perf vendor events arm64: Update N2 and V2 metrics and events using Arm telemetry repo
v2 cover perf vendor events arm64: Update N2 and V2 metrics and events using Arm telemetry repo
Changes since v1:
* Split last change into two so it doesn't hit the mailing list size limit
James Clark (5):
perf: cs-etm: Don't duplicate FIELD_GET()
perf jevents: Match on highest version of Arm json file available perf vendor events arm64: Update scale units and descriptions of
common topdown metrics
perf vendor events arm64: Update N2-r0p3 and V2 metrics and events
using Arm telemetry repo
perf vendor events arm64: Update N2-r0p0 metrics and events using Arm
telemetry repo
tools/perf/arch/arm64/util/header.c | 61 +++- .../arch/arm64/arm/neoverse-n2-v2/branch.json | 8 - .../arch/arm64/arm/neoverse-n2-v2/bus.json | 20 -- .../arch/arm64/arm/neoverse-n2-v2/cache.json | 155 -------- .../arm64/arm/neoverse-n2-v2/exception.json | 47 --- .../arm64/arm/neoverse-n2-v2/instruction.json | 143 -------- .../arch/arm64/arm/neoverse-n2-v2/memory.json | 41 --- .../arm64/arm/neoverse-n2-v2/metrics.json | 273 -------------- .../arm64/arm/neoverse-n2-v2/pipeline.json | 23 -- .../arch/arm64/arm/neoverse-n2-v2/spe.json | 14 - .../arch/arm64/arm/neoverse-n2-v2/trace.json | 29 -- .../arch/arm64/arm/neoverse-n2r0p0/bus.json | 18 + .../arm64/arm/neoverse-n2r0p0/exception.json | 62 ++++ .../arm/neoverse-n2r0p0/fp_operation.json | 22 ++ .../arm64/arm/neoverse-n2r0p0/general.json | 10 + .../arm64/arm/neoverse-n2r0p0/l1d_cache.json | 54 +++ .../arm64/arm/neoverse-n2r0p0/l1i_cache.json | 14 + .../arm64/arm/neoverse-n2r0p0/l2_cache.json | 50 +++ .../arm64/arm/neoverse-n2r0p0/l3_cache.json | 22 ++ .../arm64/arm/neoverse-n2r0p0/ll_cache.json | 10 + .../arm64/arm/neoverse-n2r0p0/memory.json | 46 +++ .../arm64/arm/neoverse-n2r0p0/metrics.json | 332 ++++++++++++++++++ .../arm64/arm/neoverse-n2r0p0/retired.json | 30 ++ .../arch/arm64/arm/neoverse-n2r0p0/spe.json | 18 + .../arm/neoverse-n2r0p0/spec_operation.json | 110 ++++++ .../arch/arm64/arm/neoverse-n2r0p0/stall.json | 30 ++ .../arch/arm64/arm/neoverse-n2r0p0/sve.json | 50 +++ .../arch/arm64/arm/neoverse-n2r0p0/tlb.json | 66 ++++ .../arch/arm64/arm/neoverse-n2r0p0/trace.json | 38 ++ .../arm64/arm/neoverse-n2r0p3-v2/bus.json | 18 + .../arm/neoverse-n2r0p3-v2/exception.json | 62 ++++ .../arm/neoverse-n2r0p3-v2/fp_operation.json | 22 ++ .../arm64/arm/neoverse-n2r0p3-v2/general.json | 10 + .../arm/neoverse-n2r0p3-v2/l1d_cache.json | 54 +++ .../arm/neoverse-n2r0p3-v2/l1i_cache.json | 14 + .../arm/neoverse-n2r0p3-v2/l2_cache.json | 50 +++ .../arm/neoverse-n2r0p3-v2/l3_cache.json | 22 ++ .../arm/neoverse-n2r0p3-v2/ll_cache.json | 10 + .../arm64/arm/neoverse-n2r0p3-v2/memory.json | 46 +++ .../arm64/arm/neoverse-n2r0p3-v2/metrics.json | 331 +++++++++++++++++ .../arm64/arm/neoverse-n2r0p3-v2/retired.json | 30 ++ .../arm64/arm/neoverse-n2r0p3-v2/spe.json | 18 + .../neoverse-n2r0p3-v2/spec_operation.json | 110 ++++++ .../arm64/arm/neoverse-n2r0p3-v2/stall.json | 30 ++ .../arm64/arm/neoverse-n2r0p3-v2/sve.json | 50 +++ .../arm64/arm/neoverse-n2r0p3-v2/tlb.json | 66 ++++ .../arm64/arm/neoverse-n2r0p3-v2/trace.json | 38 ++ tools/perf/pmu-events/arch/arm64/mapfile.csv | 5 +- tools/perf/pmu-events/arch/arm64/sbsa.json | 24 +- tools/perf/pmu-events/jevents.py | 49 +-- tools/perf/tests/pmu-events.c | 34 ++ tools/perf/util/cs-etm.c | 14 +- 52 files changed, 2088 insertions(+), 815 deletions(-) delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/branch.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/bus.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/cache.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/exception.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/instruction.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/memory.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/metrics.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/pipeline.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/spe.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/trace.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/bus.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/exception.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/fp_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/general.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l1d_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l1i_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l2_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l3_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/ll_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/memory.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/metrics.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/retired.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/spe.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/spec_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/stall.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/sve.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/tlb.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/trace.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/bus.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/exception.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/fp_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/general.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l1d_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l1i_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l2_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l3_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/ll_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/memory.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/retired.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/spe.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/spec_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/stall.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/sve.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/tlb.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/trace.json
Signature · 1 line
2.34.1
PATCH v2 1/5 perf: cs-etm: Don't duplicate FIELD_GET()
linux/bitfield.h can be included as long as linux/kernel.h is included first, so change the order of the includes and drop the duplicate macro.
--- tools/perf/util/cs-etm.c | 14 ++------------ 1 file changed, 2 insertions(+), 12 deletions(-)
tools/perf/util/cs-etm.c +2 -12
@@ -6,10 +6,11 @@* Author: Mathieu Poirier <[email protected]> */ +#include <linux/kernel.h> +#include <linux/bitfield.h> #include <linux/bitops.h> #include <linux/coresight-pmu.h> #include <linux/err.h> -#include <linux/kernel.h> #include <linux/log2.h> #include <linux/types.h> #include <linux/zalloc.h>
@@ -281,17 +282,6 @@ static int cs_etm__metadata_set_trace_id(u8 trace_chan_id, u64 *cpu_metadata)return 0 ; } -/* -* FIELD_GET (linux/bitfield.h) not available outside kernel code, -* and the header contains too many dependencies to just copy over, -* so roll our own based on the original -*/ -#define __bf_shf (x) (__builtin_ffsll(x) - 1) -#define FIELD_GET (_mask, _reg) \ - ({ \ - (typeof(_mask))(((_reg) & (_mask)) >> __bf_shf(_mask)); \ - }) -/* * Get a metadata for a specific cpu from an array. *
Signature · 1 line
2.34.1
PATCH v2 2/5 perf jevents: Match on highest version of Arm json file available
In reply to James ClarkCurrently version and revision fields are masked out of the MIDR so there can only be one set of jsons per CPU. In a later commit multiple revisions of Neoverse N2 json files will be provided.
The highest valid version of json files should be used, but to make this work the mapfile has to be reverse sorted on the CPUID field so that the highest is found first. It's possible, but error prone, to do this manually so instead add an explicit sort into jevents.py. If the CPUID is a string then the rows are string sorted rather than numerically.
