From e7b96b1e71108dbf7c818e02f7a90edbb909778f Mon Sep 17 00:00:00 2001
From: jatang <jatang@amd.com>
Date: Tue, 7 May 2024 15:40:56 -0400
Subject: [PATCH] SWDEV-458392 - Add gfx12 support.

Change-Id: I91bb6a3329bf77f26005a345c18b63b86922028a
---
 build.sh                                      |  2 +-
 src/core/counters/derived/metrics.xml         | 27 +++++++++-
 src/core/counters/metrics/basic_counters.xml  | 46 +++++++++++++++++
 .../counters/metrics/derived_counters.xml     | 32 ++++++++++++
 src/core/counters/metrics/metrics.h           |  7 +--
 src/core/metrics.h                            |  3 +-
 test/tool/gfx_metrics.xml                     | 49 +++++++++++++++++++
 test/tool/metrics.xml                         | 33 +++++++++++++
 tests-v2/featuretests/profiler/CMakeLists.txt |  2 +-
 .../featuretests/profiler/profiler_gtest.cpp  |  9 ++--
 10 files changed, 199 insertions(+), 11 deletions(-)

diff --git a/build.sh b/build.sh
index 8cfcbe21cf..0ce2e8320e 100755
--- a/build.sh
+++ b/build.sh
@@ -75,7 +75,7 @@ if [ -n "$ROCM_RPATH" ] ; then LD_RUNPATH_FLAG=" -Wl,--enable-new-dtags -Wl,--rp
 if [ -z "$TO_CLEAN" ] ; then TO_CLEAN=yes; fi
 if [ -z "$RUN_TEST" ] ; then RUN_TEST=no; fi
 if [ -z "$ASAN" ] ; then ASAN=False; fi
-if [ -z "$GPU_LIST" ] ; then GPU_LIST="gfx900 gfx906 gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100 gfx1101 gfx1102 gfx1031"; fi
+if [ -z "$GPU_LIST" ] ; then GPU_LIST="gfx900 gfx906 gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100 gfx1101 gfx1102 gfx1031 gfx1200 gfx1201"; fi
 
 ROCPROFILER_ROOT=$(cd $ROCPROFILER_ROOT && echo $PWD)
 
diff --git a/src/core/counters/derived/metrics.xml b/src/core/counters/derived/metrics.xml
index edc12850ad..db60b778f6 100755
--- a/src/core/counters/derived/metrics.xml
+++ b/src/core/counters/derived/metrics.xml
@@ -462,4 +462,29 @@
     <metric name="LDSBankConflict" expr="100*SQC_LDS_BANK_CONFLICT/SQC_LDS_IDX_ACTIVE" descr="The percentage of GPUTime LDS is stalled by bank conflicts. Value range: 0% (optimal) to 100% (bad)."></metric>
   </gfx11>
 
