Re: [PATCH v3 3/7] perf vendor events: Add JSON metrics for CMN

[Jing Zhang]

在 2023/5/31 上午9:18, Ian Rogers 写道:
> On Tue, May 30, 2023 at 2:19 AM Jing Zhang <renyu.zj@xxxxxxxxxxxxxxxxx> wrote:
>>
>> Add JSON metrics for arm CMN. Currently just add part of CMN PMU
>> metrics which are general and compatible for any SoC and CMN-ANY.
>>
>> Signed-off-by: Jing Zhang <renyu.zj@xxxxxxxxxxxxxxxxx>
>> ---
>>  .../pmu-events/arch/arm64/arm/cmn/sys/metrics.json | 74 ++++++++++++++++++++++
>>  tools/perf/pmu-events/jevents.py                   |  1 +
>>  2 files changed, 75 insertions(+)
>>  create mode 100644 tools/perf/pmu-events/arch/arm64/arm/cmn/sys/metrics.json
>>
>> diff --git a/tools/perf/pmu-events/arch/arm64/arm/cmn/sys/metrics.json b/tools/perf/pmu-events/arch/arm64/arm/cmn/sys/metrics.json
>> new file mode 100644
>> index 0000000..e70ac1a
>> --- /dev/null
>> +++ b/tools/perf/pmu-events/arch/arm64/arm/cmn/sys/metrics.json
>> @@ -0,0 +1,74 @@
>> +[
>> +       {
>> +               "MetricName": "slc_miss_rate",
>> +               "BriefDescription": "The system level cache miss rate include.",
> 
> Nit, partial sentence?

Yes my mistake.

> 
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "hnf_cache_miss / hnf_slc_sf_cache_access",
>> +               "ScaleUnit": "100%",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "hnf_message_retry_rate",
>> +               "BriefDescription": "HN-F message retry rate indicates whether a lack of credits is causing the bottlenecks.",
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "hnf_pocq_retry / hnf_pocq_reqs_recvd",
>> +               "ScaleUnit": "100%",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "sf_hit_rate",
>> +               "BriefDescription": "Snoop filter hit rate can be used to measure the Snoop Filter efficiency.",
> 
> Nit, inconsistent capitalization?

Ok, fix it in next version.

> 
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "hnf_sf_hit / hnf_slc_sf_cache_access",
>> +               "ScaleUnit": "100%",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "mc_message_retry_rate",
>> +               "BriefDescription": "The memory controller request retries rate indicates whether the memory controller is the bottleneck.",
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "hnf_mc_retries / hnf_mc_reqs",
>> +               "ScaleUnit": "100%",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "rni_actual_read_bandwidth.all",
>> +               "BriefDescription": "This event measure the actual bandwidth(MB/sec) that RN-I bridge sends to the interconnect.",
> 
> Nit, the MB/sec is in the ScaleUnit so could be dropped here.

ok

> 
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "rnid_rxdat_flits * 32 / 1e6 / duration_time",
>> +               "ScaleUnit": "1MB/s",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "rni_actual_write_bandwidth.all",
>> +               "BriefDescription": "This event measures the actual write bandwidth(MB/sec) at RN-I bridges.",
> 
> Nit, same thing.

ok

> 
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "rnid_txdat_flits * 32 / 1e6 / duration_time",
>> +               "ScaleUnit": "1MB/s",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "rni_retry_rate",
>> +               "BriefDescription": "RN-I bridge retry rate indicates whether the memory controller is the bottleneck.",
>> +               "MetricGroup": "arm_cmn",
>> +               "MetricExpr": "rnid_txreq_flits_retried / rnid_txreq_flits_total",
>> +               "ScaleUnit": "100%",
>> +               "Unit": "arm_cmn",
>> +               "Compat": "arm_cmn600;arm_cmn650;arm_cmn700;arm_ci700"
>> +       },
>> +       {
>> +               "MetricName": "sbsx_actual_write_bandwidth.all",
>> +               "BriefDescription": "sbsx actual write bandwidth(MB/sec).",
> 
> Nit, same thing.
> 

Ok, Thanks Ian.