--- tools/perf/arch/arm64/util/header.c | 61 ++++++++++++++++++++++------- tools/perf/pmu-events/jevents.py | 49 ++++++++++++----------- tools/perf/tests/pmu-events.c | 34 ++++++++++++++++ 3 files changed, 108 insertions(+), 36 deletions(-)
tools/perf/arch/arm64/util/header.c +47 -14
@@ -1,3 +1,6 @@ +#include <linux/kernel.h> +#include <linux/bits.h> +#include <linux/bitfield.h> #include <stdio.h> #include <stdlib.h> #include <perf/cpumap.h>
@@ -10,14 +13,12 @@#define MIDR "/regs/identification/midr_el1"#define MIDR_SIZE 19 -#define MIDR_REVISION_MASK 0xf -#define MIDR_VARIANT_SHIFT 20 -#define MIDR_VARIANT_MASK (0xf << MIDR_VARIANT_SHIFT) +#define MIDR_REVISION_MASK GENMASK(3, 0) +#define MIDR_VARIANT_MASK GENMASK(23, 20)static int _get_cpuid (char *buf,size_t sz,struct perf_cpu_map *cpus) {const char *sysfs =sysfs__mountpoint (); -u64 midr =0 ;int cpu;if (!sysfs || sz <MIDR_SIZE )
@@ -44,21 +45,11 @@ static int _get_cpuid(char *buf, size_t sz, struct perf_cpu_map *cpus) }fclose (file); -/* Ignore/clear Variant[23:20] and -* Revision[3:0] of MIDR -*/ - midr =strtoul (buf,NULL ,16 ); - midr &= (~(MIDR_VARIANT_MASK |MIDR_REVISION_MASK )); -scnprintf (buf,MIDR_SIZE ,"0x%016lx" , midr);/* got midr break loop */ break ; }perf_cpu_map__put (cpus); - -if (!midr) -return EINVAL ; -return 0 ; }
@@ -99,3 +90,45 @@ char *get_cpuid_str(struct perf_pmu *pmu)return buf; } + + +int strcmp_cpuid_str (const char *mapcpuid,const char *idstr) +{ +u64 map_id =strtoull (mapcpuid,NULL ,16 ); +char map_id_variant =FIELD_GET (MIDR_VARIANT_MASK , map_id); +char map_id_revision =FIELD_GET (MIDR_REVISION_MASK , map_id); +u64 id =strtoull (idstr,NULL ,16 ); +char id_variant =FIELD_GET (MIDR_VARIANT_MASK , id); +char id_revision =FIELD_GET (MIDR_REVISION_MASK , id); +u64 id_fields = ~(MIDR_VARIANT_MASK |MIDR_REVISION_MASK ); + +/* Compare without version first */ +if ((map_id & id_fields) != (id & id_fields)) +return 1 ; + +/* +* ID matches, now compare version. +* +* Arm revisions (like r0p0) are compared here like two digit semver +* values eg. 1.3 < 2.0 < 2.1 < 2.2. The events json file with the +* highest matching version is used. +* +* r = high value = 'Variant' field in MIDR +* p = low value = 'Revision' field in MIDR +* +* Because the Variant field is further to the left, iterating through a +* reverse sorted mapfile.csv gives the correct comparison behavior. +* This relies on jevents.py sorting the list in print_mapping_table(). +*/ +if (id_variant > map_id_variant) +return 0 ; + +if (id_variant == map_id_variant && id_revision >= map_id_revision) +return 0 ; + +/* +* variant is less than mapfile variant or variants are the same but +* the revision doesn't match. Return no match. +*/ +return 1 ; +}
tools/perf/pmu-events/jevents.py +27 -22
62 lines not on this page · raw message
tools/perf/tests/pmu-events.c +34 -0
@@ -14,6 +14,7 @@#include "util/evlist.h" #include "util/expr.h" #include "util/hashmap.h" +#include "util/header.h" #include "util/parse-events.h" #include "metricgroup.h" #include "stat.h"
@@ -1027,6 +1028,38 @@ static int test__parsing_threshold(struct test_suite *test __maybe_unused,return pmu_for_each_sys_metric (test__parsing_threshold_callback,NULL ); } +static int test__cpuid_match (struct test_suite *test__maybe_unused , +int subtest__maybe_unused ) +{ +#ifdef __aarch64__ +/* midr with no leading zeros matches */ +if (strcmp_cpuid_str ("0x410fd0c0" ,"0x00000000410fd0c0" )) +return -1 ; +/* Upper case matches */ +if (strcmp_cpuid_str ("0x410fd0c0" ,"0x00000000410FD0C0" )) +return -1 ; +/* r0p0 = r0p0 matches */ +if (strcmp_cpuid_str ("0x00000000410fd480" ,"0x00000000410fd480" )) +return -1 ; +/* r0p1 > r0p0 matches */ +if (strcmp_cpuid_str ("0x00000000410fd480" ,"0x00000000410fd481" )) +return -1 ; +/* r1p0 > r0p0 matches*/ +if (strcmp_cpuid_str ("0x00000000410fd480" ,"0x00000000411fd480" )) +return -1 ; +/* r0p0 < r0p1 doesn't match */ +if (!strcmp_cpuid_str ("0x00000000410fd481" ,"0x00000000410fd480" )) +return -1 ; +/* r0p0 < r1p0 doesn't match */ +if (!strcmp_cpuid_str ("0x00000000411fd480" ,"0x00000000410fd480" )) +return -1 ; +/* Different CPU doesn't match */ +if (!strcmp_cpuid_str ("0x00000000410fd4c0" ,"0x00000000430f0af0" )) +return -1 ; +#endif +return 0 ; +} +static struct test_case pmu_events_tests[] = {TEST_CASE ("PMU event table sanity" , pmu_event_table),TEST_CASE ("PMU event map aliases" , aliases),
@@ -1034,6 +1067,7 @@ static struct test_case pmu_events_tests[] = { "some metrics failed"),TEST_CASE ("Parsing of PMU event table metrics with fake PMUs" , parsing_fake),TEST_CASE ("Parsing of metric thresholds with fake PMUs" , parsing_threshold), +TEST_CASE ("CPUID matching" , cpuid_match), { .name = NULL, } };
Signature · 1 line
2.34.1
John Garry
reviewerRe: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
On 10/07/2023 15:19, James Clark wrote:
+
Hi James,
It would be good to cc some additional people and lists.
Quoted 1 line
Currently version and revision fields are masked out of the MIDR so
Do you mean variant and revision?
Quoted 6 lines
there can only be one set of jsons per CPU. In a later commit multiple revisions of Neoverse N2 json files will be provided.
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Then to solve that are you saying that the highest version of the JSONs should used, upto and including the same CPU version (???) - correct?
The cover letter doesn't mention anything relevant :(
Quoted 7 lines
but to make this
work the mapfile has to be reverse sorted on the CPUID field so that the highest is found first.
It's possible, but error prone, to do this
manually so instead add an explicit sort intojevents.py. If the CPUID is a string then the rows are string sorted rather than numerically.
Thanks,
John
_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Re: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 8 lines · John Garry
On 10/07/2023 15:19, James Clark wrote:
+
Hi James,
It would be good to cc some additional people and lists.
Hi John,
Resent a v3 with the normal get_maintainer list. I did this with the first one but it bounced and for some reason with the v2 I reduced the CC list. Not sure why really...
Quoted 5 lines
Quoted 2 lines
Currently version and revision fields are masked out of the MIDR so
Do you mean variant and revision?
Yes that should have been variant. Fixed in v3, thanks.
Quoted 11 lines
Quoted 6 lines
there can only be one set of jsons per CPU. In a later commit multiple revisions of Neoverse N2 json files will be provided.
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Yes. In this case we changed how a metric is calculated in N2-r0p3 because the workaround that was needed for r0p0 is no longer needed (CPU_CYCLES should not subtracted from stalls for topdown metrics anymore). But it would also support the case where the event list was slightly different between versions.
Quoted 3 lines
Then to solve that are you saying that the highest version of the JSONs should used, upto and including the same CPU version (???) - correct?
Yeah pretty much. It's highest version of JSONs starting from (including) the same CPU version, up to and not including the next version of JSON files provided. If only r0p0 is provided then all CPU versions match.
So if these JSONs were provided:
r0p0 - r0p4 - r1p1
Would match these CPUs:
CPU | JSON matched ------|--------------- r0p0 | r0p0 r0p3 | r0p0 r0p4 | r0p4 r1p0 | r0p4 r1p1+ | r1p1
Quoted 3 lines
The cover letter doesn't mention anything relevant :(
Hopefully I expanded more in the v3 cover letter.
Thanks
James
Quoted 11 lines
Quoted 7 lines
but to make this
work the mapfile has to be reverse sorted on the CPUID field so that the highest is found first.
It's possible, but error prone, to do this
manually so instead add an explicit sort intojevents.py. If the CPUID is a string then the rows are string sorted rather than numerically.Thanks,
John
John Garry
reviewerRe: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 9 lines · James Clark
Quoted 5 lines
Quoted 1 line
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Yes. In this case we changed how a metric is calculated in N2-r0p3 because the workaround that was needed for r0p0 is no longer needed (
CPU_CYCLESshould not subtracted from stalls for topdown metrics anymore).
If there are only very subtle differences, then could we solve with metric expression function? We already do something else like this for literal #slots for arm64
Quoted 2 lines
But it would also support the case where the event list was slightly different between versions.
I am diff'ing folders ../arm64/arm/neoverse-n2r0p3-v2 and ../arm64/arm/neoverse-n2r0p0, and all the events are the same - am I correct? It only seems like metric json is different.
Quoted 22 lines
Quoted 3 lines
Then to solve that are you saying that the highest version of the JSONs should used, upto and including the same CPU version (???) - correct?
Yeah pretty much. It's highest version of JSONs starting from (including) the same CPU version, up to and not including the next version of JSON files provided. If only r0p0 is provided then all CPU versions match.
So if these JSONs were provided:
r0p0 - r0p4 - r1p1
Would match these CPUs:
CPU | JSON matched ------|--------------- r0p0 | r0p0 r0p3 | r0p0 r0p4 | r0p4 r1p0 | r0p4 r1p1+ | r1p1
Quoted 3 lines
Quoted 2 lines
The cover letter doesn't mention anything relevant 🙁
Hopefully I expanded more in the v3 cover letter.
