC_EA_WRREQ_64B_sum * 64) + ((TCC_EA_WRREQ_sum - TCC_EA_WRREQ_64B_sum)
               * 32)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA_WRREQ_64B_sum * 64) + ((TCC_EA_WRREQ_sum - TCC_EA_WRREQ_64B_sum)
-              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Read Latency:
             value: AVG(((TCC_EA_RDREQ_LEVEL_sum / TCC_EA_RDREQ_sum) if (TCC_EA_RDREQ_sum
@@ -211,9 +211,9 @@ Panel Config:
           sL1D Cache BW:
             value: AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Hit Rate:
             value: AVG(((100 * SQC_ICACHE_HITS) / (SQC_ICACHE_HITS + SQC_ICACHE_MISSES)))
@@ -224,9 +224,9 @@ Panel Config:
           L1I BW:
             value: AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 32))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 32))) / ((($sclk
-              / 1000) * 32) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 32))) / ((($max_sclk
+              / 1000) * 32) * $sqc_per_gpu))
             tips: 
           L1I Fetch Latency:
             value: AVG((SQ_ACCUM_PREV_HIRES / SQ_IFETCH))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/0300_mem_chart.yaml

@@ -74,7 +74,7 @@ Panel Config:
             tips:
           Num CUs:
             #alias: num_cu_
-            value: $numCU
+            value: $cu_per_gpu
             tips:
           VGPR:
             #alias: vgpr_
@@ -110,7 +110,7 @@ Panel Config:
           LDS Util:
             #alias: lds_util_
             value:
-              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU))),
+              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu))),
               0)
             tips:
           LDS Latency:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/0600_shader-processor-input.yaml

@@ -26,9 +26,9 @@ Panel Config:
             unit: Pct
             tips: 
           Scheduler-Pipe Utilization:
-            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
+            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
             unit: Pct
             tips: 
           Workgroup Manager Utilization:
@@ -38,15 +38,15 @@ Panel Config:
             unit: Pct
             tips: 
           Shader Engine Utilization:
-            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
+            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
             unit: Pct
             tips: 
           SIMD Utilization:
-            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Dispatched Workgroups:
@@ -91,77 +91,77 @@ Panel Config:
           tips: Tips
         metric:
           Not-scheduled Rate (Workgroup Manager):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Not-scheduled Rate (Scheduler-Pipe):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Scheduler-Pipe Stall Rate:
-            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
             unit: Pct
             tips: 
           Scratch Stall Rate:
-            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
+            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
             unit: Pct
             tips: 
           Insufficient SIMD Waveslots:
-            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD VGPRs:
-            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD SGPRs:
-            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU LDS:
-            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU Barriers:
-            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Workgroup Limit:
-            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Wavefront Limit:
-            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1100_compute-unit-compute-pipeline.yaml

@@ -92,15 +92,15 @@ Panel Config:
             unit: Instr/cycle
             tips: 
           SALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VMEM Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1200_lds.yaml

@@ -18,16 +18,16 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips: 
           Access Rate:
-            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips: 
           Theoretical Bandwidth (% of Peak):
-            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             unit: Pct of Peak
             tips: 
           Bank Conflict Rate:
@@ -56,11 +56,11 @@ Panel Config:
             unit: (Instr  + $normUnit)
             tips: 
           Theoretical Bandwidth:
-            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
             unit: (Bytes  + $normUnit)
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1300_instruction-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_ICACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_ICACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips: 
@@ -28,7 +28,7 @@ Panel Config:
             unit: Pct of Peak
             tips: 
           L1I-L2 Bandwidth:
-            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($sclk * $numSQC) 
+            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($max_sclk * $sqc_per_gpu) 
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1400_constant-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_DCACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_DCACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -29,7 +29,7 @@ Panel Config:
             tips:
           sL1D-L2 BW:
             value: AVG(((SQC_TC_DATA_READ_REQ + SQC_TC_DATA_WRITE_REQ + SQC_TC_DATA_ATOMIC_REQ) * 100000)
-                        / (2 * ($sclk * $numSQC) * (End_Timestamp - Start_Timestamp)))
+                        / (2 * ($max_sclk * $sqc_per_gpu) * (End_Timestamp - Start_Timestamp)))
             unit: Pct of Peak
             tips:
         comparable: false # for now

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1500_TA_and_TD.yaml

@@ -20,27 +20,27 @@ Panel Config:
           tips: Tips
         metric:
           Address Processing Unit Busy:
-            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data Stall:
-            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data-Processor → Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Total Instructions:
@@ -128,15 +128,15 @@ Panel Config:
           tips: Tips
         metric:
           Data-Return Busy:
-            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Cache RAM → Data-Return Stall:
-            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Workgroup manager → Data-Return Stall:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1600_L1_cache.yaml

@@ -26,7 +26,7 @@ Panel Config:
             tips: 
           Bandwidth:
             value: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             unit: Pct of Peak
             tips: 
           Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx906/1700_L2_cache.yaml

@@ -18,11 +18,11 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($L2Banks) * GRBM_GUI_ACTIVE)))
+            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($total_l2_chan) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips:
           Bandwidth:
-            value: ((100 * AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp)))) / ((($sclk / 1000) * 64) * TO_INT($L2Banks)))
+            value: ((100 * AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp)))) / ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan)))
             unit: pct
             tips: 
           Hit Rate:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/0200_system-speed-of-light.yaml

@@ -23,62 +23,62 @@ Panel Config:
           VALU FLOPs:
             value: None # No perf counter
             unit: GFLOPs
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: None # No perf counter
             tips:
           VALU IOPs:
             value: None # No perf counter
             unit: GIOPs
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (BF16):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 512) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 512) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (F16):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (F32):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: None # No perf counter
             tips:
           MFMA FLOPs (F64):
             value: None # No perf counter
             unit: GFLOPs
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: None # No perf counter
             tips:
           MFMA IOPs (Int8):
             value: None # No perf counter
             unit: GOPs
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: None # No perf counter
             tips:
           Active CUs:
             value: $numActiveCUs
             unit: CUs
-            peak: $numCU
-            pop: ((100 * $numActiveCUs) / $numCU)
+            peak: $cu_per_gpu
+            pop: ((100 * $numActiveCUs) / $cu_per_gpu)
             tips: 
           SALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           VALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips:
           MFMA Utilization:
             value: None # No HW module
@@ -115,18 +115,18 @@ Panel Config:
           Wavefront Occupancy:
             value: AVG((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE))
             unit: Wavefronts
-            peak: ($maxWavesPerCU * $numCU)
-            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($maxWavesPerCU
-              * $numCU))))
+            peak: ($max_waves_per_cu * $cu_per_gpu)
+            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($max_waves_per_cu
+              * $cu_per_gpu))))
             coll_level: SQ_LEVEL_WAVES
             tips: 
           Theoretical LDS Bandwidth:
-            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: (($sclk * $numCU) * 0.128)
-            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            peak: (($max_sclk * $cu_per_gpu) * 0.128)
+            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             tips: 
           LDS Bank Conflicts/Access:
             value: AVG(((SQ_LDS_BANK_CONFLICT / (SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT))
@@ -151,9 +151,9 @@ Panel Config:
           vL1D Cache BW:
             value: AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numCU)
+            peak: ((($max_sclk / 1000) * 64) * $cu_per_gpu)
             pop: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             tips: 
           L2 Cache Hit Rate:
             value: AVG((((100 * TCC_HIT_sum) / (TCC_HIT_sum + TCC_MISS_sum)) if ((TCC_HIT_sum
@@ -166,25 +166,25 @@ Panel Config:
           L2 Cache BW:
             value: AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * TO_INT($L2Banks))
+            peak: ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan))
             pop: ((100 * AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * TO_INT($L2Banks)))
+              / ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan)))
             tips: 
           L2-Fabric Read BW:
             value: AVG((((TCC_EA_RDREQ_32B_sum * 32) + ((TCC_EA_RDREQ_sum - TCC_EA_RDREQ_32B_sum)
               * 64)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA_RDREQ_32B_sum * 32) + ((TCC_EA_RDREQ_sum - TCC_EA_RDREQ_32B_sum)
-              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Write BW:
             value: AVG((((TCC_EA_WRREQ_64B_sum * 64) + ((TCC_EA_WRREQ_sum - TCC_EA_WRREQ_64B_sum)
               * 32)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA_WRREQ_64B_sum * 64) + ((TCC_EA_WRREQ_sum - TCC_EA_WRREQ_64B_sum)
-              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Read Latency:
             value: AVG(((TCC_EA_RDREQ_LEVEL_sum / TCC_EA_RDREQ_sum) if (TCC_EA_RDREQ_sum
@@ -211,9 +211,9 @@ Panel Config:
           sL1D Cache BW:
             value: AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Hit Rate:
             value: AVG(((100 * SQC_ICACHE_HITS) / (SQC_ICACHE_HITS + SQC_ICACHE_MISSES)))
@@ -224,9 +224,9 @@ Panel Config:
           L1I BW:
             value: AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips:
           L1I Fetch Latency:
             value: AVG((SQ_ACCUM_PREV_HIRES / SQ_IFETCH))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/0300_mem_chart.yaml

@@ -74,7 +74,7 @@ Panel Config:
             tips:
           Num CUs:
             #alias: num_cu_
-            value: $numCU
+            value: $cu_per_gpu
             tips:
           VGPR:
             #alias: vgpr_
@@ -110,7 +110,7 @@ Panel Config:
           LDS Util:
             #alias: lds_util_
             value:
-              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU))),
+              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu))),
               0)
             tips:
           LDS Latency:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/0600_shader-processor-input.yaml

@@ -26,9 +26,9 @@ Panel Config:
             unit: Pct
             tips: 
           Scheduler-Pipe Utilization:
-            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
+            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
             unit: Pct
             tips: 
           Workgroup Manager Utilization:
@@ -38,15 +38,15 @@ Panel Config:
             unit: Pct
             tips: 
           Shader Engine Utilization:
-            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
+            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
             unit: Pct
             tips: 
           SIMD Utilization:
-            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Dispatched Workgroups:
@@ -91,77 +91,77 @@ Panel Config:
           tips: Tips
         metric:
           Not-scheduled Rate (Workgroup Manager):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Not-scheduled Rate (Scheduler-Pipe):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Scheduler-Pipe Stall Rate:
-            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
             unit: Pct
             tips: 
           Scratch Stall Rate:
-            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
+            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
             unit: Pct
             tips: 
           Insufficient SIMD Waveslots:
-            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD VGPRs:
-            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD SGPRs:
-            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU LDS:
-            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU Barriers:
-            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Workgroup Limit:
-            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Wavefront Limit:
-            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1100_compute-unit-compute-pipeline.yaml

