Skip to contentSkip to conversation

perf vendor events: Update PMC used in PM_RUN_INST_CMPL event for power10 platform

PATCH Applied Kajol Jain 4 participants 4 messages Oct 16, 2023 – Oct 20, 2023 perf events

lore
Share & tools
lore

Kajol Jain

author
#0 lore raw

PATCH 1/1 perf vendor events: Update PMC used in PM_RUN_INST_CMPL event for power10 platform

The CPI_STALL_RATIO metric group can be used to present the high level CPI stall breakdown metrics in powerpc, which will show:

- DISPATCH_STALL_CPI ( Dispatch stall cycles per insn )
- ISSUE_STALL_CPI ( Issue stall cycles per insn )
- EXECUTION_STALL_CPI ( Execution stall cycles per insn )
- COMPLETION_STALL_CPI ( Completion stall cycles per insn )

Commit cf26e043c2a9 ("perf vendor events power10: Add JSON metric events to present CPI stall cycles in powerpc)" which added the CPI_STALL_RATIO metric group, also modified
the PMC value used in PM_RUN_INST_CMPL event from PMC4 to PMC5, to avoid multiplexing of events.
But that got revert in recent changes. Fix this issue by changing back the PMC value used in PM_RUN_INST_CMPL to PMC5.

Result with the fix:

./perf stat --metric-no-group -M CPI_STALL_RATIO <workload>

Performance counter stats for 'workload':

        68,745,426      PM_CMPL_STALL                    #     0.21 COMPLETION_STALL_CPI
         7,692,827      PM_ISSUE_STALL                   #     0.02 ISSUE_STALL_CPI
       322,638,223      PM_RUN_INST_CMPL                 #     0.05 DISPATCH_STALL_CPI
                                                  #     0.48 EXECUTION_STALL_CPI
        16,858,553      PM_DISP_STALL_CYC
       153,880,133      PM_EXEC_STALL

       0.089774592 seconds time elapsed

"--metric-no-group" is used for forcing PM_RUN_INST_CMPL to be scheduled in all group for more accuracy.

Fixes 7d473f475b2a ("perf vendor events: Move JSON/events to appropriate files for power10 platform")
---
 tools/perf/pmu-events/arch/powerpc/power10/pmc.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/pmu-events/arch/powerpc/power10/pmc.json +1 -1
index c606ae03cd27..0e0253d0e757 100644
@@ -195,7 +195,7 @@
     "BriefDescription": "Threshold counter exceeded a value of 128."
   },
   {
-    "EventCode": "0x400FA",
+    "EventCode": "0x500FA",
     "EventName": "PM_RUN_INST_CMPL",
     "BriefDescription": "PowerPC instruction completed while the run latch is set."
   }
Signature · 1 line
2.39.3
Reviewed-by on v1 1/1
#1 lore raw
Quoted 35 lines · Kajol Jain

The CPI_STALL_RATIO metric group can be used to present the high level CPI stall breakdown metrics in powerpc, which will show:

- DISPATCH_STALL_CPI ( Dispatch stall cycles per insn )
- ISSUE_STALL_CPI ( Issue stall cycles per insn )
- EXECUTION_STALL_CPI ( Execution stall cycles per insn )
- COMPLETION_STALL_CPI ( Completion stall cycles per insn )

Commit cf26e043c2a9 ("perf vendor events power10: Add JSON metric events to present CPI stall cycles in powerpc)" which added the CPI_STALL_RATIO metric group, also modified
the PMC value used in PM_RUN_INST_CMPL event from PMC4 to PMC5, to avoid multiplexing of events.
But that got revert in recent changes. Fix this issue by changing back the PMC value used in PM_RUN_INST_CMPL to PMC5.

Result with the fix:

./perf stat --metric-no-group -M CPI_STALL_RATIO <workload>

Performance counter stats for 'workload':

       68,745,426      PM_CMPL_STALL                    #     0.21 COMPLETION_STALL_CPI
        7,692,827      PM_ISSUE_STALL                   #     0.02 ISSUE_STALL_CPI
      322,638,223      PM_RUN_INST_CMPL                 #     0.05 DISPATCH_STALL_CPI
                                                 #     0.48 EXECUTION_STALL_CPI
       16,858,553      PM_DISP_STALL_CYC
      153,880,133      PM_EXEC_STALL

      0.089774592 seconds time elapsed
Quoted 6 lines

"--metric-no-group" is used for forcing PM_RUN_INST_CMPL to be scheduled in all group for more accuracy.