Thanks,
John
_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Re: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 6 lines · John Garry
Quoted 5 lines · James Clark
Quoted 5 lines
Quoted 1 line
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Quoted 8 lines
Quoted 4 lines
Yes. In this case we changed how a metric is calculated in N2-r0p3 because the workaround that was needed for r0p0 is no longer needed (
CPU_CYCLESshould not subtracted from stalls for topdown metrics anymore).If there are only very subtle differences, then could we solve with metric expression function? We already do something else like this for literal #slots for arm64
Possibly, but I'd have to add a new mechanism to expose the version numbers to the metrics so it can be used in an expression. And I think it could get very messy quite quickly and make the expression hard to read. In fact in this specific case I'm not even sure how it would look. Something like this (where #p is the p in r0p3)?
{
"ArchStdEvent": "bad_speculation",
"MetricExpr": "(100 * (((1 - (OP_RETIRED / OP_SPEC)) * (1 -
((STALL_SLOT - (CPU_CYCLES if (#p >= 3) else 0) / (CPU_CYCLES * #slots)))) + ((BR_MIS_PRED * 4) / CPU_CYCLES)))"
},
In this case it's not so bad because we don't need to compare r as well, but I think it could get worse if r was non 0 or there were more significant differences to the metric.
It also doesn't scale in the same way when we're using auto generated metrics with the data that Arm is publishing. We're going to treat CPU versions as a completely new set of JSONs if there are any differences/fixes in the PMU stuff and just publish a whole new set. That way any tool that's consuming these can just check the MIDR and use whichever JSONS are most appropriate. In theory this shouldn't require any manual intervention or re-writing of expressions. And it saves every tool having to solve the same problem.
It's possible in the future that there are also small changes to the description text or availability of the events between versions. It hasn't happened yet, but it gives us the flexibility to do that too.
Quoted 8 lines
Quoted 3 lines
But it would also support the case where the event list was slightly different between versions.
I am diff'ing folders ../arm64/arm/neoverse-n2r0p3-v2 and ../arm64/arm/neoverse-n2r0p0, and all the events are the same - am I correct? It only seems like metric json is different.
Yes that's correct, just the topdown metrics where CPU_CYCLES was subtracted as a workaround have been removed. N2 r0p3 no longer needs that workaround. But everything else is the same.
I also used it as an opportunity to remove #slots from N2-r0p0. That set of metrics will only be used in one place where slots is always 5. It's better to hard code it so that it also has a chance to work on kernels that don't have the change to expose slots yet.
Quoted 29 lines
Quoted 20 lines
Quoted 3 lines
Then to solve that are you saying that the highest version of the JSONs should used, upto and including the same CPU version (???) - correct?
Yeah pretty much. It's highest version of JSONs starting from (including) the same CPU version, up to and not including the next version of JSON files provided. If only r0p0 is provided then all CPU versions match.
So if these JSONs were provided:
r0p0 - r0p4 - r1p1
Would match these CPUs:
CPU | JSON matched
------|---------------r0p0 | r0p0 r0p3 | r0p0 r0p4 | r0p4 r1p0 | r0p4Quoted 5 lines
r1p1+ | r1p1
Quoted 2 lines
The cover letter doesn't mention anything relevant 🙁
Hopefully I expanded more in the v3 cover letter.
Thanks,
John
John Garry
reviewerRe: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 24 lines · James Clark
Quoted 6 lines · John Garry
Quoted 5 lines · James Clark
Quoted 5 lines
Quoted 1 line
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Quoted 8 lines
Quoted 4 lines
Yes. In this case we changed how a metric is calculated in N2-r0p3 because the workaround that was needed for r0p0 is no longer needed (
CPU_CYCLESshould not subtracted from stalls for topdown metrics anymore).If there are only very subtle differences, then could we solve with metric expression function? We already do something else like this for literal #slots for arm64
Possibly, but I'd have to add a new mechanism to expose the version numbers to the metrics so it can be used in an expression. And I think it could get very messy quite quickly and make the expression hard to read. In fact in this specific case I'm not even sure how it would look. Something like this (where #p is the p in r0p3)?
Quoted 5 lines
{
"ArchStdEvent": "bad_speculation",
"MetricExpr": "(100 * (((1 - (OP_RETIRED/OP_SPEC)) * (1 -
((STALL_SLOT- (CPU_CYCLESif (#p >= 3) else 0) / (CPU_CYCLES* #slots)))) + ((BR_MIS_PRED* 4) /CPU_CYCLES)))"
Could this formula for "CPU_CYCLES if (#p >= 3) else 0" be generalised with a literal (like what we do for #slots)?
Quoted 6 lines
},In this case it's not so bad because we don't need to compare r as well, but I think it could get worse if r was non 0 or there were more significant differences to the metric.
Quoted 6 lines
It also doesn't scale in the same way when we're using auto generated metrics with the data that Arm is publishing. We're going to treat CPU versions as a completely new set of JSONs if there are any differences/fixes in the PMU stuff and just publish a whole new set. That way any tool that's consuming these can just check the MIDR and use whichever JSONS are most appropriate.
So far we have got away with ignoring revision and variant - why change this now just for these arm parts is a good question..
Quoted 3 lines
In theory this shouldn't require
any manual intervention or re-writing of expressions. And it saves every tool having to solve the same problem.
Sure, but we have tried to reduce duplication as much as possible, and example would be ArchStdEvents
Quoted 4 lines
It's possible in the future that there are also small changes to the description text or availability of the events between versions. It hasn't happened yet, but it gives us the flexibility to do that too.
Here the (handful of) metrics are subtly different between these rev/variant, but all the events are the same. It's hard to justify duplicating almost everything just for that.
Quoted 1 line
I am not sure on how to handle this, since we might be bombarded with support for more CPUs with the same MIDR issue and having quirks for all becomes unworkable.
Can you just give the idea above on literals/functions in the metricexpr a go, to see how it looks? Maybe Ian has a good idea on similar solutions.
Thanks,
John
Re: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 25 lines · John Garry
Quoted 24 lines · James Clark
Quoted 6 lines · John Garry
Quoted 5 lines · James Clark
Quoted 5 lines
Quoted 1 line
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Quoted 8 lines
Quoted 4 lines
Yes. In this case we changed how a metric is calculated in N2-r0p3 because the workaround that was needed for r0p0 is no longer needed (
CPU_CYCLESshould not subtracted from stalls for topdown metrics anymore).If there are only very subtle differences, then could we solve with metric expression function? We already do something else like this for literal #slots for arm64
Possibly, but I'd have to add a new mechanism to expose the version numbers to the metrics so it can be used in an expression. And I think it could get very messy quite quickly and make the expression hard to read. In fact in this specific case I'm not even sure how it would look. Something like this (where #p is the p in r0p3)?
Quoted 9 lines
Quoted 5 lines
{
"ArchStdEvent": "bad_speculation",
"MetricExpr": "(100 * (((1 - (OP_RETIRED/OP_SPEC)) * (1 - ((STALL_SLOT- (CPU_CYCLESif (#p >= 3) else 0) / (CPU_CYCLES* #slots)))) + ((BR_MIS_PRED* 4) /CPU_CYCLES)))"Could this formula for "
CPU_CYCLESif (#p >= 3) else 0" be generalised with a literal (like what we do for #slots)?
Yes that could work, then we wouldn't need the logic inside the formula.
Quoted 16 lines
Quoted 6 lines
},
In this case it's not so bad because we don't need to compare r as well, but I think it could get worse if r was non 0 or there were more significant differences to the metric.
Quoted 6 lines
It also doesn't scale in the same way when we're using auto generated metrics with the data that Arm is publishing. We're going to treat CPU versions as a completely new set of JSONs if there are any differences/fixes in the PMU stuff and just publish a whole new set. That way any tool that's consuming these can just check the MIDR and use whichever JSONS are most appropriate.
So far we have got away with ignoring revision and variant - why change this now just for these arm parts is a good question..
Quoted 7 lines
Quoted 3 lines
In theory this shouldn't require
any manual intervention or re-writing of expressions. And it saves every tool having to solve the same problem.Sure, but we have tried to reduce duplication as much as possible, and example would be ArchStdEvents
What's the end goal of reducing duplication? Is it to reduce the binary size or to reduce human input?
For the binary size if most of these strings are the same then they are de-duplicated and the impact on the binary size is very small.
If it's to save on human input, then not doing it this way has the potential to be more work in the long run. At the moment we can just run a script and generate the Arm JSONs using what Arm is publishing. We might understand all these subtle de-duplication efforts, but anyone coming in the future is going to have a hard time understanding. Especially if they just want to re-run the script to update some text, then there is a big risk of making a mistake when re-applying the manual changes.