@@ -92,15 +92,15 @@ Panel Config:
             unit: Instr/cycle
             tips: 
           SALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VMEM Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1200_lds.yaml

@@ -18,17 +18,17 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips: 
           Access Rate:
-            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips: 
             unit: pct
           Theoretical Bandwidth:
-            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             unit: Pct of Peak
             tips:
           Bank Conflict Rate:
@@ -57,11 +57,11 @@ Panel Config:
             unit: (Instr  + $normUnit)
             tips: 
           Theoretical Bandwidth:
-            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
             unit: (Bytes  + $normUnit)
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1300_instruction-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_ICACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_ICACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips: 
@@ -28,7 +28,7 @@ Panel Config:
             unit: Pct of Peak
             tips: 
           L1I-L2 Bandwidth:
-            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($sclk * $numSQC) 
+            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($max_sclk * $sqc_per_gpu) 
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1400_constant-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_DCACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_DCACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -29,7 +29,7 @@ Panel Config:
             tips:
           sL1D-L2 BW:
             value: AVG(((SQC_TC_DATA_READ_REQ + SQC_TC_DATA_WRITE_REQ + SQC_TC_DATA_ATOMIC_REQ) * 100000)
-                        / (2 * ($sclk * $numSQC) * (End_Timestamp - Start_Timestamp)))
+                        / (2 * ($max_sclk * $sqc_per_gpu) * (End_Timestamp - Start_Timestamp)))
             unit: Pct of Peak
             tips:
         comparable: false # for now

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1500_TA_and_TD.yaml

@@ -20,27 +20,27 @@ Panel Config:
           tips: Tips
         metric:
           Address Processing Unit Busy:
-            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data Stall:
-            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data-Processor → Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Total Instructions:
@@ -128,15 +128,15 @@ Panel Config:
           tips: Tips
         metric:
           Data-Return Busy:
-            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Cache RAM → Data-Return Stall:
-            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Workgroup manager → Data-Return Stall:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1600_L1_cache.yaml

@@ -26,7 +26,7 @@ Panel Config:
             tips: 
           Bandwidth:
             value: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             unit: Pct of Peak
             tips: 
           Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx908/1700_L2_cache.yaml

@@ -18,11 +18,11 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($L2Banks) * GRBM_GUI_ACTIVE)))
+            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($total_l2_chan) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips:
           Bandwidth:
-            value: ((100 * AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp)))) / ((($sclk / 1000) * 64) * TO_INT($L2Banks)))
+            value: ((100 * AVG(((TCC_REQ_sum * 64) / (End_Timestamp - Start_Timestamp)))) / ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan)))
             unit: pct
             tips: 
           Hit Rate:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/0200_system-speed-of-light.yaml

@@ -27,93 +27,93 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (Int8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           Active CUs:
             value: $numActiveCUs
             unit: CUs
-            peak: $numCU
-            pop: ((100 * $numActiveCUs) / $numCU)
+            peak: $cu_per_gpu
+            pop: ((100 * $numActiveCUs) / $cu_per_gpu)
             tips: 
           SALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           VALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           MFMA Utilization:
-            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             tips: 
           VMEM Utilization:
-            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           Branch Utilization:
-            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           VALU Active Threads:
             value: AVG(((SQ_THREAD_CYCLES_VALU / SQ_ACTIVE_INST_VALU) if (SQ_ACTIVE_INST_VALU
@@ -132,18 +132,18 @@ Panel Config:
           Wavefront Occupancy:
             value: AVG((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE))
             unit: Wavefronts
-            peak: ($maxWavesPerCU * $numCU)
-            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($maxWavesPerCU
-              * $numCU))))
+            peak: ($max_waves_per_cu * $cu_per_gpu)
+            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($max_waves_per_cu
+              * $cu_per_gpu))))
             coll_level: SQ_LEVEL_WAVES
             tips: 
           Theoretical LDS Bandwidth:
-            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: (($sclk * $numCU) * 0.128)
-            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            peak: (($max_sclk * $cu_per_gpu) * 0.128)
+            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             tips: 
           LDS Bank Conflicts/Access:
             value: AVG(((SQ_LDS_BANK_CONFLICT / (SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT))
@@ -168,9 +168,9 @@ Panel Config:
           vL1D Cache BW:
             value: AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numCU)
+            peak: ((($max_sclk / 1000) * 64) * $cu_per_gpu)
             pop: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             tips: 
           L2 Cache Hit Rate:
             value: AVG((((100 * TCC_HIT_sum) / (TCC_HIT_sum + TCC_MISS_sum)) if ((TCC_HIT_sum
@@ -183,25 +183,25 @@ Panel Config:
           L2 Cache BW:
             value: AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 128) * TO_INT($L2Banks))
+            peak: ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan))
             pop: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+              / ((($max_sclk / 1000) * 64) * TO_INT($total_l2_chan)))
             tips: 
           L2-Fabric Read BW:
             value: AVG((((TCC_EA_RDREQ_32B_sum * 32) + ((TCC_EA_RDREQ_sum - TCC_EA_RDREQ_32B_sum)
               * 64)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA_RDREQ_32B_sum * 32) + ((TCC_EA_RDREQ_sum - TCC_EA_RDREQ_32B_sum)
-              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Write BW:
             value: AVG((((TCC_EA_WRREQ_64B_sum * 64) + ((TCC_EA_WRREQ_sum - TCC_EA_WRREQ_64B_sum)
               * 32)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA_WRREQ_64B_sum * 64) + ((TCC_EA_WRREQ_sum - TCC_EA_WRREQ_64B_sum)
-              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Read Latency:
             value: AVG(((TCC_EA_RDREQ_LEVEL_sum / TCC_EA_RDREQ_sum) if (TCC_EA_RDREQ_sum
@@ -228,9 +228,9 @@ Panel Config:
           sL1D Cache BW:
             value: AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Hit Rate:
             value: AVG(((100 * SQC_ICACHE_HITS) / (SQC_ICACHE_HITS + SQC_ICACHE_MISSES)))
@@ -241,9 +241,9 @@ Panel Config:
           L1I BW:
             value: AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Fetch Latency:
             value: AVG((SQ_ACCUM_PREV_HIRES / SQ_IFETCH))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/0300_mem_chart.yaml

@@ -74,7 +74,7 @@ Panel Config:
             tips:
           Num CUs:
             #alias: num_cu_
-            value: $numCU
+            value: $cu_per_gpu
             tips:
           VGPR:
             #alias: vgpr_
@@ -111,7 +111,7 @@ Panel Config:
           LDS Util:
             #alias: lds_util_
             value:
-              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU))),
+              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu))),
               0)
             tips:
           LDS Latency:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/0600_shader-processor-input.yaml

@@ -26,9 +26,9 @@ Panel Config:
             unit: Pct
             tips: 
           Scheduler-Pipe Utilization:
-            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
+            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
             unit: Pct
             tips: 
           Workgroup Manager Utilization:
@@ -38,15 +38,15 @@ Panel Config:
             unit: Pct
             tips: 
           Shader Engine Utilization:
-            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
+            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
             unit: Pct
             tips: 
           SIMD Utilization:
-            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Dispatched Workgroups:
@@ -91,77 +91,77 @@ Panel Config:
           tips: Tips
         metric:
           Not-scheduled Rate (Workgroup Manager):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Not-scheduled Rate (Scheduler-Pipe):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Scheduler-Pipe Stall Rate:
-            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
             unit: Pct
             tips: 
           Scratch Stall Rate:
-            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
+            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
             unit: Pct
             tips: 
           Insufficient SIMD Waveslots:
-            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD VGPRs:
-            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD SGPRs:
-            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU LDS:
-            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU Barriers:
-            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Workgroup Limit:
-            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Wavefront Limit:
-            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1100_compute-unit-compute-pipeline.yaml

@@ -26,55 +26,55 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (INT8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
 
     - metric_table:
@@ -107,27 +107,27 @@ Panel Config:
             unit: Instr/cycle
             tips: 
           SALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VMEM Utilization:
-            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           Branch Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Active Threads:
@@ -140,9 +140,9 @@ Panel Config:
             unit: Threads
             tips: 
           MFMA Utilization:
-            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
+            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips: 
           MFMA Instr Cycles:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1200_lds.yaml

@@ -18,16 +18,16 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips: 
           Access Rate:
-            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips: 
           Theoretical Bandwidth:
-            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             unit: Pct of Peak
             tips: 
           Bank Conflict Rate:
@@ -56,11 +56,11 @@ Panel Config:
             unit: (Instr  + $normUnit)
             tips: 
           Theoretical Bandwidth:
-            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
             unit: (Bytes  + $normUnit)
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1300_instruction-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_ICACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_ICACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips: 
@@ -28,7 +28,7 @@ Panel Config:
             unit: Pct of Peak
             tips: 
           L1I-L2 Bandwidth:
-            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($sclk * $numSQC) 
+            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($max_sclk * $sqc_per_gpu) 
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1400_constant-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_DCACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_DCACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -29,7 +29,7 @@ Panel Config:
             tips:
           sL1D-L2 BW:
             value: AVG(((SQC_TC_DATA_READ_REQ + SQC_TC_DATA_WRITE_REQ + SQC_TC_DATA_ATOMIC_REQ) * 100000)
-                        / (2 * ($sclk * $numSQC) * (End_Timestamp - Start_Timestamp)))
+                        / (2 * ($max_sclk * $sqc_per_gpu) * (End_Timestamp - Start_Timestamp)))
             unit: Pct of Peak
             tips:
         comparable: false # for now

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1500_TA_and_TD.yaml

@@ -20,27 +20,27 @@ Panel Config:
           tips: Tips
         metric:
           Address Processing Unit Busy:
-            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data Stall:
-            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data-Processor → Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Total Instructions:
@@ -128,21 +128,21 @@ Panel Config:
           tips: Tips
         metric:
           Data-Return Busy:
-            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Cache RAM → Data-Return Stall:
-            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Workgroup manager → Data-Return Stall:
-            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Coalescable Instructions:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1600_L1_cache.yaml

@@ -26,7 +26,7 @@ Panel Config:
             tips: 
           Bandwidth:
             value: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             unit: Pct of Peak
             tips: 
           Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx90a/1700_L2_cache.yaml