-</metrics>
\ No newline at end of file
+  <gfx12>
+    <metric name="GPU_UTIL" expr="100*GRBM_GUI_ACTIVE/GRBM_COUNT" descr="Percentage of the time that GUI is active"></metric>
+    <metric name="WAVE_DEP_WAIT" expr="100*SQ_WAIT_ANY/SQ_WAVE_CYCLES" descr="Percentage of the SQ_WAVE_CYCLE time spent waiting for anything."></metric>
+    <metric name="WAVE_ISSUE_WAIT" expr="100*SQ_WAIT_INST_ANY/SQ_WAVE_CYCLES" descr="Percentage of the SQ_WAVE_CYCLE time spent waiting for any instruction issue."></metric>
+    <metric name="TA_BUSY_avr" expr="avr(TA_TA_BUSY,16)" descr="TA block is busy. Average over TA instances."></metric>
+    <metric name="TA_BUSY_max" expr="max(TA_TA_BUSY,16)" descr="TA block is busy. Max over TA instances."></metric>
+    <metric name="TA_BUSY_min" expr="min(TA_TA_BUSY,16)" descr="TA block is busy. Min over TA instances."></metric>
+    <metric name="TA_BUFFER_LOAD_WAVEFRONTS_sum" expr="sum(TA_BUFFER_LOAD_WAVEFRONTS,16)" descr="Number of buffer load vec32 packets processed by the TA. Sum over TA instances."></metric>
+    <metric name="TA_BUFFER_STORE_WAVEFRONTS_sum" expr="sum(TA_BUFFER_STORE_WAVEFRONTS,16)" descr="Number of buffer store vec32 packets processed by the TA. Sum over TA instances."></metric>
+    <metric name="GL2C_HIT_sum" expr="sum(GL2C_HIT,16)" descr="Number of cache hits. Sum over GL2C instances."></metric>
+    <metric name="GL2C_MISS_sum" expr="sum(GL2C_MISS,16)" descr="Number of cache misses. Sum over GL2C instances."></metric>
+    <metric name="GL2C_EA_RDREQ_32B_sum" expr="sum(GL2C_EA_RDREQ_32B,16)" descr="Number of 32-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+    <metric name="GL2C_EA_RDREQ_64B_sum" expr="sum(GL2C_EA_RDREQ_64B,16)" descr="Number of 64-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+    <metric name="GL2C_EA_RDREQ_96B_sum" expr="sum(GL2C_EA_RDREQ_96B,16)" descr="Number of 96-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+    <metric name="GL2C_EA_RDREQ_128B_sum" expr="sum(GL2C_EA_RDREQ_128B,16)" descr="Number of 128-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+    <metric name="GL2C_MC_RDREQ_sum" expr="sum(GL2C_MC_RDREQ,16)" descr="Number of GL2C/EA read requests (either 32-byte or 64-byte or 128-byte). Sum over GL2C instances."></metric>
+    <metric name="GL2C_MC_WRREQ_sum" expr="sum(GL2C_MC_WRREQ,16)" descr="Number of transactions (either 32-byte or 64-byte) going over the GL2C_MC_wrreq interface. Sum over GL2C instances."></metric>
+    <metric name="GL2C_EA_WRREQ_64B_sum" expr="sum(GL2C_EA_WRREQ_64B,16)" descr="Number of 64-byte transactions going (64-byte write or CMPSWAP) over the GL2C_EA_wrreq interface. Sum over GL2C instances."></metric>
+    <metric name="GL2C_WRREQ_STALL_max" expr="max(GL2C_MC_WRREQ_STALL,16)" descr="Number of cycles a write request was stalled. Max over GL2C instances."></metric>
+    <metric name="L2CacheHit" expr="100*sum(GL2C_HIT,16)/(sum(GL2C_HIT,16)+sum(GL2C_MISS,16))" descr="The percentage of fetch, write, atomic, and other instructions that hit the data in L2 cache. Value range: 0% (no hit) to 100% (optimal)."></metric>
+    <metric name="FETCH_SIZE" expr="(GL2C_EA_RDREQ_32B_sum*32+GL2C_EA_RDREQ_64B_sum*64+GL2C_EA_RDREQ_96B_sum*96+GL2C_EA_RDREQ_128B_sum*128)/1024" descr="The total kilobytes fetched from the video memory. This is measured with all extra fetches and any cache or memory effects taken into account."></metric>
+    <metric name="WriteUnitStalled" expr="100*GL2C_WRREQ_STALL_max/GRBM_GUI_ACTIVE" descr="The percentage of GPUTime the Write unit is stalled. Value range: 0% to 100% (bad)."></metric>
+    <metric name="LDSBankConflict" expr="100*SQC_LDS_BANK_CONFLICT/SQC_LDS_IDX_ACTIVE" descr="The percentage of GPUTime LDS is stalled by bank conflicts. Value range: 0% (optimal) to 100% (bad)."></metric>
+  </gfx12>
+
+</metrics>
diff --git a/src/core/counters/metrics/basic_counters.xml b/src/core/counters/metrics/basic_counters.xml
index 3fc2efc294..22c8680087 100755
--- a/src/core/counters/metrics/basic_counters.xml
+++ b/src/core/counters/metrics/basic_counters.xml
@@ -737,3 +737,49 @@
 <gfx1101 base="gfx11"></gfx1101>
 <gfx1102 base="gfx11"></gfx1102>
 