Fixes 7d473f475b2a ("perf vendor events: Move JSON/events to appropriate files for power10 platform")

Thanks
Athira

Quoted 20 lines

---
tools/perf/pmu-events/arch/powerpc/power10/pmc.json | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/pmu-events/arch/powerpc/power10/pmc.json +1 -1
index c606ae03cd27..0e0253d0e757 100644
@@ -195,7 +195,7 @@
    "BriefDescription": "Threshold counter exceeded a value of 128."
  },
  {
-    "EventCode": "0x400FA",
+    "EventCode": "0x500FA",
    "EventName": "PM_RUN_INST_CMPL",
    "BriefDescription": "PowerPC instruction completed while the run latch is set."
  }
Signature · 1 line
2.39.3
Tested-by on v1 1/1
#2 lore raw
In reply to Kajol Jain
Quoted 31 lines · Kajol Jain

The CPI_STALL_RATIO metric group can be used to present the high level CPI stall breakdown metrics in powerpc, which will show:

- DISPATCH_STALL_CPI ( Dispatch stall cycles per insn )
- ISSUE_STALL_CPI ( Issue stall cycles per insn )
- EXECUTION_STALL_CPI ( Execution stall cycles per insn )
- COMPLETION_STALL_CPI ( Completion stall cycles per insn )

Commit cf26e043c2a9 ("perf vendor events power10: Add JSON metric events to present CPI stall cycles in powerpc)" which added the CPI_STALL_RATIO metric group, also modified
the PMC value used in PM_RUN_INST_CMPL event from PMC4 to PMC5, to avoid multiplexing of events.
But that got revert in recent changes. Fix this issue by changing back the PMC value used in PM_RUN_INST_CMPL to PMC5.

Result with the fix:

  ./perf stat --metric-no-group -M CPI_STALL_RATIO <workload>

Performance counter stats for 'workload':

         68,745,426      PM_CMPL_STALL                    #     0.21 COMPLETION_STALL_CPI
          7,692,827      PM_ISSUE_STALL                   #     0.02 ISSUE_STALL_CPI
        322,638,223      PM_RUN_INST_CMPL                 #     0.05 DISPATCH_STALL_CPI
                                                   #     0.48 EXECUTION_STALL_CPI
         16,858,553      PM_DISP_STALL_CYC
        153,880,133      PM_EXEC_STALL

        0.089774592 seconds time elapsed
Quoted 6 lines

"--metric-no-group" is used for forcing PM_RUN_INST_CMPL to be scheduled in all group for more accuracy.

Fixes 7d473f475b2a ("perf vendor events: Move JSON/events to appropriate files for power10 platform")

Thanks for the fix Kajol.
With this patch applied getting correct PMC value for PM_RUN_INST_CMPL event.

# ./perf stat -vv -e PM_RUN_INST_CMPL sleep 1
Using CPUID 00800200
Attempt to add: cpu/PM_RUN_INST_CMPL/
..after resolving event: cpu/event=0x500fa/
PM_RUN_INST_CMPL -> cpu/event=0x500fa/
Control descriptor is not initialized
------------------------------------------------------------
perf_event_attr:
  type                             4 (PERF_TYPE_RAW)
  size                             136
  config                           0x500fa
  sample_type                      IDENTIFIER
  read_format                      TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
  disabled                         1
  inherit                          1
  enable_on_exec                   1
  exclude_guest                    1
------------------------------------------------------------
sys_perf_event_open: pid 750989  cpu -1  group_fd -1  flags 0x8 = 3
PM_RUN_INST_CMPL: -1: 1593381 483118 483118
PM_RUN_INST_CMPL: 1593381 483118 483118

Performance counter stats for 'sleep 1':

         1,593,381      PM_RUN_INST_CMPL

       1.001107900 seconds time elapsed

       0.001087000 seconds user
       0.000000000 seconds sys
Quoted 17 lines
---
  tools/perf/pmu-events/arch/powerpc/power10/pmc.json | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/pmu-events/arch/powerpc/power10/pmc.json +1 -1
index c606ae03cd27..0e0253d0e757 100644
@@ -195,7 +195,7 @@
      "BriefDescription": "Threshold counter exceeded a value of 128."
    },
    {
-    "EventCode": "0x400FA",
+    "EventCode": "0x500FA",
      "EventName": "PM_RUN_INST_CMPL",
      "BriefDescription": "PowerPC instruction completed while the run latch is set."
    }
maintainer Applied
#3 lore raw
In reply to Kajol Jain
Quoted 9 lines · Kajol Jain

The CPI_STALL_RATIO metric group can be used to present the high level CPI stall breakdown metrics in powerpc, which will show:

- DISPATCH_STALL_CPI ( Dispatch stall cycles per insn )
- ISSUE_STALL_CPI ( Issue stall cycles per insn )
- EXECUTION_STALL_CPI ( Execution stall cycles per insn )
- COMPLETION_STALL_CPI ( Completion stall cycles per insn )

[...]

Applied to perf-tools-next, thanks!