@@ -18,11 +18,11 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($L2Banks) * GRBM_GUI_ACTIVE)))
+            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($total_l2_chan) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips:
           Bandwidth:
-            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             unit: pct
             tips: 
           Hit Rate:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/0200_system-speed-of-light.yaml

@@ -27,93 +27,93 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (Int8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           Active CUs:
             value: $numActiveCUs
             unit: CUs
-            peak: $numCU
-            pop: ((100 * $numActiveCUs) / $numCU)
+            peak: $cu_per_gpu
+            pop: ((100 * $numActiveCUs) / $cu_per_gpu)
             tips: 
           SALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           VALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           MFMA Utilization:
-            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             tips: 
           VMEM Utilization:
-            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           Branch Utilization:
-            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           VALU Active Threads:
             value: AVG(((SQ_THREAD_CYCLES_VALU / SQ_ACTIVE_INST_VALU) if (SQ_ACTIVE_INST_VALU
@@ -132,18 +132,18 @@ Panel Config:
           Wavefront Occupancy:
             value: AVG((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE))
             unit: Wavefronts
-            peak: ($maxWavesPerCU * $numCU)
-            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($maxWavesPerCU
-              * $numCU))))
+            peak: ($max_waves_per_cu * $cu_per_gpu)
+            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($max_waves_per_cu
+              * $cu_per_gpu))))
             coll_level: SQ_LEVEL_WAVES
             tips: 
           Theoretical LDS Bandwidth:
-            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: (($sclk * $numCU) * 0.128)
-            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            peak: (($max_sclk * $cu_per_gpu) * 0.128)
+            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             tips: 
           LDS Bank Conflicts/Access:
             value: AVG(((SQ_LDS_BANK_CONFLICT / (SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT))
@@ -168,9 +168,9 @@ Panel Config:
           vL1D Cache BW:
             value: AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numCU)
+            peak: ((($max_sclk / 1000) * 64) * $cu_per_gpu)
             pop: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             tips: 
           L2 Cache Hit Rate:
             value: AVG((((100 * TCC_HIT_sum) / (TCC_HIT_sum + TCC_MISS_sum)) if ((TCC_HIT_sum
@@ -183,25 +183,25 @@ Panel Config:
           L2 Cache BW:
             value: AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 128) * TO_INT($L2Banks))
+            peak: ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan))
             pop: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+              / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             tips: 
           L2-Fabric Read BW:
             value: AVG((((TCC_EA0_RDREQ_32B_sum * 32) + ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_32B_sum)
               * 64)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA0_RDREQ_32B_sum * 32) + ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_32B_sum)
-              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Write BW:
             value: AVG((((TCC_EA0_WRREQ_64B_sum * 64) + ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_64B_sum)
               * 32)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA0_WRREQ_64B_sum * 64) + ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_64B_sum)
-              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Read Latency:
             value: AVG(((TCC_EA0_RDREQ_LEVEL_sum / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum
@@ -228,9 +228,9 @@ Panel Config:
           sL1D Cache BW:
             value: AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Hit Rate:
             value: AVG(((100 * SQC_ICACHE_HITS) / (SQC_ICACHE_HITS + SQC_ICACHE_MISSES)))
@@ -241,9 +241,9 @@ Panel Config:
           L1I BW:
             value: AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Fetch Latency:
             value: AVG((SQ_ACCUM_PREV_HIRES / SQ_IFETCH))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/0300_mem_chart.yaml

@@ -74,7 +74,7 @@ Panel Config:
             tips:
           Num CUs:
             #alias: num_cu_
-            value: $numCU
+            value: $cu_per_gpu
             tips:
           VGPR:
             #alias: vgpr_
@@ -111,7 +111,7 @@ Panel Config:
           LDS Util:
             #alias: lds_util_
             value:
-              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU))),
+              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu))),
               0)
             tips:
           LDS Latency:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/0600_shader-processor-input.yaml

@@ -26,9 +26,9 @@ Panel Config:
             unit: Pct
             tips: 
           Scheduler-Pipe Utilization:
-            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
+            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
             unit: Pct
             tips: 
           Workgroup Manager Utilization:
@@ -38,15 +38,15 @@ Panel Config:
             unit: Pct
             tips: 
           Shader Engine Utilization:
-            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
+            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
             unit: Pct
             tips: 
           SIMD Utilization:
-            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Dispatched Workgroups:
@@ -91,77 +91,77 @@ Panel Config:
           tips: Tips
         metric:
           Not-scheduled Rate (Workgroup Manager):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Not-scheduled Rate (Scheduler-Pipe):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Scheduler-Pipe Stall Rate:
-            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
             unit: Pct
             tips: 
           Scratch Stall Rate:
-            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
+            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
             unit: Pct
             tips: 
           Insufficient SIMD Waveslots:
-            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD VGPRs:
-            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD SGPRs:
-            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU LDS:
-            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU Barriers:
-            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Workgroup Limit:
-            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Wavefront Limit:
-            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1100_compute-unit-compute-pipeline.yaml

@@ -26,55 +26,55 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (INT8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
 
     - metric_table:
@@ -107,27 +107,27 @@ Panel Config:
             unit: Instr/cycle
             tips: 
           SALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VMEM Utilization:
-            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           Branch Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Active Threads:
@@ -140,9 +140,9 @@ Panel Config:
             unit: Threads
             tips: 
           MFMA Utilization:
-            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
+            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips: 
           MFMA Instr Cycles:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1200_lds.yaml

@@ -18,16 +18,16 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips:
           Access Rate:
-            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips:
           Theoretical Bandwidth (% of Peak):
-            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             unit: Pct of Peak
             tips:
           Bank Conflict Rate:
@@ -56,11 +56,11 @@ Panel Config:
             unit: (Instr  + $normUnit)
             tips:
           Theoretical Bandwidth:
-            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
             unit: (Bytes  + $normUnit)
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1300_instruction-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_ICACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_ICACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -28,7 +28,7 @@ Panel Config:
             unit: Pct of Peak
             tips:
           L1I-L2 Bandwidth:
-            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($sclk * $numSQC) 
+            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($max_sclk * $sqc_per_gpu) 
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1400_constant-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_DCACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_DCACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -29,7 +29,7 @@ Panel Config:
             tips:
           sL1D-L2 BW:
             value: AVG(((SQC_TC_DATA_READ_REQ + SQC_TC_DATA_WRITE_REQ + SQC_TC_DATA_ATOMIC_REQ) * 100000)
-                        / (2 * ($sclk * $numSQC) * (End_Timestamp - Start_Timestamp)))
+                        / (2 * ($max_sclk * $sqc_per_gpu) * (End_Timestamp - Start_Timestamp)))
             unit: Pct of Peak
             tips:
         comparable: false # for now

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1500_TA_and_TD.yaml

@@ -20,27 +20,27 @@ Panel Config:
           tips: Tips
         metric:
           Address Processing Unit Busy:
-            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data Stall:
-            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data-Processor → Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Total Instructions:
@@ -128,21 +128,21 @@ Panel Config:
           tips: Tips
         metric:
           Data-Return Busy:
-            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Cache RAM → Data-Return Stall:
-            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Workgroup manager → Data-Return Stall:
-            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Coalescable Instructions:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1600_L1_cache.yaml

@@ -26,7 +26,7 @@ Panel Config:
             tips:
           Bandwidth:
             value: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             unit: Pct of Peak
             tips:
           Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1700_L2_cache.yaml

@@ -18,11 +18,11 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($L2Banks) * GRBM_GUI_ACTIVE)))
+            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($total_l2_chan) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips:
           Bandwidth:
-            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             unit: pct
             tips: 
           Hit Rate:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx940/1800_L2_cache_per_channel.yaml

@@ -117,7 +117,7 @@ Panel Config:
               (((100 * TCC_HIT[::_1]) / (TCC_HIT[::_1] + TCC_MISS[::_1])) if ((TCC_HIT[::_1]
               + TCC_MISS[::_1]) != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -130,7 +130,7 @@ Panel Config:
           "::_1":
             expr: (TO_INT(TCC_REQ[::_1]) / $denom)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -147,7 +147,7 @@ Panel Config:
             write req: AVG((TO_INT(TCC_WRITE[::_1]) / $denom))
             atomic req: AVG((TO_INT(TCC_ATOMIC[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -164,7 +164,7 @@ Panel Config:
             write req: AVG((TO_INT(TCC_EA0_WRREQ[::_1]) / $denom))
             atomic req: AVG((TO_INT(TCC_EA0_ATOMIC[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     # - metric_table:
@@ -187,7 +187,7 @@ Panel Config:
     #           AVG(((TCC_EA0_ATOMIC_LEVEL[::_1] / TCC_EA0_ATOMIC[::_1]) if
     #           (TCC_EA0_ATOMIC[::_1] != 0) else 0))
     #       placeholder_range:
-    #         "::_1": $totalL2Banks
+    #         "::_1": $total_l2_chan
     #     cli_style: simple_multiple_bar
 
     - metric_table:
@@ -202,7 +202,7 @@ Panel Config:
               ((TCC_EA0_RDREQ_LEVEL[::_1] / TCC_EA0_RDREQ[::_1]) if (TCC_EA0_RDREQ[::_1]
               != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -217,7 +217,7 @@ Panel Config:
               ((TCC_EA0_WRREQ_LEVEL[::_1] / TCC_EA0_WRREQ[::_1]) if (TCC_EA0_WRREQ[::_1]
               != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -231,7 +231,7 @@ Panel Config:
             expr: ((TCC_EA0_ATOMIC_LEVEL[::_1] / TCC_EA0_ATOMIC[::_1]) if
               (TCC_EA0_ATOMIC[::_1] != 0) else 0)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -248,7 +248,7 @@ Panel Config:
             ea read stall - gmi: None # Missing perfmon
             ea read stall - dram: None # Missing perfmon
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -267,7 +267,7 @@ Panel Config:
             ea write stall - dram: None # Missing perfmon
             ea write stall - starve: AVG((TO_INT(TCC_TOO_MANY_EA_WRREQS_STALL[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -280,7 +280,7 @@ Panel Config:
           "::_1":
             expr: TCC_TAG_STALL[::_1]
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -293,6 +293,6 @@ Panel Config:
           "::_1":
             expr: TCC_BUBBLE[::_1]
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
           # tips: Number of 128-byte read requests sent to EA
         cli_style: simple_box

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/0200_system-speed-of-light.yaml