+<gfx12>
+  <metric name="GRBM_COUNT" block=GRBM event=0 descr="Tie High - Count Number of Clocks"></metric>
+  <metric name="GRBM_GUI_ACTIVE" block=GRBM event=2 descr="The GUI is Active"></metric>
+  <metric name="GL2C_HIT" block=GL2C event=42 descr="Number of cache hits"></metric>
+  <metric name="GL2C_MISS" block=GL2C event=43 descr="Number of cache misses.  UC reads count as misses."></metric>
+  <metric name="GL2C_MC_WRREQ" block=GL2C event=83 descr="Number of transactions (either 32-byte or 64-byte) going over the GL2C_EA_wrreq interface. Atomics may travel over the same interface and are generally classified as write requests. This does not include probe commands"></metric>
+  <metric name="GL2C_EA_WRREQ_64B" block=GL2C event=85 descr="Number of 64-byte transactions going (64-byte write or CMPSWAP) over the TC_EA_wrreq interface."></metric>
+  <metric name="GL2C_MC_WRREQ_STALL" block=GL2C event=88 descr="Number of cycles a write request was stalled."></metric>
+  <metric name="GL2C_MC_RDREQ" block=GL2C event=96 descr="Number of GL2C/EA read requests (either 32-byte or 64-byte or 128-byte)."></metric>
+  <metric name="GL2C_EA_RDREQ_32B" block=GL2C event=99 descr="Number of 32-byte GL2C/EA read requests"></metric>
+  <metric name="GL2C_EA_RDREQ_64B" block=GL2C event=100 descr="Number of 64-byte GL2C/EA read requests"></metric>
+  <metric name="GL2C_EA_RDREQ_96B" block=GL2C event=101 descr="Number of 96-byte GL2C/EA read requests"></metric>
+  <metric name="GL2C_EA_RDREQ_128B" block=GL2C event=102 descr="Number of 128-byte GL2C/EA read requests"></metric>
+  <metric name="SQ_ACCUM_PREV" block=SQ event=1 descr="For counter N, increment by the value of counter N-1."></metric>
+  <metric name="SQ_BUSY_CYCLES" block=SQ event=3 descr="Clock cycles while SQ is reporting that it is busy. {nondeterministic, global, C2}"></metric>
+  <metric name="SQ_WAVES" block=SQ event=4 descr="Count number of waves sent to SQs. {emulated, global, C1}"></metric>
+  <metric name="SQ_WAVE_CYCLES" block=SQ event=24 descr="Number of clock cycles spent by waves in the SQs. Incremented by number of living (valid) waves each cycle. {nondeterministic, C1}"></metric>
+  <metric name="SQ_WAIT_INST_ANY" block=SQ event=26 descr="Number of clock-cycles spent waiting for any instruction issue. In units of cycles. (nondeterministic)"></metric>
+  <metric name="SQ_WAIT_ANY" block=SQ event=35 descr="Number of wave-cycles spent waiting for anything (nondeterministic, C1)"></metric>
+  <metric name="SQ_INSTS_WAVE32" block=SQ event=70 descr="Number of wave32 instructions issued, for flat, lds, valu, tex. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_WAVE32_LDS" block=SQ event=72 descr="Number of wave32 LDS indexed instructions issued. Wave64 may count 1 or 2, depending on what gets issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_WAVE32_VALU" block=SQ event=73 descr="Number of wave32 valu instructions issued. Wave64 may count 1 or 2, depending on what gets issued. {emulated, C1}"></metric>
+  <metric name="SQ_WAVE32_INSTS" block=SQ event=82 descr="Number of instructions issued by wave32 waves. Skipped instructions are not counted. {emulated}"></metric>
+  <metric name="SQ_WAVE64_INSTS" block=SQ event=83 descr="Number of instructions issued by wave64 waves. Skipped instructions are not counted. {emulated}"></metric>
+  <metric name="SQ_INST_LEVEL_GDS" block=SQ event=87 descr="Number of in-flight GDS instructions. Set next counter to ACCUM_PREV and divide by INSTS_GDS for average latency. {level, nondeterministic, C1}"></metric>
+  <metric name="SQ_INST_LEVEL_LDS" block=SQ event=88 descr="Number of in-flight LDS instructions. Set next counter to ACCUM_PREV and divide by INSTS_LDS for average latency. Includes FLAT instructions. {level, nondeterministic, C1}"></metric>
+  <metric name="SQ_INST_CYCLES_VMEM" block=SQ event=106 descr="Number of cycles needed to send addr and data for VMEM (lds, buffer, image, flat, scratch, global) instructions, windowed by perf_en. {emulated, C1}"></metric>
+  <metric name="SQC_LDS_BANK_CONFLICT" block=SQ event=256 descr="Number of cycles LDS is stalled by bank conflicts. (emulated, C1)"></metric>
+  <metric name="SQC_LDS_IDX_ACTIVE" block=SQ event=261 descr="Number of cycles LDS is used for indexed (non-direct,non-interpolation) operations. {per-simd, emulated, C1}"></metric>
+  <metric name="SQ_INSTS_VALU" block=SQ event=62 descr="Number of VALU instructions issued excluding skipped instructions. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_SALU" block=SQ event=58 descr="Number of SALU instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_SMEM" block=SQ event=59 descr="Number of SMEM instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_FLAT" block=SQ event=56 descr="Number of FLAT instructions issued. {emulated, C2}"></metric>
+  <metric name="SQ_INSTS_LDS" block=SQ event=57 descr="Number of LDS indexed instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_GDS" block=SQ event=54 descr="Number of GDS instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_TEX_LOAD" block=SQ event=66 descr="Number of buffer load, image load, sample, or atomic (with return) instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_TEX_STORE" block=SQ event=67 descr="Number of buffer store, image store, or atomic (without return) instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_WAIT_INST_LDS" block=SQ event=29 descr="Number of clock cycles spent waiting for LDS (indexed) instruction issue. In units of cycles. {nondeterministic, C1}"></metric>
+  <metric name="TA_TA_BUSY" block=TA event=15 descr="TA block is busy. Perf_Windowing not supported for this counter."></metric>
+  <metric name="TA_BUFFER_LOAD_WAVEFRONTS" block=TA event=45 descr="Number of buffer load vec32 packets processed by TA"></metric>
+  <metric name="TA_BUFFER_STORE_WAVEFRONTS" block=TA event=46 descr="Number of buffer store vec32 packets processed by TA"></metric>
+</gfx12>
+
+<gfx1200 base="gfx12"></gfx1200>
+<gfx1201 base="gfx12"></gfx1201>
+
diff --git a/src/core/counters/metrics/derived_counters.xml b/src/core/counters/metrics/derived_counters.xml
index d9217bc50d..2a873cf2dd 100755
--- a/src/core/counters/metrics/derived_counters.xml
+++ b/src/core/counters/metrics/derived_counters.xml
@@ -555,6 +555,35 @@
   <metric name="LDSBankConflict" expr=100*SQC_LDS_BANK_CONFLICT/SQC_LDS_IDX_ACTIVE descr="The percentage of GPUTime LDS is stalled by bank conflicts. Value range: 0% (optimal) to 100% (bad)."></metric>
 </gfx11_expr>
 