I'm not completely opposed to the idea of re-working this expression this time so that it works in both places, but I think we should be sure that we do it for the right reasons.
As an example for comparison, I tried a build with a single set of events and both sets (from this change) and the difference in size is 1760 bytes. Is it really worth starting to edit formulas just to save that much space?
Quoted 9 lines
Quoted 4 lines
It's possible in the future that there are also small changes to the description text or availability of the events between versions. It hasn't happened yet, but it gives us the flexibility to do that too.
Here the (handful of) metrics are subtly different between these rev/variant, but all the events are the same. It's hard to justify duplicating almost everything just for that.
If it saves time and reduces the chance of making mistakes then personally I don't see an issue with the duplication.
Quoted 6 lines
Quoted 1 line
I am not sure on how to handle this, since we might be bombarded with support for more CPUs with the same MIDR issue and having quirks for all becomes unworkable.
I really don't see this being used very often. And it will always be used much less often than the task of adding entirely new CPUs, so it won't cause the number of JSONs to grow faster than that already existing work. To be honest there is a non-zero chance that it never gets used again.
Although even if we never use it again we still might want to re-generate the JSONS because of a typo fix or something so that task is just a script run rather than script run, and then re-applying the expression fixes.
Quoted 5 lines
Can you just give the idea above on literals/functions in the metricexpr a go, to see how it looks? Maybe Ian has a good idea on similar solutions.
Thanks,
John
Yes I can give it a go. Also interested to see what Ian thinks as well.
Thanks
James
Re: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 79 lines · James Clark
Quoted 25 lines · John Garry
Quoted 24 lines · James Clark
Quoted 6 lines · John Garry
Quoted 5 lines · James Clark
Quoted 5 lines
Quoted 1 line
The highest valid version of json files should be used,
What exactly does that mean?
So it seems that you have CPUs with matching MIDR except variant and revision, but have different events, right?
Quoted 8 lines
Quoted 4 lines
Yes. In this case we changed how a metric is calculated in N2-r0p3 because the workaround that was needed for r0p0 is no longer needed (
CPU_CYCLESshould not subtracted from stalls for topdown metrics anymore).If there are only very subtle differences, then could we solve with metric expression function? We already do something else like this for literal #slots for arm64
Possibly, but I'd have to add a new mechanism to expose the version numbers to the metrics so it can be used in an expression. And I think it could get very messy quite quickly and make the expression hard to read. In fact in this specific case I'm not even sure how it would look. Something like this (where #p is the p in r0p3)?
Quoted 9 lines
Quoted 5 lines
{
"ArchStdEvent": "bad_speculation",
"MetricExpr": "(100 * (((1 - (OP_RETIRED/OP_SPEC)) * (1 - ((STALL_SLOT- (CPU_CYCLESif (#p >= 3) else 0) / (CPU_CYCLES* #slots)))) + ((BR_MIS_PRED* 4) /CPU_CYCLES)))"Could this formula for "
CPU_CYCLESif (#p >= 3) else 0" be generalised with a literal (like what we do for #slots)?Yes that could work, then we wouldn't need the logic inside the formula.
Quoted 16 lines
Quoted 6 lines
},
In this case it's not so bad because we don't need to compare r as well, but I think it could get worse if r was non 0 or there were more significant differences to the metric.
Quoted 6 lines
It also doesn't scale in the same way when we're using auto generated metrics with the data that Arm is publishing. We're going to treat CPU versions as a completely new set of JSONs if there are any differences/fixes in the PMU stuff and just publish a whole new set. That way any tool that's consuming these can just check the MIDR and use whichever JSONS are most appropriate.
So far we have got away with ignoring revision and variant - why change this now just for these arm parts is a good question..
Quoted 7 lines
Quoted 3 lines
In theory this shouldn't require
any manual intervention or re-writing of expressions. And it saves every tool having to solve the same problem.Sure, but we have tried to reduce duplication as much as possible, and example would be ArchStdEvents
What's the end goal of reducing duplication? Is it to reduce the binary size or to reduce human input?
For the binary size if most of these strings are the same then they are de-duplicated and the impact on the binary size is very small.
If it's to save on human input, then not doing it this way has the potential to be more work in the long run. At the moment we can just run a script and generate the Arm JSONs using what Arm is publishing. We might understand all these subtle de-duplication efforts, but anyone coming in the future is going to have a hard time understanding. Especially if they just want to re-run the script to update some text, then there is a big risk of making a mistake when re-applying the manual changes.
Quoted 9 lines
I'm not completely opposed to the idea of re-working this expression this time so that it works in both places, but I think we should be sure that we do it for the right reasons.
As an example for comparison, I tried a build with a single set of events and both sets (from this change) and the difference in size is 1760 bytes. Is it really worth starting to edit formulas just to save that much space?
I re-ran the test with a clean build and the difference is actually 4888 bytes. Still quite small but not as suspiciously small as 1760.
Quoted 41 lines
Quoted 9 lines
Quoted 4 lines
It's possible in the future that there are also small changes to the description text or availability of the events between versions. It hasn't happened yet, but it gives us the flexibility to do that too.
Here the (handful of) metrics are subtly different between these rev/variant, but all the events are the same. It's hard to justify duplicating almost everything just for that.
If it saves time and reduces the chance of making mistakes then personally I don't see an issue with the duplication.
Quoted 6 lines
Quoted 1 line
I am not sure on how to handle this, since we might be bombarded with support for more CPUs with the same MIDR issue and having quirks for all becomes unworkable.
I really don't see this being used very often. And it will always be used much less often than the task of adding entirely new CPUs, so it won't cause the number of JSONs to grow faster than that already existing work. To be honest there is a non-zero chance that it never gets used again.
Although even if we never use it again we still might want to re-generate the JSONS because of a typo fix or something so that task is just a script run rather than script run, and then re-applying the expression fixes.
Quoted 5 lines
Can you just give the idea above on literals/functions in the metricexpr a go, to see how it looks? Maybe Ian has a good idea on similar solutions.
Thanks,
JohnYes I can give it a go. Also interested to see what Ian thinks as well.
Thanks
James
John Garry
reviewerRe: [PATCH v2 2/5] perf jevents: Match on highest version of Arm json file available
Quoted 3 lines
Quoted 3 lines
Quoted 1 line
What's the end goal of reducing duplication? Is it to reduce the binary size or to reduce human input?
Not to reduce binary size. Well that was never the original intent.
Originally we were seeing many transcribed JSONs, with descriptions manually copied from TRMs/ARM ARM or some random implementator speadsheets. There was also lots of inconsistencies of event descriptions and classifications between implementators and implemenations when there are no differences in practice – it is not helpful in giving a consistent user experience on arm.
The motivation to reduce duplication was for the usual reasons, like:
- minimize mistakes
- ease of maintenance
- easier to update
- etc
There was no central repository of ARM events for all implementators, like intel had in (I think) 01org. So it was good to try to provide a method in perf tool to harmonize event support there.
Obviously when implementators have their own repo of events in proprietary formats - like arm does - some conversion needs to be done (to perf tool format). And using methods in perf tool there to reduce duplication becomes questionable - reducing duplication would be the job there of the original repo author.
So we could say that arm can have as much duplication as they want in their JSONs as we know that they are managing in their own repo, but having different rules for different implementators is going to be difficult to manage. But allowing duplication in arm JSONs does leave door open for inconsistent event descriptions between implementators.
Quoted 13 lines
Quoted 13 lines
For the binary size if most of these strings are the same then they are de-duplicated and the impact on the binary size is very small.
If it's to save on human input, then not doing it this way has the potential to be more work in the long run. At the moment we can just run a script and generate the Arm JSONs using what Arm is publishing. We might understand all these subtle de-duplication efforts, but anyone coming in the future is going to have a hard time understanding. Especially if they just want to re-run the script to update some text, then there is a big risk of making a mistake when re-applying the manual changes.
Quoted 3 lines
Quoted 3 lines
I'm not completely opposed to the idea of re-working this expression this time so that it works in both places, but I think we should be sure that we do it for the right reasons.
For sure. For this issue, it's ironically a bit annoying to have such a small set of differences - otherwise we could just let it in as is. But have a small set of subtle differences in a handful or metrics makes me think that we can solve this with literal / formula.
Again, if this turns out to be a continuous issue popping up, then it may become an unmanageable rule.
Quoted 8 lines
Quoted 6 lines
As an example for comparison, I tried a build with a single set of events and both sets (from this change) and the difference in size is 1760 bytes. Is it really worth starting to edit formulas just to save that much space?
I re-ran the test with a clean build and the difference is actually 4888 bytes. Still quite small but not as suspiciously small as 1760.
But it's still not huge in terms of size.
FWIW, we could prob reduce size more by doing the archstdevent fixup during runtime and not in generating pmu-events.c, which would be saving more space.
Quoted 26 lines
Quoted 26 lines
Quoted 8 lines
Quoted 4 lines
It's possible in the future that there are also small changes to the description text or availability of the events between versions. It hasn't happened yet, but it gives us the flexibility to do that too.