@@ -27,93 +27,93 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (Int8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           Active CUs:
             value: $numActiveCUs
             unit: CUs
-            peak: $numCU
-            pop: ((100 * $numActiveCUs) / $numCU)
+            peak: $cu_per_gpu
+            pop: ((100 * $numActiveCUs) / $cu_per_gpu)
             tips: 
           SALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           VALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           MFMA Utilization:
-            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             tips: 
           VMEM Utilization:
-            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           Branch Utilization:
-            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           VALU Active Threads:
             value: AVG(((SQ_THREAD_CYCLES_VALU / SQ_ACTIVE_INST_VALU) if (SQ_ACTIVE_INST_VALU
@@ -132,18 +132,18 @@ Panel Config:
           Wavefront Occupancy:
             value: AVG((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE))
             unit: Wavefronts
-            peak: ($maxWavesPerCU * $numCU)
-            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($maxWavesPerCU
-              * $numCU))))
+            peak: ($max_waves_per_cu * $cu_per_gpu)
+            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($max_waves_per_cu
+              * $cu_per_gpu))))
             coll_level: SQ_LEVEL_WAVES
             tips: 
           Theoretical LDS Bandwidth:
-            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: (($sclk * $numCU) * 0.128)
-            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            peak: (($max_sclk * $cu_per_gpu) * 0.128)
+            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             tips: 
           LDS Bank Conflicts/Access:
             value: AVG(((SQ_LDS_BANK_CONFLICT / (SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT))
@@ -168,9 +168,9 @@ Panel Config:
           vL1D Cache BW:
             value: AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numCU)
+            peak: ((($max_sclk / 1000) * 64) * $cu_per_gpu)
             pop: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             tips: 
           L2 Cache Hit Rate:
             value: AVG((((100 * TCC_HIT_sum) / (TCC_HIT_sum + TCC_MISS_sum)) if ((TCC_HIT_sum
@@ -183,25 +183,25 @@ Panel Config:
           L2 Cache BW:
             value: AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 128) * TO_INT($L2Banks))
+            peak: ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan))
             pop: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+              / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             tips: 
           L2-Fabric Read BW:
             value: AVG((((TCC_EA0_RDREQ_32B_sum * 32) + ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_32B_sum)
               * 64)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA0_RDREQ_32B_sum * 32) + ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_32B_sum)
-              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Write BW:
             value: AVG((((TCC_EA0_WRREQ_64B_sum * 64) + ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_64B_sum)
               * 32)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA0_WRREQ_64B_sum * 64) + ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_64B_sum)
-              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Read Latency:
             value: AVG(((TCC_EA0_RDREQ_LEVEL_sum / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum
@@ -228,9 +228,9 @@ Panel Config:
           sL1D Cache BW:
             value: AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Hit Rate:
             value: AVG(((100 * SQC_ICACHE_HITS) / (SQC_ICACHE_HITS + SQC_ICACHE_MISSES)))
@@ -241,9 +241,9 @@ Panel Config:
           L1I BW:
             value: AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Fetch Latency:
             value: AVG((SQ_ACCUM_PREV_HIRES / SQ_IFETCH))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/0300_mem_chart.yaml

@@ -74,7 +74,7 @@ Panel Config:
             tips:
           Num CUs:
             #alias: num_cu_
-            value: $numCU
+            value: $cu_per_gpu
             tips:
           VGPR:
             #alias: vgpr_
@@ -111,7 +111,7 @@ Panel Config:
           LDS Util:
             #alias: lds_util_
             value:
-              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU))),
+              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu))),
               0)
             tips:
           LDS Latency:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/0600_shader-processor-input.yaml

@@ -26,9 +26,9 @@ Panel Config:
             unit: Pct
             tips: 
           Scheduler-Pipe Utilization:
-            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
+            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
             unit: Pct
             tips: 
           Workgroup Manager Utilization:
@@ -38,15 +38,15 @@ Panel Config:
             unit: Pct
             tips: 
           Shader Engine Utilization:
-            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
+            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
             unit: Pct
             tips: 
           SIMD Utilization:
-            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Dispatched Workgroups:
@@ -91,77 +91,77 @@ Panel Config:
           tips: Tips
         metric:
           Not-scheduled Rate (Workgroup Manager):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Not-scheduled Rate (Scheduler-Pipe):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Scheduler-Pipe Stall Rate:
-            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
             unit: Pct
             tips: 
           Scratch Stall Rate:
-            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
+            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
             unit: Pct
             tips: 
           Insufficient SIMD Waveslots:
-            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD VGPRs:
-            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD SGPRs:
-            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU LDS:
-            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU Barriers:
-            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Workgroup Limit:
-            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Wavefront Limit:
-            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1100_compute-unit-compute-pipeline.yaml

@@ -26,55 +26,55 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (INT8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
 
     - metric_table:
@@ -107,27 +107,27 @@ Panel Config:
             unit: Instr/cycle
             tips: 
           SALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VMEM Utilization:
-            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           Branch Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Active Threads:
@@ -140,9 +140,9 @@ Panel Config:
             unit: Threads
             tips: 
           MFMA Utilization:
-            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
+            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips: 
           MFMA Instr Cycles:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1200_lds.yaml

@@ -18,16 +18,16 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips:
           Access Rate:
-            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips:
           Theoretical Bandwidth (% of Peak):
-            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             unit: Pct of Peak
             tips:
           Bank Conflict Rate:
@@ -56,11 +56,11 @@ Panel Config:
             unit: (Instr  + $normUnit)
             tips:
           Theoretical Bandwidth:
-            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
             unit: (Bytes  + $normUnit)
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1300_instruction-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_ICACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_ICACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -28,7 +28,7 @@ Panel Config:
             unit: Pct of Peak
             tips:
           L1I-L2 Bandwidth:
-            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($sclk * $numSQC) 
+            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($max_sclk * $sqc_per_gpu) 
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1400_constant-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_DCACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_DCACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -29,7 +29,7 @@ Panel Config:
             tips:
           sL1D-L2 BW:
             value: AVG(((SQC_TC_DATA_READ_REQ + SQC_TC_DATA_WRITE_REQ + SQC_TC_DATA_ATOMIC_REQ) * 100000)
-                        / (2 * ($sclk * $numSQC) * (End_Timestamp - Start_Timestamp)))
+                        / (2 * ($max_sclk * $sqc_per_gpu) * (End_Timestamp - Start_Timestamp)))
             unit: Pct of Peak
             tips:
         comparable: false # for now

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1500_TA_and_TD.yaml

@@ -20,27 +20,27 @@ Panel Config:
           tips: Tips
         metric:
           Address Processing Unit Busy:
-            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data Stall:
-            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data-Processor → Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Total Instructions:
@@ -128,21 +128,21 @@ Panel Config:
           tips: Tips
         metric:
           Data-Return Busy:
-            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Cache RAM → Data-Return Stall:
-            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Workgroup manager → Data-Return Stall:
-            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Coalescable Instructions:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1600_L1_cache.yaml

@@ -26,7 +26,7 @@ Panel Config:
             tips:
           Bandwidth:
             value: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             unit: Pct of Peak
             tips:
           Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1700_L2_cache.yaml

@@ -18,11 +18,11 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($L2Banks) * GRBM_GUI_ACTIVE)))
+            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($total_l2_chan) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips:
           Bandwidth:
-            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             unit: pct
             tips: 
           Hit Rate:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx941/1800_L2_cache_per_channel.yaml

@@ -117,7 +117,7 @@ Panel Config:
               (((100 * TCC_HIT[::_1]) / (TCC_HIT[::_1] + TCC_MISS[::_1])) if ((TCC_HIT[::_1]
               + TCC_MISS[::_1]) != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -130,7 +130,7 @@ Panel Config:
           "::_1":
             expr: (TO_INT(TCC_REQ[::_1]) / $denom)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -147,7 +147,7 @@ Panel Config:
             write req: AVG((TO_INT(TCC_WRITE[::_1]) / $denom))
             atomic req: AVG((TO_INT(TCC_ATOMIC[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -164,7 +164,7 @@ Panel Config:
             write req: AVG((TO_INT(TCC_EA0_WRREQ[::_1]) / $denom))
             atomic req: AVG((TO_INT(TCC_EA0_ATOMIC[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     # - metric_table:
@@ -187,7 +187,7 @@ Panel Config:
     #           AVG(((TCC_EA0_ATOMIC_LEVEL[::_1] / TCC_EA0_ATOMIC[::_1]) if
     #           (TCC_EA0_ATOMIC[::_1] != 0) else 0))
     #       placeholder_range:
-    #         "::_1": $totalL2Banks
+    #         "::_1": $total_l2_chan
     #     cli_style: simple_multiple_bar
 
     - metric_table:
@@ -202,7 +202,7 @@ Panel Config:
               ((TCC_EA0_RDREQ_LEVEL[::_1] / TCC_EA0_RDREQ[::_1]) if (TCC_EA0_RDREQ[::_1]
               != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -217,7 +217,7 @@ Panel Config:
               ((TCC_EA0_WRREQ_LEVEL[::_1] / TCC_EA0_WRREQ[::_1]) if (TCC_EA0_WRREQ[::_1]
               != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -231,7 +231,7 @@ Panel Config:
             expr: ((TCC_EA0_ATOMIC_LEVEL[::_1] / TCC_EA0_ATOMIC[::_1]) if
               (TCC_EA0_ATOMIC[::_1] != 0) else 0)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -248,7 +248,7 @@ Panel Config:
             ea read stall - gmi: None # Missing perfmon
             ea read stall - dram: None # Missing perfmon
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -267,7 +267,7 @@ Panel Config:
             ea write stall - dram: None # Missing perfmon
             ea write stall - starve: TCC_TOO_MANY_EA_WRREQS_STALL
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -280,7 +280,7 @@ Panel Config:
           "::_1":
             expr: TCC_TAG_STALL[::_1]
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -293,6 +293,6 @@ Panel Config:
           "::_1":
             expr: TCC_BUBBLE[::_1]
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
           # tips: Number of 128-byte read requests sent to EA
         cli_style: simple_box

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/0200_system-speed-of-light.yaml