+<gfx12_expr>
+  <metric name="KERNEL_DURATION" expr=1 descr="The duration of the kernel dispatch"></metric>
+  <metric name="GPU_UTIL" expr=100*GRBM_GUI_ACTIVE/GRBM_COUNT descr="Percentage of the time that GUI is active"></metric>
+  <metric name="WAVE_DEP_WAIT" expr=100*SQ_WAIT_ANY/SQ_WAVE_CYCLES descr="Percentage of the SQ_WAVE_CYCLE time spent waiting for anything."></metric>
+  <metric name="WAVE_ISSUE_WAIT" expr=100*SQ_WAIT_INST_ANY/SQ_WAVE_CYCLES descr="Percentage of the SQ_WAVE_CYCLE time spent waiting for any instruction issue."></metric>
+  <metric name="TA_BUSY_avr" expr=avr(TA_TA_BUSY,16) descr="TA block is busy. Average over TA instances."></metric>
+  <metric name="TA_BUSY_max" expr=max(TA_TA_BUSY,16) descr="TA block is busy. Max over TA instances."></metric>
+  <metric name="TA_BUSY_min" expr=min(TA_TA_BUSY,16) descr="TA block is busy. Min over TA instances."></metric>
+  <metric name="TA_BUFFER_LOAD_WAVEFRONTS_sum" expr=sum(TA_BUFFER_LOAD_WAVEFRONTS,16) descr="Number of buffer load vec32 packets processed by the TA. Sum over TA instances."></metric>
+  <metric name="TA_BUFFER_STORE_WAVEFRONTS_sum" expr=sum(TA_BUFFER_STORE_WAVEFRONTS,16) descr="Number of buffer store vec32 packets processed by the TA. Sum over TA instances."></metric>
+  <metric name="GL2C_HIT_sum" expr=sum(GL2C_HIT,32) descr="Number of cache hits. Sum over GL2C instances."></metric>
+  <metric name="GL2C_MISS_sum" expr=sum(GL2C_MISS,32) descr="Number of cache misses. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_32B_sum" expr=sum(GL2C_EA_RDREQ_32B,32) descr="Number of 32-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_64B_sum" expr=sum(GL2C_EA_RDREQ_64B,32) descr="Number of 64-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_96B_sum" expr=sum(GL2C_EA_RDREQ_96B,32) descr="Number of 96-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_128B_sum" expr=sum(GL2C_EA_RDREQ_128B,32) descr="Number of 128-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_MC_RDREQ_sum" expr=sum(GL2C_MC_RDREQ,32) descr="Number of GL2C/EA read requests (either 32-byte or 64-byte or 128-byte). Sum over GL2C instances."></metric>
+  <metric name="GL2C_MC_WRREQ_sum" expr=sum(GL2C_MC_WRREQ,32) descr="Number of transactions (either 32-byte or 64-byte) going over the GL2C_MC_wrreq interface. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_WRREQ_64B_sum" expr=sum(GL2C_EA_WRREQ_64B,32) descr="Number of 64-byte transactions going (64-byte write or CMPSWAP) over the GL2C_EA_wrreq interface. Sum over GL2C instances."></metric>
+  <metric name="GL2C_WRREQ_STALL_max" expr=max(GL2C_MC_WRREQ_STALL,32) descr="Number of cycles a write request was stalled. Max over GL2C instances."></metric>
+  <metric name="L2CacheHit" expr=100*sum(GL2C_HIT,32)/(sum(GL2C_HIT,32)+sum(GL2C_MISS,32)) descr="The percentage of fetch, write, atomic, and other instructions that hit the data in L2 cache. Value range: 0% (no hit) to 100% (optimal)."></metric>
+  <metric name="FETCH_SIZE" expr=(GL2C_EA_RDREQ_32B_sum*32+GL2C_EA_RDREQ_64B_sum*64+GL2C_EA_RDREQ_96B_sum*96+GL2C_EA_RDREQ_128B_sum*128)/1024 descr="The total kilobytes fetched from the video memory. This is measured with all extra fetches and any cache or memory effects taken into account."></metric>
+  <metric name="WriteUnitStalled" expr=100*GL2C_WRREQ_STALL_max/GRBM_GUI_ACTIVE descr="The percentage of GPUTime the Write unit is stalled. Value range: 0% to 100% (bad)."></metric>
+  <metric name="LDSBankConflict" expr=100*SQC_LDS_BANK_CONFLICT/SQC_LDS_IDX_ACTIVE descr="The percentage of GPUTime LDS is stalled by bank conflicts. Value range: 0% (optimal) to 100% (bad)."></metric>
+</gfx12_expr>
+
+<gfx1200_expr base="gfx12_expr"></gfx1200_expr>
+<gfx1201_expr base="gfx12_expr"></gfx1201_expr>
+
 <gfx1100_expr base="gfx11_expr"></gfx1100_expr>
 <gfx1101_expr base="gfx11_expr"></gfx1101_expr>
 <gfx1102_expr base="gfx11_expr"></gfx1102_expr>
@@ -581,6 +610,9 @@
 <gfx1100 base="gfx1100_expr"></gfx1100>
 <gfx1101 base="gfx1101_expr"></gfx1101>
 <gfx1102 base="gfx1102_expr"></gfx1102>
+#Navi4x
+<gfx1200 base="gfx1200_expr"></gfx1200>
+<gfx1201 base="gfx1201_expr"></gfx1201>
 