Here the (handful of) metrics are subtly different between these rev/variant, but all the events are the same. It's hard to justify duplicating almost everything just for that.
If it saves time and reduces the chance of making mistakes then personally I don't see an issue with the duplication.
Quoted 4 lines
I am not sure on how to handle this, since we might be bombarded with support for more CPUs with the same MIDR issue and having quirks for all becomes unworkable.
I really don't see this being used very often. And it will always be used much less often than the task of adding entirely new CPUs, so it won't cause the number of JSONs to grow faster than that already existing work. To be honest there is a non-zero chance that it never gets used again.
Although even if we never use it again we still might want to re-generate the JSONS because of a typo fix or something so that task is just a script run rather than script run, and then re-applying the expression fixes.
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Can you just give the idea above on literals/functions in the metricexpr a go, to see how it looks? Maybe Ian has a good idea on similar solutions.
Thanks,
JohnYes I can give it a go. Also interested to see what Ian thinks as well.
Cheers for that.
John
PATCH v2 3/5 perf vendor events arm64: Update scale units and descriptions of common topdown metrics
In reply to James ClarkMetrics will be published here [1] going forwards, but they have slightly different scale units. To allow autogenerated metrics to be added more easily, update the scale units to match.
The more detailed descriptions have also been taken and added to the common file.
[1]: https://gitlab.arm.com/telemetry-solution/telemetry-solution/-/tree/main/data/pmu/cpu/
--- tools/perf/pmu-events/arch/arm64/sbsa.json | 24 +++++++++++----------- 1 file changed, 12 insertions(+), 12 deletions(-)
tools/perf/pmu-events/arch/arm64/sbsa.json +12 -12
@@ -1,34 +1,34 @@ [ { - "MetricExpr": "stall_slot_frontend / (#slots * cpu_cycles)", - "BriefDescription": "Frontend bound L1 topdown metric", + "MetricExpr": "100 * (stall_slot_frontend / (#slots * cpu_cycles))", + "BriefDescription": "This metric is the percentage of total slots that were stalled due to resource constraints in the frontend of the processor.", "DefaultMetricgroupName": "TopdownL1", "MetricGroup": "Default;TopdownL1", "MetricName": "frontend_bound", - "ScaleUnit": "100%" + "ScaleUnit": "1percent of slots" }, { - "MetricExpr": "(1 - op_retired / op_spec) * (1 - stall_slot / (#slots * cpu_cycles))", - "BriefDescription": "Bad speculation L1 topdown metric", + "MetricExpr": "100 * ((1 - op_retired / op_spec) * (1 - stall_slot / (#slots * cpu_cycles)))", + "BriefDescription": "This metric is the percentage of total slots that executed operations and didn't retire due to a pipeline flush.\nThis indicates cycles that were utilized but inefficiently.", "DefaultMetricgroupName": "TopdownL1", "MetricGroup": "Default;TopdownL1", "MetricName": "bad_speculation", - "ScaleUnit": "100%" + "ScaleUnit": "1percent of slots" }, { - "MetricExpr": "(op_retired / op_spec) * (1 - stall_slot / (#slots * cpu_cycles))", - "BriefDescription": "Retiring L1 topdown metric", + "MetricExpr": "100 * ((op_retired / op_spec) * (1 - stall_slot / (#slots * cpu_cycles)))", + "BriefDescription": "This metric is the percentage of total slots that retired operations, which indicates cycles that were utilized efficiently.", "DefaultMetricgroupName": "TopdownL1", "MetricGroup": "Default;TopdownL1", "MetricName": "retiring", - "ScaleUnit": "100%" + "ScaleUnit": "1percent of slots" }, { - "MetricExpr": "stall_slot_backend / (#slots * cpu_cycles)", - "BriefDescription": "Backend Bound L1 topdown metric", + "MetricExpr": "100 * (stall_slot_backend / (#slots * cpu_cycles))", + "BriefDescription": "This metric is the percentage of total slots that were stalled due to resource constraints in the backend of the processor.", "DefaultMetricgroupName": "TopdownL1", "MetricGroup": "Default;TopdownL1", "MetricName": "backend_bound", - "ScaleUnit": "100%" + "ScaleUnit": "1percent of slots" } ]
Signature · 1 line
2.34.1
PATCH v2 4/5 perf vendor events arm64: Update N2-r0p3 and V2 metrics and events using Arm telemetry repo
In reply to James ClarkThe new metrics contain a fix for N2 r0p3 where CPU_CYCLES should not be subtracted from stalls for topdown metrics anymore. The current metrics assume that the fix should be applied anywhere where slots != 5, but this is only the case for V2 and not N2 r0p3.
Split the metrics into a new version for N2-r0p3 and V2 which still share the same metrics. Apart from some slight naming and grouping differences the new metrics are functionally the same as the existing ones. Any missing metrics were manually appended to the end of the auto generated file.
For the events, the new data includes descriptions that may have product specific details and new groupings that will be consistent with other products.
After generating the metrics from the telemetry repo [1], the following manual steps were performed:
* Change the hard coded slots in neoverse-n2r0p3-v2 to #slots so that it will work on both N2 and V2.
* Append some metrics from the old N2/V2 data that aren't present in the telemetry data. These will possibly be added to the telemetry-solution repo at a later time:
l3d_cache_mpki, l3d_cache_miss_rate, branch_pki, ipc_rate, spec_ipc, retired_rate, wasted_rate, load_spec_rate, store_spec_rate, advanced_simd_spec_rate, float_point_spec_rate, branch_immed_spec_rate, branch_return_spec_rate, branch_indirect_spec_rate
--- .../arm64/arm/neoverse-n2r0p3-v2/bus.json | 18 + .../arm/neoverse-n2r0p3-v2/exception.json | 62 ++++ .../arm/neoverse-n2r0p3-v2/fp_operation.json | 22 ++ .../arm64/arm/neoverse-n2r0p3-v2/general.json | 10 + .../arm/neoverse-n2r0p3-v2/l1d_cache.json | 54 +++ .../arm/neoverse-n2r0p3-v2/l1i_cache.json | 14 + .../arm/neoverse-n2r0p3-v2/l2_cache.json | 50 +++ .../arm/neoverse-n2r0p3-v2/l3_cache.json | 22 ++ .../arm/neoverse-n2r0p3-v2/ll_cache.json | 10 + .../arm64/arm/neoverse-n2r0p3-v2/memory.json | 46 +++ .../arm64/arm/neoverse-n2r0p3-v2/metrics.json | 331 ++++++++++++++++++ .../arm64/arm/neoverse-n2r0p3-v2/retired.json | 30 ++ .../arm64/arm/neoverse-n2r0p3-v2/spe.json | 18 + .../neoverse-n2r0p3-v2/spec_operation.json | 110 ++++++ .../arm64/arm/neoverse-n2r0p3-v2/stall.json | 30 ++ .../arm64/arm/neoverse-n2r0p3-v2/sve.json | 50 +++ .../arm64/arm/neoverse-n2r0p3-v2/tlb.json | 66 ++++ .../arm64/arm/neoverse-n2r0p3-v2/trace.json | 38 ++ tools/perf/pmu-events/arch/arm64/mapfile.csv | 3 +- 19 files changed, 983 insertions(+), 1 deletion(-) create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/bus.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/exception.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/fp_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/general.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l1d_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l1i_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l2_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l3_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/ll_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/memory.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/retired.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/spe.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/spec_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/stall.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/sve.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/tlb.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/trace.json
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/bus.json new file +18 -0
@@ -0,0 +1,18 @@ +[ + { + "ArchStdEvent": "BUS_ACCESS", + "PublicDescription": "Counts memory transactions issued by the CPU to the external bus, including snoop requests and snoop responses. Each beat of data is counted individually." + }, + { + "ArchStdEvent": "BUS_CYCLES", + "PublicDescription": "Counts bus cycles in the CPU. Bus cycles represent a clock cycle in which a transaction could be sent or received on the interface from the CPU to the external bus. Since that interface is driven at the same clock speed as the CPU, this event is a duplicate of CPU_CYCLES." + }, + { + "ArchStdEvent": "BUS_ACCESS_RD", + "PublicDescription": "Counts memory read transactions seen on the external bus. Each beat of data is counted individually." + }, + { + "ArchStdEvent": "BUS_ACCESS_WR", + "PublicDescription": "Counts memory write transactions seen on the external bus. Each beat of data is counted individually." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/exception.json new file +62 -0
62 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/fp_operation.json new file +22 -0
@@ -0,0 +1,22 @@ +[ + { + "ArchStdEvent": "FP_HP_SPEC", + "PublicDescription": "Counts speculatively executed half precision floating point operations." + }, + { + "ArchStdEvent": "FP_SP_SPEC", + "PublicDescription": "Counts speculatively executed single precision floating point operations." + }, + { + "ArchStdEvent": "FP_DP_SPEC", + "PublicDescription": "Counts speculatively executed double precision floating point operations." + }, + { + "ArchStdEvent": "FP_SCALE_OPS_SPEC", + "PublicDescription": "Counts speculatively executed scalable single precision floating point operations." + }, + { + "ArchStdEvent": "FP_FIXED_OPS_SPEC", + "PublicDescription": "Counts speculatively executed non-scalable single precision floating point operations." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/general.json new file +10 -0
@@ -0,0 +1,10 @@ +[ + { + "ArchStdEvent": "CPU_CYCLES", + "PublicDescription": "Counts CPU clock cycles (not timer cycles). The clock measured by this event is defined as the physical clock driving the CPU logic." + }, + { + "ArchStdEvent": "CNT_CYCLES", + "PublicDescription": "Counts constant frequency cycles" + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l1d_cache.json new file +54 -0
54 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l1i_cache.json new file +14 -0
