MI355X Support - PC Sampling and updating counter_defs.yaml (#206)

* Update mi350/gfx950 counter_defs.yaml (#131)

* Update gfx950 counter_defs.yaml

* Update F8 MFMA for gfx950

* Update counter_defs.yaml

* Update counter_defs.yaml

* add simd_util counter

* add new rdc ops gfx950

* Update counter_defs.yaml

* New mi350 CPC counters

* Update counter_defs.yaml

* New mi350 spi counters

* Update new mi350 sq counter_defs.yaml

* Update TA counter_defs.yaml

* Update TD GFX950counter_defs.yaml

* Update TCP gfx950 counter_defs.yaml

* Update new gfx950 tcc counter_defs.yaml

* Update TCP_PENDING_STALL_CYCLES counter_defs.yaml

* MI355X Host-Trap PC sampling Support (#130)

* Adding gfx12 to CU_NUM

* Add ELFABIVERSION_AMDGPU_HSA_V6

* add gfx950 to TEST_YAML_LOAD metric

* add gfx950 to append counters tests

* Updated CHANGELOG.md

---------

Co-authored-by: Kandula, Venkateshwar reddy <Venkateshwarreddy.Kandula@amd.com>
Co-authored-by: Indic, Vladimir <Vladimir.Indic@amd.com>
Co-authored-by: Bhardwaj, Gopesh <Gopesh.Bhardwaj@amd.com>
Co-authored-by: Madsen, Jonathan <Jonathan.Madsen@amd.com>

[ROCm/rocprofiler-sdk commit: 09c7d44cc4]

projects/rocprofiler-sdk/CHANGELOG.md

@@ -168,6 +168,7 @@ Full documentation for ROCprofiler-SDK is available at [rocm.docs.amd.com/projec
 ### Added
 
 - Added support for rocJPEG API Tracing
+- Added MI350X/MI355X support
 - Added rocprofiler_create_counter to allow for adding custom derived counters at runtime.
 
 ### Changed

projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/tests/core.cpp

@@ -757,7 +757,7 @@ TEST(core, check_load_counter_def_append)
     const std::string test_yaml = R"(
 TEST_YAML_LOAD:
   architectures:
-    gfx942/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx908/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
+    gfx950/gfx942/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx908/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
       expression: reduce(GRBM_GUI_ACTIVE,max)*CU_NUM
   description: 'Unit: cycles'
     )";
@@ -784,13 +784,13 @@ TEST(core, check_load_counter_def)
     const std::string test_yaml = R"(
 GRBM_GUI_ACTIVE:
   architectures:
-    gfx942/gfx941/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx940/gfx908/gfx900/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
+    gfx950/gfx942/gfx941/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx940/gfx908/gfx900/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
       block: GRBM
       event: 2
   description: The GUI is Active
 TEST_YAML_LOAD:
   architectures:
-    gfx942/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx908/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
+    gfx950/gfx942/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx908/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
       expression: reduce(GRBM_GUI_ACTIVE,max)
   description: cycles
     )";

projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/pc_sampling/ioctl/ioctl_adapter.cpp

@@ -238,6 +238,16 @@ is_pc_sampling_supported(const rocprofiler_agent_t* agent)
         else
             return ROCPROFILER_STATUS_ERROR_INCOMPATIBLE_KERNEL;
     }
+    else if(agent_name.find("gfx95") == 0)
+    {
+        // As I am not sure if the PCS IOCTL is going to be bumped for gfx950,
+        // I introduced a separate branch for it.
+        // We expect PC sampling IOCTL to be at least 0.3 for gfx950.
+        if(pcs_ioctl_version.major_version > 0 || pcs_ioctl_version.minor_version >= 3)
+            return ROCPROFILER_STATUS_SUCCESS;
+        else
+            return ROCPROFILER_STATUS_ERROR_INCOMPATIBLE_KERNEL;
+    }
     else
     {
         // The agent does not support PC sampling.

projects/rocprofiler-sdk/tests/rocprofv3/counter-collection/extra_counters/extra_counters.yaml

@@ -1,5 +1,5 @@
 TEST_YAML_LOAD:
   architectures:
-    gfx942/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx908/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
+    gfx950/gfx942/gfx10/gfx1010/gfx1030/gfx1031/gfx11/gfx1032/gfx1102/gfx906/gfx1100/gfx1101/gfx908/gfx90a/gfx9/gfx12/gfx1200/gfx1201:
       expression: reduce(GRBM_GUI_ACTIVE,max)*CU_NUM
   description: 'Unit: cycles'

projects/rocprofiler-sdk/tests/rocprofv3/pc-sampling/host-trap/transpose-multiple-agents/validate.py

@@ -39,7 +39,9 @@ def test_multi_agent_support(
 
     mi2xx_mi3xx_agents_df = input_agent_info_csv[
         input_agent_info_csv["Name"].apply(
-            lambda name: name == "gfx90a" or name.startswith("gfx94")
+            lambda name: name == "gfx90a"
+            or name.startswith("gfx94")
+            or name.startswith("gfx95")
         )
     ]