 
 <global>
diff --git a/src/core/counters/metrics/metrics.h b/src/core/counters/metrics/metrics.h
index 609e126bab..78a92d2294 100644
--- a/src/core/counters/metrics/metrics.h
+++ b/src/core/counters/metrics/metrics.h
@@ -225,10 +225,11 @@ class MetricsDict {
       agent_name_ = agent_name_.substr(0, agent_name_.find(':'));
 
     std::unordered_set<std::string> supported_agent_names = {
-        "gfx906",  "gfx908", "gfx90a",  // Vega
-        "gfx940",  "gfx941", "gfx942",  // Mi300
+        "gfx906",  "gfx908", "gfx90a",    // Vega
+        "gfx940",  "gfx941", "gfx942",    // Mi300
         "gfx1030", "gfx1031", "gfx1032",  // Navi2x
-        "gfx1100", "gfx1101", "gfx1102"  // Navi3x
+        "gfx1100", "gfx1101", "gfx1102",  // Navi3x
+        "gfx1200", "gfx1201",             // Navi4x
     };
     if (supported_agent_names.find(agent_name_) != supported_agent_names.end()) {
       ImportMetrics(agent_info, agent_name_);
diff --git a/src/core/metrics.h b/src/core/metrics.h
index 2a5b8f37ef..2b8f4286d8 100644
--- a/src/core/metrics.h
+++ b/src/core/metrics.h
@@ -211,7 +211,8 @@ class MetricsDict {
           "gfx906",  "gfx908",  "gfx90a",   // Vega
           "gfx940",  "gfx941",  "gfx942",   // Mi300
           "gfx1030", "gfx1031", "gfx1032",  // Navi2x
-          "gfx1100", "gfx1101", "gfx1102"   // Navi3x
+          "gfx1100", "gfx1101", "gfx1102",  // Navi3x
+          "gfx1200", "gfx1201",             // Navi4x
       };
       if (supported_agent_names.find(agent_name_) != supported_agent_names.end()) {
         ImportMetrics(agent_info, agent_name_);
diff --git a/test/tool/gfx_metrics.xml b/test/tool/gfx_metrics.xml
index 81dc9dde74..f915b653a0 100755
--- a/test/tool/gfx_metrics.xml
+++ b/test/tool/gfx_metrics.xml
@@ -741,3 +741,52 @@
 <gfx1101 base="gfx11"></gfx1101>
 <gfx1102 base="gfx11"></gfx1102>
 