@@ -0,0 +1,14 @@ +[ + { + "ArchStdEvent": "L1I_CACHE_REFILL", + "PublicDescription": "Counts cache line refills in the level 1 instruction cache caused by a missed instruction fetch. Instruction fetches may include accessing multiple instructions, but the single cache line allocation is counted once." + }, + { + "ArchStdEvent": "L1I_CACHE", + "PublicDescription": "Counts instruction fetches which access the level 1 instruction cache. Instruction cache accesses caused by cache maintenance operations are not counted." + }, + { + "ArchStdEvent": "L1I_CACHE_LMISS", + "PublicDescription": "Counts cache line refills into the level 1 instruction cache, that incurred additional latency." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l2_cache.json new file +50 -0
50 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/l3_cache.json new file +22 -0
@@ -0,0 +1,22 @@ +[ + { + "ArchStdEvent": "L3D_CACHE_ALLOCATE", + "PublicDescription": "Counts level 3 cache line allocates that do not fetch data from outside the level 3 data or unified cache. For example, allocates due to streaming stores." + }, + { + "ArchStdEvent": "L3D_CACHE_REFILL", + "PublicDescription": "Counts level 3 accesses that receive data from outside the L3 cache." + }, + { + "ArchStdEvent": "L3D_CACHE", + "PublicDescription": "Counts level 3 cache accesses. level 3 cache is a unified cache for data and instruction accesses. Accesses are for misses in the lower level caches or translation resolutions due to accesses." + }, + { + "ArchStdEvent": "L3D_CACHE_RD", + "PublicDescription": "TBD" + }, + { + "ArchStdEvent": "L3D_CACHE_LMISS_RD", + "PublicDescription": "Counts any cache line refill into the level 3 cache from memory read operations that incurred additional latency." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/ll_cache.json new file +10 -0
@@ -0,0 +1,10 @@ +[ + { + "ArchStdEvent": "LL_CACHE_RD", + "PublicDescription": "Counts read transactions that were returned from outside the core cluster. This event counts when the system register CPUECTLR.EXTLLC bit is set. This event counts read transactions returned from outside the core if those transactions are either hit in the system level cache or missed in the SLC and are returned from any other external sources." + }, + { + "ArchStdEvent": "LL_CACHE_MISS_RD", + "PublicDescription": "Counts read transactions that were returned from outside the core cluster but missed in the system level cache. This event counts when the system register CPUECTLR.EXTLLC bit is set. This event counts read transactions returned from outside the core if those transactions are missed in the System level Cache. The data source of the transaction is indicated by a field in the CHI transaction returning to the CPU. This event does not count reads caused by cache maintenance operations." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/memory.json new file +46 -0
46 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json new file +331 -0
331 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/retired.json new file +30 -0
@@ -0,0 +1,30 @@ +[ + { + "ArchStdEvent": "SW_INCR", + "PublicDescription": "Counts software writes to the PMSWINC_EL0 (software PMU increment) register. The PMSWINC_EL0 register is a manually updated counter for use by application software.\n\nThis event could be used to measure any user program event, such as accesses to a particular data structure (by writing to the PMSWINC_EL0 register each time the data structure is accessed).\n\nTo use the PMSWINC_EL0 register and event, developers must insert instructions that write to the PMSWINC_EL0 register into the source code.\n\nSince the SW_INCR event records writes to the PMSWINC_EL0 register, there is no need to do a read/increment/write sequence to the PMSWINC_EL0 register." + }, + { + "ArchStdEvent": "INST_RETIRED", + "PublicDescription": "Counts instructions that have been architecturally executed." + }, + { + "ArchStdEvent": "CID_WRITE_RETIRED", + "PublicDescription": "Counts architecturally executed writes to the CONTEXTIDR register, which usually contain the kernel PID and can be output with hardware trace." + }, + { + "ArchStdEvent": "TTBR_WRITE_RETIRED", + "PublicDescription": "Counts architectural writes to TTBR0/1_EL1. If virtualization host extensions are enabled (by setting the HCR_EL2.E2H bit to 1), then accesses to TTBR0/1_EL1 that are redirected to TTBR0/1_EL2, or accesses to TTBR0/1_EL12, are counted. TTBRn registers are typically updated when the kernel is swapping user-space threads or applications." + }, + { + "ArchStdEvent": "BR_RETIRED", + "PublicDescription": "Counts architecturally executed branches, whether the branch is taken or not. Instructions that explicitly write to the PC are also counted." + }, + { + "ArchStdEvent": "BR_MIS_PRED_RETIRED", + "PublicDescription": "Counts branches counted by BR_RETIRED which were mispredicted and caused a pipeline flush." + }, + { + "ArchStdEvent": "OP_RETIRED", + "PublicDescription": "Counts micro-operations that are architecturally executed. This is a count of number of micro-operations retired from the commit queue in a single cycle." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/spe.json new file +18 -0
@@ -0,0 +1,18 @@ +[ + { + "ArchStdEvent": "SAMPLE_POP", + "PublicDescription": "Counts statistical profiling sample population, the count of all operations that could be sampled but may or may not be chosen for sampling." + }, + { + "ArchStdEvent": "SAMPLE_FEED", + "PublicDescription": "Counts statistical profiling samples taken for sampling." + }, + { + "ArchStdEvent": "SAMPLE_FILTRATE", + "PublicDescription": "Counts statistical profiling samples taken which are not removed by filtering." + }, + { + "ArchStdEvent": "SAMPLE_COLLISION", + "PublicDescription": "Counts statistical profiling samples that have collided with a previous sample and so therefore not taken." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/spec_operation.json new file +110 -0
110 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/stall.json new file +30 -0
@@ -0,0 +1,30 @@ +[ + { + "ArchStdEvent": "STALL_FRONTEND", + "PublicDescription": "Counts cycles when frontend could not send any micro-operations to the rename stage because of frontend resource stalls caused by fetch memory latency or branch prediction flow stalls. All the frontend slots were empty during the cycle when this event counts." + }, + { + "ArchStdEvent": "STALL_BACKEND", + "PublicDescription": "Counts cycles whenever the rename unit is unable to send any micro-operations to the backend of the pipeline because of backend resource constraints. Backend resource constraints can include issue stage fullness, execution stage fullness, or other internal pipeline resource fullness. All the backend slots were empty during the cycle when this event counts." + }, + { + "ArchStdEvent": "STALL", + "PublicDescription": "Counts cycles when no operations are sent to the rename unit from the frontend or from the rename unit to the backend for any reason (either frontend or backend stall)." + }, + { + "ArchStdEvent": "STALL_SLOT_BACKEND", + "PublicDescription": "Counts slots per cycle in which no operations are sent from the rename unit to the backend due to backend resource constraints." + }, + { + "ArchStdEvent": "STALL_SLOT_FRONTEND", + "PublicDescription": "Counts slots per cycle in which no operations are sent to the rename unit from the frontend due to frontend resource constraints." + }, + { + "ArchStdEvent": "STALL_SLOT", + "PublicDescription": "Counts slots per cycle in which no operations are sent to the rename unit from the frontend or from the rename unit to the backend for any reason (either frontend or backend stall)." + }, + { + "ArchStdEvent": "STALL_BACKEND_MEM", + "PublicDescription": "Counts cycles when the backend is stalled because there is a pending demand load request in progress in the last level core cache." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/sve.json new file +50 -0
50 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/tlb.json new file +66 -0
66 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/trace.json new file +38 -0
@@ -0,0 +1,38 @@ +[ + { + "ArchStdEvent": "TRB_WRAP", + "PublicDescription": "This event is generated each time the current write pointer is wrapped to the base pointer." + }, + { + "ArchStdEvent": "TRCEXTOUT0", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 0." + }, + { + "ArchStdEvent": "TRCEXTOUT1", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 1." + }, + { + "ArchStdEvent": "TRCEXTOUT2", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 2." + }, + { + "ArchStdEvent": "TRCEXTOUT3", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 3." + }, + { + "ArchStdEvent": "CTI_TRIGOUT4", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 4." + }, + { + "ArchStdEvent": "CTI_TRIGOUT5", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 5." + }, + { + "ArchStdEvent": "CTI_TRIGOUT6", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 6." + }, + { + "ArchStdEvent": "CTI_TRIGOUT7", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 7." + } +]
tools/perf/pmu-events/arch/arm64/mapfile.csv +2 -1
@@ -35,7 +35,8 @@ 0x00000000410fd470,v1,arm/cortex-a710,core 0x00000000410fd480,v1,arm/cortex-x2,core 0x00000000410fd490,v1,arm/neoverse-n2-v2,core -0x00000000410fd4f0,v1,arm/neoverse-n2-v2,core +0x00000000410fd493,v1,arm/neoverse-n2r0p3-v2,core +0x00000000410fd4f0,v1,arm/neoverse-n2r0p3-v2,core 0x00000000420f5160,v1,cavium/thunderx2,core 0x00000000430f0af0,v1,cavium/thunderx2,core 0x00000000460f0010,v1,fujitsu/a64fx,core
Signature · 1 line
2.34.1
Re: [PATCH v2 4/5] perf vendor events arm64: Update N2-r0p3 and V2 metrics and events using Arm telemetry repo
Quoted 35 lines · James Clark
The new metrics contain a fix for N2 r0p3 where
CPU_CYCLESshould not be subtracted from stalls for topdown metrics anymore. The current metrics assume that the fix should be applied anywhere where slots != 5, but this is only the case for V2 and not N2 r0p3.Split the metrics into a new version for N2-r0p3 and V2 which still share the same metrics. Apart from some slight naming and grouping differences the new metrics are functionally the same as the existing ones. Any missing metrics were manually appended to the end of the auto generated file.
