From 20652f775afdcb708264c97c45eed5cd89378772 Mon Sep 17 00:00:00 2001
From: Gopesh Bhardwaj <gopesh.bhardwaj@amd.com>
Date: Sat, 30 Mar 2024 03:14:38 +0530
Subject: [PATCH] SWDEV-452077 Fixing MI300 list counters and metrics issue
 (#701)

[ROCm/rocprofiler-sdk commit: ecc79b1fa34050fecdcc07c374a37e8d42d9bb76]
---
 .../lib/rocprofiler-sdk/counters/xml/derived_counters.xml   | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/xml/derived_counters.xml b/projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/xml/derived_counters.xml
index cbd122ad11..da14d3eab6 100755
--- a/projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/xml/derived_counters.xml
+++ b/projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/xml/derived_counters.xml
@@ -298,13 +298,13 @@
   <metric name="TCP_TCP_TA_DATA_STALL_CYCLES_sum" expr=reduce(TCP_TCP_TA_DATA_STALL_CYCLES,sum) descr="Total number of TCP stalls TA data interface."></metric>
   <metric name="TCP_TCP_TA_DATA_STALL_CYCLES_max" expr=reduce(TCP_TCP_TA_DATA_STALL_CYCLES,max) descr="Maximum number of TCP stalls TA data interface."></metric>
 
-  <metric name="MeanOccupancyPerCU" expr=SQ_LEVEL_WAVES*0+SQ_ACCUM_PREV_HIRES/GRBM_GUI_ACTIVE/CU_NUM descr="Mean occupancy per compute unit."></metric>
+  <metric name="MeanOccupancyPerCU" expr=reduce(SQ_LEVEL_WAVES,sum)*0+reduce(SQ_ACCUM_PREV_HIRES,sum)/reduce(GRBM_GUI_ACTIVE,sum)/CU_NUM descr="Mean occupancy per compute unit."></metric>
   <metric name="MeanOccupancyPerActiveCU" expr=SQ_LEVEL_WAVES*0+SQ_ACCUM_PREV_HIRES*4/SQ_BUSY_CYCLES/CU_NUM descr="Mean occupancy per active compute unit."></metric>
   <metric name="VFetchInsts" expr=(SQ_INSTS_VMEM_RD-TA_FLAT_READ_WAVEFRONTS_sum)/SQ_WAVES descr="The average number of vector fetch instructions from the video memory executed per work-item (affected by flow control). Excludes FLAT instructions that fetch from video memory."></metric>
   <metric name="VWriteInsts" expr=(SQ_INSTS_VMEM_WR-TA_FLAT_WRITE_WAVEFRONTS_sum)/SQ_WAVES descr="The average number of vector write instructions to the video memory executed per work-item (affected by flow control). Excludes FLAT instructions that write to video memory."></metric>
   <metric name="VALUUtilization" expr=100*SQ_THREAD_CYCLES_VALU/(SQ_ACTIVE_INST_VALU*MAX_WAVE_SIZE) descr="The percentage of active vector ALU threads in a wave. A lower number can mean either more thread divergence in a wave or that the work-group size is not a multiple of 64. Value range: 0% (bad), 100% (ideal - no thread divergence)."></metric>
-  <metric name="VALUBusy" expr=100*SQ_ACTIVE_INST_VALU*4/SIMD_NUM/GRBM_GUI_ACTIVE descr="The percentage of GPUTime vector ALU instructions are processed. Value range: 0% (bad) to 100% (optimal)."></metric>
-  <metric name="SALUBusy" expr=100*SQ_INST_CYCLES_SALU*4/SIMD_NUM/GRBM_GUI_ACTIVE descr="The percentage of GPUTime scalar ALU instructions are processed. Value range: 0% (bad) to 100% (optimal)."></metric>
+  <metric name="VALUBusy" expr=100*reduce(SQ_ACTIVE_INST_VALU,sum)*4/SIMD_NUM/reduce(GRBM_GUI_ACTIVE,sum) descr="The percentage of GPUTime vector ALU instructions are processed. Value range: 0% (bad) to 100% (optimal)."></metric>
+  <metric name="SALUBusy" expr=100*reduce(SQ_INST_CYCLES_SALU,sum)*4/SIMD_NUM/reduce(GRBM_GUI_ACTIVE,sum) descr="The percentage of GPUTime scalar ALU instructions are processed. Value range: 0% (bad) to 100% (optimal)."></metric>
   <metric name="FetchSize" expr=FETCH_SIZE 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="WriteSize" expr=WRITE_SIZE descr="The total kilobytes written to the video memory. This is measured with all extra fetches and any cache or memory effects taken into account."></metric>
   <metric name="MemWrites32B" expr=WRITE_REQ_32B descr="The total number of effective 32B write transactions to the memory"></metric>