+<gfx12>
+  <metric name="GRBM_COUNT" block=GRBM event=0 descr="Tie High - Count Number of Clocks"></metric>
+  <metric name="GRBM_GUI_ACTIVE" block=GRBM event=2 descr="The GUI is Active"></metric>
+
+  <metric name="GL2C_HIT" block=GL2C event=42 descr="Number of cache hits"></metric>
+  <metric name="GL2C_MISS" block=GL2C event=43 descr="Number of cache misses.  UC reads count as misses."></metric>
+  <metric name="GL2C_MC_WRREQ" block=GL2C event=83 descr="Number of transactions (either 32-byte or 64-byte) going over the GL2C_EA_wrreq interface. Atomics may travel over the same interface and are generally classified as write requests. This does not include probe commands"></metric>
+  <metric name="GL2C_EA_WRREQ_64B" block=GL2C event=85 descr="Number of 64-byte transactions going (64-byte write or CMPSWAP) over the TC_EA_wrreq interface."></metric>
+  <metric name="GL2C_MC_WRREQ_STALL" block=GL2C event=88 descr="Number of cycles a write request was stalled."></metric>
+  <metric name="GL2C_MC_RDREQ" block=GL2C event=96 descr="Number of GL2C/EA read requests (either 32-byte or 64-byte or 128-byte)."></metric>
+  <metric name="GL2C_EA_RDREQ_32B" block=GL2C event=99 descr="Number of 32-byte GL2C/EA read requests"></metric>
+  <metric name="GL2C_EA_RDREQ_64B" block=GL2C event=100 descr="Number of 64-byte GL2C/EA read requests"></metric>
+  <metric name="GL2C_EA_RDREQ_96B" block=GL2C event=101 descr="Number of 96-byte GL2C/EA read requests"></metric>
+  <metric name="GL2C_EA_RDREQ_128B" block=GL2C event=102 descr="Number of 128-byte GL2C/EA read requests"></metric>
+
+  <metric name="SQ_ACCUM_PREV" block=SQ event=1 descr="For counter N, increment by the value of counter N-1."></metric>
+  <metric name="SQ_BUSY_CYCLES" block=SQ event=3 descr="Clock cycles while SQ is reporting that it is busy. {nondeterministic, global, C2}"></metric>
+  <metric name="SQ_WAVES" block=SQ event=4 descr="Count number of waves sent to SQs. {emulated, global, C1}"></metric>
+  <metric name="SQ_WAVE_CYCLES" block=SQ event=24 descr="Number of clock cycles spent by waves in the SQs. Incremented by number of living (valid) waves each cycle. {nondeterministic, C1}"></metric>
+  <metric name="SQ_WAIT_INST_ANY" block=SQ event=26 descr="Number of clock-cycles spent waiting for any instruction issue. In units of cycles. (nondeterministic)"></metric>
+  <metric name="SQ_WAIT_ANY" block=SQ event=35 descr="Number of wave-cycles spent waiting for anything (nondeterministic, C1)"></metric>
+  <metric name="SQ_INSTS_WAVE32" block=SQ event=70 descr="Number of wave32 instructions issued, for flat, lds, valu, tex. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_WAVE32_LDS" block=SQ event=72 descr="Number of wave32 LDS indexed instructions issued. Wave64 may count 1 or 2, depending on what gets issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_WAVE32_VALU" block=SQ event=73 descr="Number of wave32 valu instructions issued. Wave64 may count 1 or 2, depending on what gets issued. {emulated, C1}"></metric>
+  <metric name="SQ_WAVE32_INSTS" block=SQ event=82 descr="Number of instructions issued by wave32 waves. Skipped instructions are not counted. {emulated}"></metric>
+  <metric name="SQ_WAVE64_INSTS" block=SQ event=83 descr="Number of instructions issued by wave64 waves. Skipped instructions are not counted. {emulated}"></metric>
+  <metric name="SQ_INST_LEVEL_GDS" block=SQ event=87 descr="Number of in-flight GDS instructions. Set next counter to ACCUM_PREV and divide by INSTS_GDS for average latency. {level, nondeterministic, C1}"></metric>
+  <metric name="SQ_INST_LEVEL_LDS" block=SQ event=88 descr="Number of in-flight LDS instructions. Set next counter to ACCUM_PREV and divide by INSTS_LDS for average latency. Includes FLAT instructions. {level, nondeterministic, C1}"></metric>
+  <metric name="SQ_INST_CYCLES_VMEM" block=SQ event=106 descr="Number of cycles needed to send addr and data for VMEM (lds, buffer, image, flat, scratch, global) instructions, windowed by perf_en. {emulated, C1}"></metric>
+  <metric name="SQC_LDS_BANK_CONFLICT" block=SQ event=256 descr="Number of cycles LDS is stalled by bank conflicts. (emulated, C1)"></metric>
+  <metric name="SQC_LDS_IDX_ACTIVE" block=SQ event=261 descr="Number of cycles LDS is used for indexed (non-direct,non-interpolation) operations. {per-simd, emulated, C1}"></metric>
+  <metric name="SQ_INSTS_VALU" block=SQ event=62 descr="Number of VALU instructions issued excluding skipped instructions. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_SALU" block=SQ event=58 descr="Number of SALU instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_SMEM" block=SQ event=59 descr="Number of SMEM instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_FLAT" block=SQ event=56 descr="Number of FLAT instructions issued. {emulated, C2}"></metric>
+  <metric name="SQ_INSTS_LDS" block=SQ event=57 descr="Number of LDS indexed instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_GDS" block=SQ event=54 descr="Number of GDS instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_TEX_LOAD" block=SQ event=66 descr="Number of buffer load, image load, sample, or atomic (with return) instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_INSTS_TEX_STORE" block=SQ event=67 descr="Number of buffer store, image store, or atomic (without return) instructions issued. {emulated, C1}"></metric>
+  <metric name="SQ_WAIT_INST_LDS" block=SQ event=29 descr="Number of clock cycles spent waiting for LDS (indexed) instruction issue. In units of cycles. {nondeterministic, C1}"></metric>
+  <metric name="TA_TA_BUSY" block=TA event=15 descr="TA block is busy. Perf_Windowing not supported for this counter."></metric>
+  <metric name="TA_BUFFER_LOAD_WAVEFRONTS" block=TA event=45 descr="Number of buffer load vec32 packets processed by TA"></metric>
+  <metric name="TA_BUFFER_STORE_WAVEFRONTS" block=TA event=46 descr="Number of buffer store vec32 packets processed by TA"></metric>
+
+</gfx12>
+
+<gfx1200 base="gfx12"></gfx1200>
+<gfx1201 base="gfx12"></gfx1201>
+
diff --git a/test/tool/metrics.xml b/test/tool/metrics.xml
index 75c303077c..8b63912e1c 100755
--- a/test/tool/metrics.xml
+++ b/test/tool/metrics.xml
@@ -422,6 +422,36 @@
   <metric name="LDSBankConflict" expr=100*SQC_LDS_BANK_CONFLICT/SQC_LDS_IDX_ACTIVE descr="The percentage of GPUTime LDS is stalled by bank conflicts. Value range: 0% (optimal) to 100% (bad)."></metric>
 </gfx11_expr>
 