@@ -27,93 +27,93 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (Int8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 4096) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 4096) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
             tips: 
           Active CUs:
             value: $numActiveCUs
             unit: CUs
-            peak: $numCU
-            pop: ((100 * $numActiveCUs) / $numCU)
+            peak: $cu_per_gpu
+            pop: ((100 * $numActiveCUs) / $cu_per_gpu)
             tips: 
           SALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_SCA) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           VALU Utilization:
-            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $numCU)))
+            pop: AVG(((100 * SQ_ACTIVE_INST_VALU) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             tips: 
           MFMA Utilization:
-            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            value: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             unit: pct
             peak: 100
-            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $numCU)
+            pop: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((GRBM_GUI_ACTIVE * $cu_per_gpu)
               * 4)))
             tips: 
           VMEM Utilization:
-            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           Branch Utilization:
-            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            value: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            pop: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             tips: 
           VALU Active Threads:
             value: AVG(((SQ_THREAD_CYCLES_VALU / SQ_ACTIVE_INST_VALU) if (SQ_ACTIVE_INST_VALU
@@ -132,18 +132,18 @@ Panel Config:
           Wavefront Occupancy:
             value: AVG((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE))
             unit: Wavefronts
-            peak: ($maxWavesPerCU * $numCU)
-            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($maxWavesPerCU
-              * $numCU))))
+            peak: ($max_waves_per_cu * $cu_per_gpu)
+            pop: (100 * AVG(((SQ_ACCUM_PREV_HIRES / GRBM_GUI_ACTIVE) / ($max_waves_per_cu
+              * $cu_per_gpu))))
             coll_level: SQ_LEVEL_WAVES
             tips: 
           Theoretical LDS Bandwidth:
-            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            value: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: (($sclk * $numCU) * 0.128)
-            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            peak: (($max_sclk * $cu_per_gpu) * 0.128)
+            pop: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             tips: 
           LDS Bank Conflicts/Access:
             value: AVG(((SQ_LDS_BANK_CONFLICT / (SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT))
@@ -168,9 +168,9 @@ Panel Config:
           vL1D Cache BW:
             value: AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numCU)
+            peak: ((($max_sclk / 1000) * 64) * $cu_per_gpu)
             pop: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             tips: 
           L2 Cache Hit Rate:
             value: AVG((((100 * TCC_HIT_sum) / (TCC_HIT_sum + TCC_MISS_sum)) if ((TCC_HIT_sum
@@ -183,25 +183,25 @@ Panel Config:
           L2 Cache BW:
             value: AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: ((($sclk / 1000) * 128) * TO_INT($L2Banks))
+            peak: ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan))
             pop: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+              / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             tips: 
           L2-Fabric Read BW:
             value: AVG((((TCC_EA0_RDREQ_32B_sum * 32) + ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_32B_sum)
               * 64)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA0_RDREQ_32B_sum * 32) + ((TCC_EA0_RDREQ_sum - TCC_EA0_RDREQ_32B_sum)
-              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 64)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Write BW:
             value: AVG((((TCC_EA0_WRREQ_64B_sum * 64) + ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_64B_sum)
               * 32)) / (End_Timestamp - Start_Timestamp)))
             unit: GB/s
-            peak: $hbmBW
+            peak: $hbm_bw
             pop: ((100 * AVG((((TCC_EA0_WRREQ_64B_sum * 64) + ((TCC_EA0_WRREQ_sum - TCC_EA0_WRREQ_64B_sum)
-              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbmBW)
+              * 32)) / (End_Timestamp - Start_Timestamp)))) / $hbm_bw)
             tips: 
           L2-Fabric Read Latency:
             value: AVG(((TCC_EA0_RDREQ_LEVEL_sum / TCC_EA0_RDREQ_sum) if (TCC_EA0_RDREQ_sum
@@ -228,9 +228,9 @@ Panel Config:
           sL1D Cache BW:
             value: AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_DCACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Hit Rate:
             value: AVG(((100 * SQC_ICACHE_HITS) / (SQC_ICACHE_HITS + SQC_ICACHE_MISSES)))
@@ -241,9 +241,9 @@ Panel Config:
           L1I BW:
             value: AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))
             unit: GB/s
-            peak: ((($sclk / 1000) * 64) * $numSQC)
-            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($sclk
-              / 1000) * 64) * $numSQC))
+            peak: ((($max_sclk / 1000) * 64) * $sqc_per_gpu)
+            pop: ((100 * AVG(((SQC_ICACHE_REQ / (End_Timestamp - Start_Timestamp)) * 64))) / ((($max_sclk
+              / 1000) * 64) * $sqc_per_gpu))
             tips: 
           L1I Fetch Latency:
             value: AVG((SQ_ACCUM_PREV_HIRES / SQ_IFETCH))

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/0300_mem_chart.yaml

@@ -74,7 +74,7 @@ Panel Config:
             tips:
           Num CUs:
             #alias: num_cu_
-            value: $numCU
+            value: $cu_per_gpu
             tips:
           VGPR:
             #alias: vgpr_
@@ -111,7 +111,7 @@ Panel Config:
           LDS Util:
             #alias: lds_util_
             value:
-              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU))),
+              ROUND(AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu))),
               0)
             tips:
           LDS Latency:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/0600_shader-processor-input.yaml

@@ -26,9 +26,9 @@ Panel Config:
             unit: Pct
             tips: 
           Scheduler-Pipe Utilization:
-            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
-            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $numPipes * $numSE))
+            avg: AVG(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            min: MIN(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
+            max: MAX(100 * SPI_CSN_BUSY / (GRBM_GUI_ACTIVE * $pipes_per_gpu * $se_per_gpu))
             unit: Pct
             tips: 
           Workgroup Manager Utilization:
@@ -38,15 +38,15 @@ Panel Config:
             unit: Pct
             tips: 
           Shader Engine Utilization:
-            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
-            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $numSE))
+            avg: AVG(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            min: MIN(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
+            max: MAX(100 * SQ_BUSY_CYCLES / (GRBM_GUI_ACTIVE * $se_per_gpu))
             unit: Pct
             tips: 
           SIMD Utilization:
-            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SQ_BUSY_CU_CYCLES / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Dispatched Workgroups:
@@ -91,77 +91,77 @@ Panel Config:
           tips: Tips
         metric:
           Not-scheduled Rate (Workgroup Manager):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Not-scheduled Rate (Scheduler-Pipe):
-            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
-            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((100 * SPI_RA_REQ_NO_ALLOC / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None)
             unit: Pct
             tips: 
           Scheduler-Pipe Stall Rate:
-            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            avg: AVG((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            min: MIN((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
-            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY !=
+            max: MAX((((100 * SPI_RA_RES_STALL_CSN) / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY !=
               0) else None))
             unit: Pct
             tips: 
           Scratch Stall Rate:
-            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
-            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $numSE)) if (GRBM_SPI_BUSY != 0) else None)
+            avg: AVG((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            min: MIN((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
+            max: MAX((100 * SPI_RA_TMP_STALL_CSN / (GRBM_SPI_BUSY * $se_per_gpu)) if (GRBM_SPI_BUSY != 0) else None)
             unit: Pct
             tips: 
           Insufficient SIMD Waveslots:
-            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_WAVE_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD VGPRs:
-            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_VGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient SIMD SGPRs:
-            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(100 * SPI_RA_SGPR_SIMD_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU LDS:
-            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_LDS_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Insufficient CU Barriers:
-            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_BAR_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Workgroup Limit:
-            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_TGLIM_CU_FULL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 
           Reached CU Wavefront Limit:
-            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
-            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $numCU))
+            avg: AVG(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            min: MIN(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
+            max: MAX(400 * SPI_RA_WVLIM_STALL_CSN / (GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: Pct
             tips: 

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1100_compute-unit-compute-pipeline.yaml

@@ -26,55 +26,55 @@ Panel Config:
               + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
               / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
               + SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
               + SQ_INSTS_VALU_MUL_F32) + SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32))))
               + (64 * (((SQ_INSTS_VALU_ADD_F64 + SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64)
-              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($sclk
-              * $numCU) * 64) * 2) / 1000))
+              + (2 * SQ_INSTS_VALU_FMA_F64)))) / (End_Timestamp - Start_Timestamp)))) / (((($max_sclk
+              * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           VALU IOPs:
             value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: (((($sclk * $numCU) * 64) * 2) / 1000)
+            peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
             pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
-              - Start_Timestamp)))) / (((($sclk * $numCU) * 64) * 2) / 1000))
+              - Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
             tips: 
           MFMA FLOPs (BF16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F16):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
           MFMA FLOPs (F32):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA FLOPs (F64):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GFLOP
-            peak: ((($sclk * $numCU) * 256) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 256) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 256) / 1000))
             tips: 
           MFMA IOPs (INT8):
             value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
             unit: GIOP
-            peak: ((($sclk * $numCU) * 1024) / 1000)
+            peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
             pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk * $numCU) * 1024) / 1000))
+              / ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
             tips: 
 
     - metric_table:
@@ -107,27 +107,27 @@ Panel Config:
             unit: Instr/cycle
             tips: 
           SALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_SCA) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_VALU) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VMEM Utilization:
-            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * (SQ_ACTIVE_INST_FLAT+SQ_ACTIVE_INST_VMEM)) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           Branch Utilization:
-            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
-            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $numCU))
+            avg: AVG((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            min: MIN((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
+            max: MAX((((100 * SQ_ACTIVE_INST_MISC) / GRBM_GUI_ACTIVE) / $cu_per_gpu))
             unit: pct
             tips: 
           VALU Active Threads:
@@ -140,9 +140,9 @@ Panel Config:
             unit: Threads
             tips: 
           MFMA Utilization:
-            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
-            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $numCU) * GRBM_GUI_ACTIVE)))
+            avg: AVG(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            min: MIN(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
+            max: MAX(((100 * SQ_VALU_MFMA_BUSY_CYCLES) / ((4 * $cu_per_gpu) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips: 
           MFMA Instr Cycles:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1200_lds.yaml

@@ -18,16 +18,16 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((100 * SQ_LDS_IDX_ACTIVE) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips:
           Access Rate:
-            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $numCU)))
+            value: AVG(((200 * SQ_ACTIVE_INST_LDS) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: Pct of Peak
             tips:
           Theoretical Bandwidth (% of Peak):
-            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
-              / (End_Timestamp - Start_Timestamp)) / (($sclk * $numCU) * 0.00128)))
+            value: AVG((((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
+              / (End_Timestamp - Start_Timestamp)) / (($max_sclk * $cu_per_gpu) * 0.00128)))
             unit: Pct of Peak
             tips:
           Bank Conflict Rate:
@@ -56,11 +56,11 @@ Panel Config:
             unit: (Instr  + $normUnit)
             tips:
           Theoretical Bandwidth:
-            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            avg: AVG(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            min: MIN(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
-            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($LDSBanks))
+            max: MAX(((((SQ_LDS_IDX_ACTIVE - SQ_LDS_BANK_CONFLICT) * 4) * TO_INT($lds_banks_per_cu))
               / $denom))
             unit: (Bytes  + $normUnit)
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1300_instruction-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_ICACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_ICACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -28,7 +28,7 @@ Panel Config:
             unit: Pct of Peak
             tips:
           L1I-L2 Bandwidth:
-            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($sclk * $numSQC) 
+            value: AVG(((SQC_TC_INST_REQ * 100000) / (2 * ($max_sclk * $sqc_per_gpu) 
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1400_constant-cache.yaml