For the events, the new data includes descriptions that may have product specific details and new groupings that will be consistent with other products.
After generating the metrics from the telemetry repo [1], the following manual steps were performed:
* Change the hard coded slots in neoverse-n2r0p3-v2 to #slots so that it will work on both N2 and V2.
* Append some metrics from the old N2/V2 data that aren't present in the telemetry data. These will possibly be added to the telemetry-solution repo at a later time:
l3d_cache_mpki,l3d_cache_miss_rate,branch_pki,ipc_rate,spec_ipc,retired_rate,wasted_rate,load_spec_rate,store_spec_rate,advanced_simd_spec_rate,float_point_spec_rate,branch_immed_spec_rate,branch_return_spec_rate,branch_indirect_spec_rateSigned-off-by James Clark <[email protected]>---
[...]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json line 5
Quoted 11 lines
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json new file +5
-0@@ -0,0 +1,331 @@ +[ + { + "ArchStdEvent": "backend_bound", + "MetricExpr": "(100 * ((STALL_SLOT_BACKEND / (CPU_CYCLES * #slots)) - ((BR_MIS_PRED * 3) / CPU_CYCLES)))" + },
Hi Jing,
I'm not sure if you remember, but a long time ago I said that I was going to update these N2 and V2 metrics that you upstreamed.
Now that these are coming from the Arm telemetry repo they have an adjustment for subtracting branch misses that we think is more accurate because of an issue with the STALL_SLOT_BACKEND counter. I've left the sbsa.json metrics without the adjustments as it shouldn't normally be needed.
I also had to drop the "if (#slots - 5) else" check because it won't apply to N2-r0p3, but I split the metrics into two instead to fix that issue.
The rest is pretty much the same apart from some slight grouping/naming/ scaling differences.
Let me know what you think.
Thanks
James
Quoted 18 lines
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json new file +18
-0+ { + "MetricName": "backend_stalled_cycles", + "MetricExpr": "((STALL_BACKEND / CPU_CYCLES) * 100)", + "BriefDescription": "This metric is the percentage of cycles that were stalled due to resource constraints in the backend unit of the processor.", + "MetricGroup": "Cycle_Accounting", + "ScaleUnit": "1percent of cycles" + }, + { + "ArchStdEvent": "bad_speculation", + "MetricExpr": "(100 * (((1 - (OP_RETIRED / OP_SPEC)) * (1 - (STALL_SLOT / (CPU_CYCLES * #slots)))) + ((BR_MIS_PRED * 4) / CPU_CYCLES)))" + }, + { + "MetricName": "branch_misprediction_ratio", + "MetricExpr": "(BR_MIS_PRED_RETIRED / BR_RETIRED)", + "BriefDescription": "This metric measures the ratio of branches mispredicted to the total number of branches architecturally executed. This gives an indication of the effectiveness of the branch prediction unit.", + "MetricGroup": "Miss_Ratio;Branch_Effectiveness", + "ScaleUnit": "1per branch" + },
Re: [PATCH v2 4/5] perf vendor events arm64: Update N2-r0p3 and V2 metrics and events using Arm telemetry repo
Quoted 64 lines · James Clark
Quoted 35 lines · James Clark
The new metrics contain a fix for N2 r0p3 where
CPU_CYCLESshould not be subtracted from stalls for topdown metrics anymore. The current metrics assume that the fix should be applied anywhere where slots != 5, but this is only the case for V2 and not N2 r0p3.Split the metrics into a new version for N2-r0p3 and V2 which still share the same metrics. Apart from some slight naming and grouping differences the new metrics are functionally the same as the existing ones. Any missing metrics were manually appended to the end of the auto generated file.
For the events, the new data includes descriptions that may have product specific details and new groupings that will be consistent with other products.
After generating the metrics from the telemetry repo [1], the following manual steps were performed:
* Change the hard coded slots in neoverse-n2r0p3-v2 to #slots so that it will work on both N2 and V2.
* Append some metrics from the old N2/V2 data that aren't present in the telemetry data. These will possibly be added to the telemetry-solution repo at a later time:
l3d_cache_mpki,l3d_cache_miss_rate,branch_pki,ipc_rate,spec_ipc,retired_rate,wasted_rate,load_spec_rate,store_spec_rate,advanced_simd_spec_rate,float_point_spec_rate,branch_immed_spec_rate,branch_return_spec_rate,branch_indirect_spec_rateSigned-off-by James Clark <[email protected]>---
[...]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json line 5
Quoted 11 lines
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json new file +5
-0@@ -0,0 +1,331 @@ +[ + { + "ArchStdEvent": "backend_bound", + "MetricExpr": "(100 * ((STALL_SLOT_BACKEND / (CPU_CYCLES * #slots)) - ((BR_MIS_PRED * 3) / CPU_CYCLES)))" + },Hi Jing,
I'm not sure if you remember, but a long time ago I said that I was going to update these N2 and V2 metrics that you upstreamed.
Now that these are coming from the Arm telemetry repo they have an adjustment for subtracting branch misses that we think is more accurate because of an issue with the
STALL_SLOT_BACKENDcounter. I've left the sbsa.json metrics without the adjustments as it shouldn't normally be needed.
Quoted 9 lines
I also had to drop the "if (#slots - 5) else" check because it won't apply to N2-r0p3, but I split the metrics into two instead to fix that issue.
The rest is pretty much the same apart from some slight grouping/naming/ scaling differences.
Let me know what you think.
Hi James,
Sorry, I just saw it now, I think it is ok to update the metric. But because only the metric is different, is it a bit redundant to copy all the events?
For uncore_sys_pmu_event, there is a way to deal with this problem, which is to identify different rivision identifiers through a "Compat" field. So I wonder if it is also possible to add a field similar to "Compat" when describing the metric, so that the metric can be matched to different revisions. For example:
+ {
+ "ArchStdEvent": "frontend_bound",
+ "MetricExpr": "(100 * ((STALL_SLOT_FRONTEND / (CPU_CYCLES * #slots)) - (BR_MIS_PRED / CPU_CYCLES)))"
+ "Compat": "0x00000000410fd493;0x00000000410fd4f0"
+ },
+ {
+ "ArchStdEvent": "frontend_bound",
+ "MetricExpr": "(100 * (((STALL_SLOT_FRONTEND - CPU_CYCLES) / (5 * CPU_CYCLES)) - (BR_MIS_PRED / CPU_CYCLES)))"
+ "Compat": "0x00000000410fd490"
+ },
With this “Compat” field, it only matches the specified revision, and if the “Compat” field is not added, it can match all revisions. The 'Compat' field for the Core PMU is currently not supported. However, I believe that adding this is not complex, and it also has good readability without causing redundancy. Would it be worth a try?
What do you think?
Thanks,
Jing
Re: [PATCH v2 4/5] perf vendor events arm64: Update N2-r0p3 and V2 metrics and events using Arm telemetry repo
Quoted 106 lines · Jing Zhang
Quoted 64 lines · James Clark
Quoted 35 lines · James Clark
The new metrics contain a fix for N2 r0p3 where
CPU_CYCLESshould not be subtracted from stalls for topdown metrics anymore. The current metrics assume that the fix should be applied anywhere where slots != 5, but this is only the case for V2 and not N2 r0p3.Split the metrics into a new version for N2-r0p3 and V2 which still share the same metrics. Apart from some slight naming and grouping differences the new metrics are functionally the same as the existing ones. Any missing metrics were manually appended to the end of the auto generated file.