+<gfx12_expr>
+  <metric name="GPU_UTIL" expr=100*GRBM_GUI_ACTIVE/GRBM_COUNT descr="Percentage of the time that GUI is active"></metric>
+  <metric name="WAVE_DEP_WAIT" expr=100*SQ_WAIT_ANY/SQ_WAVE_CYCLES descr="Percentage of the SQ_WAVE_CYCLE time spent waiting for anything."></metric>
+  <metric name="WAVE_ISSUE_WAIT" expr=100*SQ_WAIT_INST_ANY/SQ_WAVE_CYCLES descr="Percentage of the SQ_WAVE_CYCLE time spent waiting for any instruction issue."></metric>
+
+  <metric name="TA_BUSY_avr" expr=avr(TA_TA_BUSY,16) descr="TA block is busy. Average over TA instances."></metric>
+  <metric name="TA_BUSY_max" expr=max(TA_TA_BUSY,16) descr="TA block is busy. Max over TA instances."></metric>
+  <metric name="TA_BUSY_min" expr=min(TA_TA_BUSY,16) descr="TA block is busy. Min over TA instances."></metric>
+  <metric name="TA_BUFFER_LOAD_WAVEFRONTS_sum" expr=sum(TA_BUFFER_LOAD_WAVEFRONTS,16) descr="Number of buffer load vec32 packets processed by the TA. Sum over TA instances."></metric>
+  <metric name="TA_BUFFER_STORE_WAVEFRONTS_sum" expr=sum(TA_BUFFER_STORE_WAVEFRONTS,16) descr="Number of buffer store vec32 packets processed by the TA. Sum over TA instances."></metric>
+
+  <metric name="GL2C_HIT_sum" expr=sum(GL2C_HIT,16) descr="Number of cache hits. Sum over GL2C instances."></metric>
+  <metric name="GL2C_MISS_sum" expr=sum(GL2C_MISS,16) descr="Number of cache misses. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_32B_sum" expr=sum(GL2C_EA_RDREQ_32B,16) descr="Number of 32-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_64B_sum" expr=sum(GL2C_EA_RDREQ_64B,16) descr="Number of 64-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_96B_sum" expr=sum(GL2C_EA_RDREQ_96B,16) descr="Number of 96-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_RDREQ_128B_sum" expr=sum(GL2C_EA_RDREQ_128B,16) descr="Number of 128-byte GL2C/EA read requests. Sum over GL2C instances."></metric>
+  <metric name="GL2C_MC_RDREQ_sum" expr=sum(GL2C_MC_RDREQ,16) descr="Number of GL2C/EA read requests (either 32-byte or 64-byte or 128-byte). Sum over GL2C instances."></metric>
+  <metric name="GL2C_MC_WRREQ_sum" expr=sum(GL2C_MC_WRREQ,16) descr="Number of transactions (either 32-byte or 64-byte) going over the GL2C_MC_wrreq interface. Sum over GL2C instances."></metric>
+  <metric name="GL2C_EA_WRREQ_64B_sum" expr=sum(GL2C_EA_WRREQ_64B,16) descr="Number of 64-byte transactions going (64-byte write or CMPSWAP) over the GL2C_EA_wrreq interface. Sum over GL2C instances."></metric>
+  <metric name="GL2C_WRREQ_STALL_max" expr=max(GL2C_MC_WRREQ_STALL,16) descr="Number of cycles a write request was stalled. Max over GL2C instances."></metric>
+  <metric name="L2CacheHit" expr=100*sum(GL2C_HIT,16)/(sum(GL2C_HIT,16)+sum(GL2C_MISS,16)) descr="The percentage of fetch, write, atomic, and other instructions that hit the data in L2 cache. Value range: 0% (no hit) to 100% (optimal)."></metric>
+  <metric name="FETCH_SIZE" expr=(GL2C_EA_RDREQ_32B_sum*32+GL2C_EA_RDREQ_64B_sum*64+GL2C_EA_RDREQ_96B_sum*96+GL2C_EA_RDREQ_128B_sum*128)/1024 descr="The total kilobytes fetched from the video memory. This is measured with all extra fetches and any cache or memory effects taken into account."></metric>
+  <metric name="WriteUnitStalled" expr=100*GL2C_WRREQ_STALL_max/GRBM_GUI_ACTIVE descr="The percentage of GPUTime the Write unit is stalled. Value range: 0% to 100% (bad)."></metric>
+  <metric name="LDSBankConflict" expr=100*SQC_LDS_BANK_CONFLICT/SQC_LDS_IDX_ACTIVE descr="The percentage of GPUTime LDS is stalled by bank conflicts. Value range: 0% (optimal) to 100% (bad)."></metric>
+</gfx12_expr>
+
+<gfx1200_expr base="gfx12_expr"></gfx1200_expr>
+<gfx1201_expr base="gfx12_expr"></gfx1201_expr>
+
 <gfx1100_expr base="gfx11_expr"></gfx1100_expr>
 <gfx1101_expr base="gfx11_expr"></gfx1101_expr>
 <gfx1102_expr base="gfx11_expr"></gfx1102_expr>
@@ -459,6 +489,9 @@
 <gfx1101 base="gfx1101_expr"></gfx1101>
 <gfx1102 base="gfx1102_expr"></gfx1102>
 