@@ -18,7 +18,7 @@ Panel Config:
           tips: Tips
         metric:
           Bandwidth:
-            value: AVG(((SQC_DCACHE_REQ * 100000) / (($sclk * $numSQC)
+            value: AVG(((SQC_DCACHE_REQ * 100000) / (($max_sclk * $sqc_per_gpu)
               * (End_Timestamp - Start_Timestamp))))
             unit: Pct of Peak
             tips:
@@ -29,7 +29,7 @@ Panel Config:
             tips:
           sL1D-L2 BW:
             value: AVG(((SQC_TC_DATA_READ_REQ + SQC_TC_DATA_WRITE_REQ + SQC_TC_DATA_ATOMIC_REQ) * 100000)
-                        / (2 * ($sclk * $numSQC) * (End_Timestamp - Start_Timestamp)))
+                        / (2 * ($max_sclk * $sqc_per_gpu) * (End_Timestamp - Start_Timestamp)))
             unit: Pct of Peak
             tips:
         comparable: false # for now

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1500_TA_and_TD.yaml

@@ -20,27 +20,27 @@ Panel Config:
           tips: Tips
         metric:
           Address Processing Unit Busy:
-            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_TA_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data Stall:
-            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_DATA_STALLED_BY_TC_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Data-Processor → Address Stall:
-            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TA_ADDR_STALLED_BY_TD_CYCLES_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Total Instructions:
@@ -128,21 +128,21 @@ Panel Config:
           tips: Tips
         metric:
           Data-Return Busy:
-            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TD_BUSY_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Cache RAM → Data-Return Stall:
-            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_TC_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Workgroup manager → Data-Return Stall:
-            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
-            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $numCU)))
+            avg: AVG(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            min: MIN(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
+            max: MAX(((100 * TD_SPI_STALL_sum) / (GRBM_GUI_ACTIVE * $cu_per_gpu)))
             unit: pct
             tips: 
           Coalescable Instructions:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1600_L1_cache.yaml

@@ -26,7 +26,7 @@ Panel Config:
             tips:
           Bandwidth:
             value: ((100 * AVG(((TCP_TOTAL_CACHE_ACCESSES_sum * 64) / (End_Timestamp - Start_Timestamp))))
-              / ((($sclk / 1000) * 64) * $numCU))
+              / ((($max_sclk / 1000) * 64) * $cu_per_gpu))
             unit: Pct of Peak
             tips:
           Utilization:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1700_L2_cache.yaml

@@ -18,11 +18,11 @@ Panel Config:
           tips: Tips
         metric:
           Utilization:
-            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($L2Banks) * GRBM_GUI_ACTIVE)))
+            value: AVG(((TCC_BUSY_sum * 100) / (TO_INT($total_l2_chan) * GRBM_GUI_ACTIVE)))
             unit: pct
             tips:
           Bandwidth:
-            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($sclk / 1000) * 128) * TO_INT($L2Banks)))
+            value: ((100 * AVG(((TCC_REQ_sum * 128) / (End_Timestamp - Start_Timestamp)))) / ((($max_sclk / 1000) * 128) * TO_INT($total_l2_chan)))
             unit: pct
             tips: 
           Hit Rate:

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/gfx942/1800_L2_cache_per_channel.yaml

@@ -117,7 +117,7 @@ Panel Config:
               (((100 * TCC_HIT[::_1]) / (TCC_HIT[::_1] + TCC_MISS[::_1])) if ((TCC_HIT[::_1]
               + TCC_MISS[::_1]) != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -130,7 +130,7 @@ Panel Config:
           "::_1":
             expr: (TO_INT(TCC_REQ[::_1]) / $denom)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -147,7 +147,7 @@ Panel Config:
             write req: AVG((TO_INT(TCC_WRITE[::_1]) / $denom))
             atomic req: AVG((TO_INT(TCC_ATOMIC[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -164,7 +164,7 @@ Panel Config:
             write req: AVG((TO_INT(TCC_EA0_WRREQ[::_1]) / $denom))
             atomic req: AVG((TO_INT(TCC_EA0_ATOMIC[::_1]) / $denom))
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     # - metric_table:
@@ -187,7 +187,7 @@ Panel Config:
     #           AVG(((TCC_EA0_ATOMIC_LEVEL[::_1] / TCC_EA0_ATOMIC[::_1]) if
     #           (TCC_EA0_ATOMIC[::_1] != 0) else 0))
     #       placeholder_range:
-    #         "::_1": $totalL2Banks
+    #         "::_1": $total_l2_chan
     #     cli_style: simple_multiple_bar
 
     - metric_table:
@@ -202,7 +202,7 @@ Panel Config:
               ((TCC_EA0_RDREQ_LEVEL[::_1] / TCC_EA0_RDREQ[::_1]) if (TCC_EA0_RDREQ[::_1]
               != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -217,7 +217,7 @@ Panel Config:
               ((TCC_EA0_WRREQ_LEVEL[::_1] / TCC_EA0_WRREQ[::_1]) if (TCC_EA0_WRREQ[::_1]
               != 0) else None)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -231,7 +231,7 @@ Panel Config:
             expr: ((TCC_EA0_ATOMIC_LEVEL[::_1] / TCC_EA0_ATOMIC[::_1]) if
               (TCC_EA0_ATOMIC[::_1] != 0) else 0)
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -248,7 +248,7 @@ Panel Config:
             ea read stall - gmi: None # Missing perfmon
             ea read stall - dram: None # Missing perfmon
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -267,7 +267,7 @@ Panel Config:
             ea write stall - dram: None # Missing perfmon
             ea write stall - starve: TCC_TOO_MANY_EA_WRREQS_STALL
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_multiple_bar
 
     - metric_table:
@@ -280,7 +280,7 @@ Panel Config:
           "::_1":
             expr: TCC_TAG_STALL[::_1]
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
         cli_style: simple_box
 
     - metric_table:
@@ -293,6 +293,6 @@ Panel Config:
           "::_1":
             expr: TCC_BUBBLE[::_1]
           placeholder_range:
-            "::_1": $totalL2Banks
+            "::_1": $total_l2_chan
           # tips: Number of 128-byte read requests sent to EA
         cli_style: simple_box

projects/rocprofiler-compute/src/omniperf_soc/analysis_configs/panel_config_template.yaml

@@ -34,16 +34,16 @@ Panel Config:
           tips: Tips
         metric:
           METRIC01:
-            value: AVG(100 * SQ_ACTIVE_INST_SCA / ( GRBM_GUI_ACTIVE * $numCU ))
+            value: AVG(100 * SQ_ACTIVE_INST_SCA / ( GRBM_GUI_ACTIVE * $cu_per_gpu ))
             unit: pct
             peak: 100
-            pop: AVG(100* SQ_ACTIVE_INST_SCA/(GRBM_GUI_ACTIVE*$numCU))
+            pop: AVG(100* SQ_ACTIVE_INST_SCA/(GRBM_GUI_ACTIVE*$cu_per_gpu))
             tips: *METRIC01_anchor
           METRIC02:
-            value: AVG(100 * SQ_ACTIVE_INST_VALU / ( GRBM_GUI_ACTIVE * $numCU))
+            value: AVG(100 * SQ_ACTIVE_INST_VALU / ( GRBM_GUI_ACTIVE * $cu_per_gpu))
             unit: pct
             peak: 100
-            pop: AVG(100* SQ_ACTIVE_INST_VALU/(GRBM_GUI_ACTIVE*$numCU))
+            pop: AVG(100* SQ_ACTIVE_INST_VALU/(GRBM_GUI_ACTIVE*$cu_per_gpu))
             tips:
     # CSV table sample
     - raw_csv_table:

projects/rocprofiler-compute/src/omniperf_soc/soc_base.py

@@ -48,7 +48,7 @@ class OmniSoC_Base():
         # In some cases (i.e. --specs) path will not be given
         if hasattr(self.__args, "path"):
             if self.__args.path == os.path.join(os.getcwd(), "workloads"):
-                self.__workload_dir = os.path.join(self.__args.path, self.__args.name, self._mspec.GPU)
+                self.__workload_dir = os.path.join(self.__args.path, self.__args.name, self._mspec.gpu_model)
             else:
                 self.__workload_dir = self.__args.path
 
@@ -96,17 +96,17 @@ class OmniSoC_Base():
             return
 
         # load stats from rocminfo
-        self._mspec.L1 = ""
-        self._mspec.L2 = ""
+        self._mspec.gpu_l1 = ""
+        self._mspec.gpu_l2 = ""
         for idx2, linetext in enumerate(self._mspec._rocminfo):
             key = search(r"^\s*L1:\s+ ([a-zA-Z0-9]+)\s*", linetext)
             if key != None:
-                self._mspec.L1 = key
+                self._mspec.gpu_l1 = key
                 continue
 
             key = search(r"^\s*L2:\s+ ([a-zA-Z0-9]+)\s*", linetext)
             if key != None:
-                self._mspec.L2 = key
+                self._mspec.gpu_l2 = key
                 continue
 
             key = search(r"^\s*Max Clock Freq\. \(MHz\):\s+([0-9]+)", linetext)
@@ -116,17 +116,17 @@ class OmniSoC_Base():
 
             key = search(r"^\s*Compute Unit:\s+ ([a-zA-Z0-9]+)\s*", linetext)
             if key != None:
-                self._mspec.CU = key
+                self._mspec.cu_per_gpu = key
                 continue
 
             key = search(r"^\s*SIMDs per CU:\s+ ([a-zA-Z0-9]+)\s*", linetext)
             if key != None:
-                self._mspec.SIMD = key
+                self._mspec.simd_per_cu = key
                 continue
 
             key = search(r"^\s*Shader Engines:\s+ ([a-zA-Z0-9]+)\s*", linetext)
             if key != None:
-                self._mspec.SE = key
+                self._mspec.se_per_gpu = key
                 continue
 
             key = search(r"^\s*Wavefront Size:\s+ ([a-zA-Z0-9]+)\s*", linetext)
@@ -157,13 +157,13 @@ class OmniSoC_Base():
         self._mspec.memory_partition = ""
 