+<gfx1200 base="gfx1200_expr"></gfx1200>
+<gfx1201 base="gfx1201_expr"></gfx1201>
+
 <global>
   # GPUBusy         The percentage of time GPU was busy.
   <metric
diff --git a/tests-v2/featuretests/profiler/CMakeLists.txt b/tests-v2/featuretests/profiler/CMakeLists.txt
index 0a5eb88339..0cff7269fa 100644
--- a/tests-v2/featuretests/profiler/CMakeLists.txt
+++ b/tests-v2/featuretests/profiler/CMakeLists.txt
@@ -308,7 +308,7 @@ function(generate_hsaco TARGET_ID INPUT_FILE OUTPUT_FILE)
         PARENT_SCOPE)
 endfunction(generate_hsaco)
 
-set(GPU_LIST "gfx900" "gfx906" "gfx908" "gfx90a" "gfx942" "gfx1030" "gfx1031" "gfx1032" "gfx1100" "gfx1101" "gfx1102")
+set(GPU_LIST "gfx900" "gfx906" "gfx908" "gfx90a" "gfx942" "gfx1030" "gfx1031" "gfx1032" "gfx1100" "gfx1101" "gfx1102" "gfx1200" "gfx1201")
 foreach(target_id ${GPU_LIST})
     # generate kernel bitcodes
     generate_hsaco(${target_id} ${CMAKE_CURRENT_SOURCE_DIR}/apps/copy.cl
diff --git a/tests-v2/featuretests/profiler/profiler_gtest.cpp b/tests-v2/featuretests/profiler/profiler_gtest.cpp
index 43a4dcc8f0..4b4dada40c 100644
--- a/tests-v2/featuretests/profiler/profiler_gtest.cpp
+++ b/tests-v2/featuretests/profiler/profiler_gtest.cpp
@@ -45,7 +45,7 @@ std::string test_app_path;
 std::string metrics_path;
 std::string binary_path;
 std::string profiler_api_lib_path = "";
-bool bContainsGFX11Agent = false;
+bool bSkipCounterNoneZeroCheck = false;
 
 static void init_test_path() {
   lib_path = "lib/rocprofiler/librocprofiler_tool.so";
@@ -519,8 +519,9 @@ TEST_F(LoadUnloadTest, WhenLoadingFirstTimeThenToolLoadsUnloadsSuccessfully) {
         std::string agentname;
         agentname.resize(64);
         hsa_agent_get_info(agent, HSA_AGENT_INFO_NAME, agentname.data());
-        if (agentname.find("gfx11") != std::string::npos)
-          bContainsGFX11Agent = true;
+        if ((agentname.find("gfx11") != std::string::npos) ||
+            (agentname.find("gfx12") != std::string::npos))
+          bSkipCounterNoneZeroCheck = true;
 
         hsa_device_type_t type;
         return hsa_agent_get_info(agent, HSA_AGENT_INFO_DEVICE, &type);
@@ -896,7 +897,7 @@ class ProfilerAPITest : public ::testing::Test {
         CheckApi(rocprofiler_query_kernel_info(ROCPROFILER_KERNEL_NAME, profiler_record->kernel_id,
                                                &kernel_name_c));
 
-        if (profiler_record->counters && !bContainsGFX11Agent)
+        if (profiler_record->counters && !bSkipCounterNoneZeroCheck)
           for (uint64_t i = 0; i < profiler_record->counters_count.value; i++)
             if (profiler_record->counters[i].counter_handler.handle > 0)
               EXPECT_NE(profiler_record->counters[i].value.value, 0);