         # specify gpu name for gfx942 hardware
-        self._mspec.GPU = list(SUPPORTED_ARCHS[self._mspec.arch].keys())[0].upper()
-        if self._mspec.GPU == "MI300":
-            self._mspec.GPU = list(SUPPORTED_ARCHS[self._mspec.arch].values())[0][0]
-        if (self._mspec.arch == "gfx942") and ("MI300A" in self._mspec._rocminfo):
-            self._mspec.GPU = "MI300A_A1"
-        if (self._mspec.arch == "gfx942") and ("MI300A" not in self._mspec._rocminfo):
-            self._mspec.GPU = "MI300X_A1"
+        self._mspec.gpu_model = list(SUPPORTED_ARCHS[self._mspec.gpu_arch].keys())[0].upper()
+        if self._mspec.gpu_model == "MI300":
+            self._mspec.gpu_model = list(SUPPORTED_ARCHS[self._mspec.gpu_arch].values())[0][0]
+        if (self._mspec.gpu_arch == "gfx942") and ("MI300A" in self._mspec._rocminfo):
+            self._mspec.gpu_model = "MI300A_A1"
+        if (self._mspec.gpu_arch == "gfx942") and ("MI300A" not in self._mspec._rocminfo):
+            self._mspec.gpu_model = "MI300X_A1"
 
     @demarcate
     def perfmon_filter(self, roofline_perfmon_only: bool):

projects/rocprofiler-compute/src/omniperf_soc/soc_gfx906.py

@@ -52,9 +52,9 @@ class gfx906_soc (OmniSoC_Base):
         
         # Set arch specific specs
         self._mspec.L2Banks = 16
-        self._mspec.LDSBanks = 32
-        self._mspec.numSQC = 15
-        self._mspec.numPipes = 4
+        self._mspec.lds_banks_per_cu = 32
+        self._mspec.sqc_per_gpu = 15
+        self._mspec.pipes_per_gpu = 4
 
     # -----------------------
     # Required child methods

projects/rocprofiler-compute/src/omniperf_soc/soc_gfx908.py

@@ -52,12 +52,12 @@ class gfx908_soc (OmniSoC_Base):
 
         # Set arch specific specs
         self._mspec.L2Banks = 32
-        self._mspec.LDSBanks = 32
-        self._mspec.numSQC = 30
-        self._mspec.numPipes = 4
+        self._mspec.lds_banks_per_cu = 32
+        self._mspec.sqc_per_gpu = 30
+        self._mspec.pipes_per_gpu = 4
         # --showmclkrange is broken in Mi100, hardcode freq
-        self._info['max_mclk'] = 1200
-        self._info['cur_mclk'] = 1200
+        self._mspec.max_mclk = 1200
+        self._mspec.cur_mclk = 1200
 
     @demarcate
     def get_profiler_options(self):

projects/rocprofiler-compute/src/omniperf_soc/soc_gfx90a.py

@@ -58,9 +58,9 @@ class gfx90a_soc (OmniSoC_Base):
 
         # Set arch specific specs
         self._mspec.L2Banks = 32
-        self._mspec.LDSBanks = 32
-        self._mspec.numSQC = 56
-        self._mspec.numPipes = 4
+        self._mspec.lds_banks_per_cu = 32
+        self._mspec.sqc_per_gpu = 56
+        self._mspec.pipes_per_gpu = 4
 
     # -----------------------
     # Required child methods

projects/rocprofiler-compute/src/omniperf_soc/soc_gfx940.py

@@ -63,9 +63,9 @@ class gfx940_soc (OmniSoC_Base):
 
         # Set arch specific specs
         self._mspec.L2Banks = 16
-        self._mspec.LDSBanks = 32
-        self._mspec.numSQC = 56
-        self._mspec.numPipes = 4
+        self._mspec.lds_banks_per_cu = 32
+        self._mspec.sqc_per_gpu = 56
+        self._mspec.pipes_per_gpu = 4
 
     # -----------------------
     # Required child methods

projects/rocprofiler-compute/src/omniperf_soc/soc_gfx941.py

@@ -63,9 +63,9 @@ class gfx941_soc (OmniSoC_Base):
 
         # Set arch specific specs
         self._mspec.L2Banks = 16
-        self._mspec.LDSBanks = 32
-        self._mspec.numSQC = 56
-        self._mspec.numPipes = 4
+        self._mspec.lds_banks_per_cu = 32
+        self._mspec.sqc_per_gpu = 56
+        self._mspec.pipes_per_gpu = 4
 
     # -----------------------
     # Required child methods

projects/rocprofiler-compute/src/omniperf_soc/soc_gfx942.py

@@ -63,9 +63,9 @@ class gfx942_soc (OmniSoC_Base):
 
         # Set arch specific specs
         self._mspec.L2Banks = 16
-        self._mspec.LDSBanks = 32
-        self._mspec.numSQC = 56
-        self._mspec.numPipes = 4
+        self._mspec.lds_banks_per_cu = 32
+        self._mspec.sqc_per_gpu = 56
+        self._mspec.pipes_per_gpu = 4
 
     # -----------------------
     # Required child methods

projects/rocprofiler-compute/src/roofline.py

@@ -79,7 +79,7 @@ class Roofline:
     def roof_setup(self):
         # set default workload path if not specified
         if self.__run_parameters['workload_dir'] == os.path.join(os.getcwd(), 'workloads'):
-            self.__run_parameters['workload_dir'] = os.path.join(self.__run_parameters['workload_dir'], self.__args.name, self.__mspec.GPU)
+            self.__run_parameters['workload_dir'] = os.path.join(self.__run_parameters['workload_dir'], self.__args.name, self.__mspec.gpu_model)
         # create new directory for roofline if it doesn't exist
         if not os.path.isdir(self.__run_parameters['workload_dir']):
             os.makedirs(self.__run_parameters['workload_dir'])

projects/rocprofiler-compute/src/utils/parser.py

@@ -78,9 +78,9 @@ supported_denom = {
 # Build-in defined in mongodb variables:
 build_in_vars = {
     "numActiveCUs": "TO_INT(MIN((((ROUND(AVG(((4 * SQ_BUSY_CU_CYCLES) / GRBM_GUI_ACTIVE)), \
-              0) / $maxWavesPerCU) * 8) + MIN(MOD(ROUND(AVG(((4 * SQ_BUSY_CU_CYCLES) \
-              / GRBM_GUI_ACTIVE)), 0), $maxWavesPerCU), 8)), $numCU))",
-    "kernelBusyCycles": "ROUND(AVG((((End_Timestamp - Start_Timestamp) / 1000) * $sclk)), 0)",
+              0) / $max_waves_per_cu) * 8) + MIN(MOD(ROUND(AVG(((4 * SQ_BUSY_CU_CYCLES) \
+              / GRBM_GUI_ACTIVE)), 0), $max_waves_per_cu), 8)), $cu_per_gpu))",
+    "kernelBusyCycles": "ROUND(AVG((((End_Timestamp - Start_Timestamp) / 1000) * $max_sclk)), 0)",
 }
 
 supported_call = {
@@ -414,8 +414,8 @@ def calc_builtin_var(var, sys_info):
     """
     if isinstance(var, int):
         return var
-    elif isinstance(var, str) and var.startswith("$totalL2Banks"):
-        return sys_info.totalL2Banks
+    elif isinstance(var, str) and var.startswith("$total_l2_chan"):
+        return sys_info.total_l2_chan
     else:
         print("Don't support", var)
         sys.exit(1)
@@ -679,20 +679,18 @@ def eval_metric(dfs, dfs_type, sys_info, raw_pmc_df, debug):
         print("WARNING: Dectected GRBM_GUI_ACTIVE == 0\nHaulting execution.")
         sys.exit(1)
 
-    ammolite__numSE = sys_info.SE
-    ammolite__numPipes = sys_info.numPipes
-    ammolite__numCU = sys_info.CU
-    ammolite__numSIMD = sys_info.SIMD
-    ammolite__numWavesPerCU = sys_info.max_waves_per_cu  # todo: check do we still need it
-    ammolite__numSQC = sys_info.numSQC
-    ammolite__L2Banks = sys_info.L2Banks
-    ammolite__LDSBanks = sys_info.LDSBanks
-    ammolite__freq = sys_info.cur_sclk  # todo: check do we still need it
-    ammolite__mclk = sys_info.cur_mclk
-    ammolite__sclk = sys_info.max_sclk
-    ammolite__maxWavesPerCU = sys_info.max_waves_per_cu
-    ammolite__hbmBW = sys_info.hbmBW
-    ammolite__totalL2Banks = calc_builtin_var("$totalL2Banks", sys_info)
+    ammolite__se_per_gpu = sys_info.se_per_gpu
+    ammolite__pipes_per_gpu = sys_info.pipes_per_gpu
+    ammolite__cu_per_gpu = sys_info.cu_per_gpu
+    ammolite__simd_per_cu = sys_info.simd_per_cu # not used
+    ammolite__sqc_per_gpu = sys_info.sqc_per_gpu
+    ammolite__lds_banks_per_cu = sys_info.lds_banks_per_cu
+    ammolite__cur_sclk = sys_info.cur_sclk  # not used
+    ammolite__mclk = sys_info.cur_mclk # not used
+    ammolite__max_sclk = sys_info.max_sclk
+    ammolite__max_waves_per_cu = sys_info.max_waves_per_cu
+    ammolite__hbm_bw = sys_info.hbm_bw
+    ammolite__total_l2_chan = calc_builtin_var("$total_l2_chan", sys_info)
 
     # TODO: fix all $normUnit in Unit column or title
 

projects/rocprofiler-compute/src/utils/specs.py

@@ -54,7 +54,7 @@ VERSION_LOC = [
 class MachineSpecs:
     def __init__(self, args, sysinfo=None):
         if not sysinfo is None:
-            self.arch = sysinfo.iloc[0]["arch"]
+            self.gpu_arch = sysinfo.iloc[0]["gpu_arch"]
             return
         # read timestamp info
         now = datetime.now()
@@ -76,16 +76,22 @@ class MachineSpecs:
         os_release = path("/etc/os-release").read_text()
         
         self.hostname: str = socket.gethostname()
-        self.CPU: str = search(r"^model name\s*: (.*?)$", cpuinfo)
+        self.cpu_model: str = search(r"^model name\s*: (.*?)$", cpuinfo)
         self.sbios: str = (
             path("/sys/class/dmi/id/bios_vendor").read_text().strip()
             + path("/sys/class/dmi/id/bios_version").read_text().strip()
         )
-        self.kernel_version: str = search(r"version (\S*)", version)
-        self.ram: str = search(r"MemTotal:\s*(\S*)", meminfo)
-        self.distro: str = search(r'PRETTY_NAME="(.*?)"', os_release)
-        if self.distro is None:
-            self.distro = ""
+        self.linux_kernel_version: str = search(r"version (\S*)", version)
+        self.amd_gpu_kernel_version: str = None
+        if self.amd_gpu_kernel_version is None: #TODO: Extract amdgpu kernel version
+            self.amd_gpu_kernel_version = ""
+        self.cpu_memory: str = search(r"MemTotal:\s*(\S*)", meminfo)
+        self.gpu_memory: str = None #TODO: Extract total gpu memory
+        if self.gpu_memory is None:
+            self.gpu_memory = ""
+        self.linux_distro: str = search(r'PRETTY_NAME="(.*?)"', os_release)
+        if self.linux_distro is None:
+            self.linux_distro = ""
         self.rocm_version: str = get_rocm_ver().strip()
         #FIXME: use device
         self.vbios: str = search(
@@ -105,12 +111,12 @@ class MachineSpecs:
         ##########################################
         ## B. SoC Specs
         ##########################################
-        self.arch: str = self.detect_arch()[0]
-        self.L1: str = None
-        self.L2: str = None
-        self.CU: str = None
-        self.SIMD: str = None
-        self.SE: str = None
+        self.gpu_arch: str = self.detect_arch()[0]
+        self.gpu_l1: str = None
+        self.gpu_l2: str = None
+        self.cu_per_gpu: str = None
+        self.simd_per_cu: str = None
+        self.se_per_gpu: str = None
         self.wave_size: str = None
         self.workgroup_max_size: str = None
         self.max_sclk: str = None
@@ -118,26 +124,26 @@ class MachineSpecs:
         self.cur_sclk: str = None
         self.cur_mclk: str = None
         self.max_waves_per_cu: str = None
-        self.GPU: str = None
-        self.L2Banks: str = None
-        self.LDSBanks: str = None
-        self.numSQC: str = None
-        self.numPipes: str = None
-        self.totalL2Banks: str = None
-        self.hbmBW: str = None
+        self.gpu_model: str = None
+        self.L2Banks: str = None #NB: This only used in flatten_tcc_info_across_hbm_stacks()
+        self.lds_banks_per_cu: str = None 
+        self.sqc_per_gpu: str = None
+        self.pipes_per_gpu: str = None
+        self.total_l2_chan: str = None
+        self.hbm_bw: str = None
         # Load above SoC specs via module import
         try:
-            soc_module = importlib.import_module('omniperf_soc.soc_'+ self.arch)
+            soc_module = importlib.import_module('omniperf_soc.soc_'+ self.gpu_arch)
         except ModuleNotFoundError as e:
-            error("Arch %s marked as supported, but couldn't find class implementation %s." % (self.arch, e))
-        soc_class = getattr(soc_module, self.arch+'_soc')
+            error("Arch %s marked as supported, but couldn't find class implementation %s." % (self.gpu_arch, e))
+        soc_class = getattr(soc_module, self.gpu_arch+'_soc')
         self._rocminfo = self._rocminfo[self.detect_arch()[1] + 1 :] # update rocminfo for target section
         soc_obj = soc_class(args, self)
         # Update arch specific specs
-        self.totalL2Banks: str = total_l2_banks(
-            self.GPU, int(self.L2Banks), self.memory_partition
+        self.total_l2_chan: str = total_l2_banks(
+            self.gpu_model, int(self.L2Banks), self.memory_partition
         )
-        self.hbmBW: str = str(int(self.max_mclk) / 1000 * 32 * self.get_hbm_channels())
+        self.hbm_bw: str = str(int(self.max_mclk) / 1000 * 32 * self.get_hbm_channels())
 
 
     def detect_arch(self):
@@ -156,10 +162,10 @@ class MachineSpecs:
             return (gpu_arch, idx1)
         
     def get_hbm_channels(self):
-        hbmchannels = int(self.totalL2Banks)
+        hbmchannels = int(self.total_l2_chan)
         if (
-            self.GPU.lower() == "mi300a_a0"
-            or self.GPU.lower() == "mi300a_a1"
+            self.gpu_model.lower() == "mi300a_a0"
+            or self.gpu_model.lower() == "mi300a_a1"
         ) and self.memory_partition.lower() == "nps1":
             # we have an extra 32 channels for the CCD
             hbmchannels += 32
@@ -174,12 +180,12 @@ class MachineSpecs:
                 attr_value = getattr(self, attr_name)
                 if attr_value is None:
                     #TODO: use proper logging function when that's merged
-                    logging.warning(f"WARNING: Incomplete class definition for {self.arch}. Expecting populated {attr_name} but detected None.")
+                    logging.warning(f"WARNING: Incomplete class definition for {self.gpu_arch}. Expecting populated {attr_name} but detected None.")
                     all_populated = False
                 data[attr_name] = attr_value
 
         if not all_populated:
-            error("Missing specs fields for %s" % self.arch)
+            error("Missing specs fields for %s" % self.gpu_arch)
         return pd.DataFrame(data, index=[0])
     
     
@@ -187,37 +193,38 @@ class MachineSpecs:
         return dedent(
             f"""\
         Host info:
-            hostname:       {self.hostname}
-            CPU:            {self.CPU}
-            sbios:          {self.sbios}
-            ram:            {self.ram}
-            distro:         {self.distro}
-            kernel_version: {self.kernel_version}
-            rocm_version:   {self.rocm_version}
+            Hostname:                   {self.hostname}
+            CPU Model:                  {self.cpu_model}
+            sbios:                      {self.sbios}
+            CPU Memory:                 {self.cpu_memory}
+            GPU Memory:                 {self.gpu_memory}
+            Linux Distro:               {self.linux_distro}
+            Linux Kernel Version:       {self.linux_kernel_version}
+            AMD GPU Kernel Version:     {self.amd_gpu_kernel_version}
+            ROCm Version:               {self.rocm_version}
         Device info:
-            GPU:                {self.GPU}
-            arch:               {self.arch}
-            vbios:              {self.vbios}
-            L1:                 {self.L1} KB
-            L2:                 {self.L2} KB
-            max_sclk:           {self.max_sclk} MHz
-            max_mclk:           {self.max_mclk} MHz
-            cur_sclk:           {self.cur_sclk} MHz
-            cur_mclk:           {self.cur_mclk} MHz
-            CU:                 {self.CU}
-            SIMD:               {self.SIMD}
-            SE:                 {self.SE}
-            wave_size:          {self.wave_size}
-            workgroup_max_size: {self.workgroup_max_size}
-            max_waves_per_cu:   {self.max_waves_per_cu}
-            L2Banks:            {self.L2Banks}
-            totalL2Banks:       {self.totalL2Banks}
-            LDSBanks:           {self.LDSBanks}
-            numSQC:             {self.numSQC}
-            numPipes:           {self.numPipes}
-            hbmBW:              {self.hbmBW} MB/s
-            compute_partition:  {self.compute_partition}
-            memory_partition:   {self.memory_partition}
+            GPU Model:                  {self.gpu_model}
+            GPU Arch:                   {self.gpu_arch}
+            vbios:                      {self.vbios}
+            GPU L1:                     {self.gpu_l1} KB
+            GPU L2:                     {self.gpu_l2} KB
+            Max SCLK:                   {self.max_sclk} MHz
+            Max MCLK:                   {self.max_mclk} MHz
+            Cur SCLK:                   {self.cur_sclk} MHz
+            Cur MCLK:                   {self.cur_mclk} MHz
+            CU per GPU:                 {self.cu_per_gpu}
+            SIMD per CU:                {self.simd_per_cu}
+            SE per GPU:                 {self.se_per_gpu}
+            Wave Size:                  {self.wave_size}
+            Workgroup Max Size:         {self.workgroup_max_size}
+            Max Waves per CU:           {self.max_waves_per_cu}
+            Total L2 Channels:          {self.total_l2_chan}
+            LDS Banks per CU:           {self.lds_banks_per_cu}
+            SQC per GPU:                {self.sqc_per_gpu}
+            Pipes per GPU:              {self.pipes_per_gpu}
+            HBM BW:                     {self.hbm_bw} MB/s
+            Compute Partition:          {self.compute_partition}
+            Memory Partition:           {self.memory_partition}
         """
         )
 

projects/rocprofiler-compute/src/utils/utils.py

@@ -210,7 +210,7 @@ def run_prof(fname, profiler_options, workload_dir, mspec):
 
     # set required env var for mi300
     new_env = None
-    if  (mspec.GPU.lower() == "mi300x_a0" or mspec.GPU.lower() == "mi300x_a1" or mspec.GPU.lower() == "mi300a_a0" or mspec.GPU.lower() == "mi300a_a1") and (
+    if  (mspec.gpu_model.lower() == "mi300x_a0" or mspec.gpu_model.lower() == "mi300x_a1" or mspec.gpu_model.lower() == "mi300a_a0" or mspec.gpu_model.lower() == "mi300a_a1") and (
         os.path.basename(fname) == "pmc_perf_13.txt"
         or os.path.basename(fname) == "pmc_perf_14.txt"
         or os.path.basename(fname) == "pmc_perf_15.txt"
@@ -232,7 +232,7 @@ def run_prof(fname, profiler_options, workload_dir, mspec):
     if new_env:
         # flatten tcc for applicable mi300 input
         f = path(workload_dir + "/out/pmc_1/results_" + fbase + ".csv")
-        hbm_stack_num = get_hbm_stack_num(mspec.GPU, mspec.memory_partition)
+        hbm_stack_num = get_hbm_stack_num(mspec.gpu_model, mspec.memory_partition)
         df = flatten_tcc_info_across_hbm_stacks(
             f, hbm_stack_num, int(mspec.L2Banks)
         )
@@ -308,7 +308,7 @@ def gen_sysinfo(workload_name, workload_dir, ip_blocks, app_cmd, skip_roof, roof
         blocks += t
     else:
         blocks += ip_blocks
-    if mspec.arch == "gfx90a" and (not skip_roof):
+    if mspec.gpu_arch == "gfx90a" and (not skip_roof):
         blocks.append("roofline")
     df['ip_blocks'] = "|".join(blocks)
 
@@ -396,7 +396,7 @@ def mibench(args, mspec):
             + "-"
             + distro_map[target_binary["distro"]]
             + "-"
-            + mspec.GPU.lower()
+            + mspec.gpu_model.lower()
             + "-rocm"
             + target_binary["rocm_ver"]
         )