For the events, the new data includes descriptions that may have product specific details and new groupings that will be consistent with other products.
After generating the metrics from the telemetry repo [1], the following manual steps were performed:
* Change the hard coded slots in neoverse-n2r0p3-v2 to #slots so that it will work on both N2 and V2.
* Append some metrics from the old N2/V2 data that aren't present in the telemetry data. These will possibly be added to the telemetry-solution repo at a later time:
l3d_cache_mpki,l3d_cache_miss_rate,branch_pki,ipc_rate,spec_ipc,retired_rate,wasted_rate,load_spec_rate,store_spec_rate,advanced_simd_spec_rate,float_point_spec_rate,branch_immed_spec_rate,branch_return_spec_rate,branch_indirect_spec_rateSigned-off-by James Clark <[email protected]>---
[...]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json line 5
Quoted 11 lines
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json new file +5
-0@@ -0,0 +1,331 @@ +[ + { + "ArchStdEvent": "backend_bound", + "MetricExpr": "(100 * ((STALL_SLOT_BACKEND / (CPU_CYCLES * #slots)) - ((BR_MIS_PRED * 3) / CPU_CYCLES)))" + },Hi Jing,
I'm not sure if you remember, but a long time ago I said that I was going to update these N2 and V2 metrics that you upstreamed.
Now that these are coming from the Arm telemetry repo they have an adjustment for subtracting branch misses that we think is more accurate because of an issue with the
STALL_SLOT_BACKENDcounter. I've left the sbsa.json metrics without the adjustments as it shouldn't normally be needed.Quoted 9 lines
I also had to drop the "if (#slots - 5) else" check because it won't apply to N2-r0p3, but I split the metrics into two instead to fix that issue.
The rest is pretty much the same apart from some slight grouping/naming/ scaling differences.
Let me know what you think.
Hi James,
Sorry, I just saw it now, I think it is ok to update the metric. But because only the metric is different, is it a bit redundant to copy all the events?
For
uncore_sys_pmu_event, there is a way to deal with this problem, which is to identify different rivision identifiers through a "Compat" field. So I wonder if it is also possible to add a field similar to "Compat" when describing the metric, so that the metric can be matched to different revisions. For example:tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p3-v2/metrics.json new file +10
-0+ { + "ArchStdEvent": "frontend_bound", + "MetricExpr": "(100 * ((STALL_SLOT_FRONTEND / (CPU_CYCLES * #slots)) - (BR_MIS_PRED / CPU_CYCLES)))" + "Compat": "0x00000000410fd493;0x00000000410fd4f0" + }, + { + "ArchStdEvent": "frontend_bound", + "MetricExpr": "(100 * (((STALL_SLOT_FRONTEND - CPU_CYCLES) / (5 * CPU_CYCLES)) - (BR_MIS_PRED / CPU_CYCLES)))" + "Compat": "0x00000000410fd490" + },With this “Compat” field, it only matches the specified revision, and if the “Compat” field is not added, it can match all revisions. The 'Compat' field for the Core PMU is currently not supported. However, I believe that adding this is not complex, and it also has good readability without causing redundancy. Would it be worth a try?
Quoted 5 lines
What do you think?
Thanks,
Jing
Hi Jing,
I posted a V4 on the list. I went with John's approach of modifying the metric expression literal so that the V2 and N2 metrics could be re-used. I thought that was simpler than the "Compat" field. And I thought that it might be duplicating some of what the mapfile.csv already does.
Instead of adding this compat field for core PMUs, if we want to do something like this, maybe we could make it so that new versions of CPUs can be added in the mapfile, but they append and modify all the previous matching versions of JSONs found. So in this case it would work like I add some new JSONs for N2 r0p3, but they only contain the modified metrics, rather than everything, and that overwrites that part of the r0p0 metrics. That way we could add new items in the existing mapfile but they choose how much or how little of the previous revisions they would like to replace.
Otherwise we could end up being in a situation where instead of the mapfile we just add the compat field to all metrics and then we can delete the mapfile and everything still works. So it seems like it could end up being two different ways to do the same thing. I'm not completely against this idea, but I suppose in this exact scenario the new literal was less change and works pretty well.
Thanks
James
PATCH v2 5/5 perf vendor events arm64: Update N2-r0p0 metrics and events using Arm telemetry repo
In reply to James ClarkApart from some slight naming and grouping differences the new metrics are functionally the same as the existing ones. Any missing metrics were manually appended to the end of the auto generated file.
For the events, the new data includes descriptions that may have product specific details and new groupings that will be consistent with other products.
After generating the metrics from the telemetry repo [1], the following manual steps were performed:
* Append some metrics from the old N2/V2 data that aren't present in the telemetry data. These will possibly be added to the telemetry-solution repo at a later time:
l3d_cache_mpki, l3d_cache_miss_rate, branch_pki, ipc_rate, spec_ipc, retired_rate, wasted_rate, load_spec_rate, store_spec_rate, advanced_simd_spec_rate, float_point_spec_rate, branch_immed_spec_rate, branch_return_spec_rate, branch_indirect_spec_rate
--- .../arch/arm64/arm/neoverse-n2-v2/branch.json | 8 - .../arch/arm64/arm/neoverse-n2-v2/bus.json | 20 -- .../arch/arm64/arm/neoverse-n2-v2/cache.json | 155 -------- .../arm64/arm/neoverse-n2-v2/exception.json | 47 --- .../arm64/arm/neoverse-n2-v2/instruction.json | 143 -------- .../arch/arm64/arm/neoverse-n2-v2/memory.json | 41 --- .../arm64/arm/neoverse-n2-v2/metrics.json | 273 -------------- .../arm64/arm/neoverse-n2-v2/pipeline.json | 23 -- .../arch/arm64/arm/neoverse-n2-v2/spe.json | 14 - .../arch/arm64/arm/neoverse-n2-v2/trace.json | 29 -- .../arch/arm64/arm/neoverse-n2r0p0/bus.json | 18 + .../arm64/arm/neoverse-n2r0p0/exception.json | 62 ++++ .../arm/neoverse-n2r0p0/fp_operation.json | 22 ++ .../arm64/arm/neoverse-n2r0p0/general.json | 10 + .../arm64/arm/neoverse-n2r0p0/l1d_cache.json | 54 +++ .../arm64/arm/neoverse-n2r0p0/l1i_cache.json | 14 + .../arm64/arm/neoverse-n2r0p0/l2_cache.json | 50 +++ .../arm64/arm/neoverse-n2r0p0/l3_cache.json | 22 ++ .../arm64/arm/neoverse-n2r0p0/ll_cache.json | 10 + .../arm64/arm/neoverse-n2r0p0/memory.json | 46 +++ .../arm64/arm/neoverse-n2r0p0/metrics.json | 332 ++++++++++++++++++ .../arm64/arm/neoverse-n2r0p0/retired.json | 30 ++ .../arch/arm64/arm/neoverse-n2r0p0/spe.json | 18 + .../arm/neoverse-n2r0p0/spec_operation.json | 110 ++++++ .../arch/arm64/arm/neoverse-n2r0p0/stall.json | 30 ++ .../arch/arm64/arm/neoverse-n2r0p0/sve.json | 50 +++ .../arch/arm64/arm/neoverse-n2r0p0/tlb.json | 66 ++++ .../arch/arm64/arm/neoverse-n2r0p0/trace.json | 38 ++ tools/perf/pmu-events/arch/arm64/mapfile.csv | 2 +- 29 files changed, 983 insertions(+), 754 deletions(-) delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/branch.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/bus.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/cache.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/exception.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/instruction.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/memory.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/metrics.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/pipeline.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/spe.json delete mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/trace.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/bus.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/exception.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/fp_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/general.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l1d_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l1i_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l2_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l3_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/ll_cache.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/memory.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/metrics.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/retired.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/spe.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/spec_operation.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/stall.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/sve.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/tlb.json create mode 100644 tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/trace.json
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/branch.json deleted +0 -8
@@ -1,8 +0,0 @@ -[ - { - "ArchStdEvent": "BR_MIS_PRED" - }, - { - "ArchStdEvent": "BR_PRED" - } -]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/bus.json deleted +0 -20
@@ -1,20 +0,0 @@ -[ - { - "ArchStdEvent": "CPU_CYCLES" - }, - { - "ArchStdEvent": "BUS_ACCESS" - }, - { - "ArchStdEvent": "BUS_CYCLES" - }, - { - "ArchStdEvent": "BUS_ACCESS_RD" - }, - { - "ArchStdEvent": "BUS_ACCESS_WR" - }, - { - "ArchStdEvent": "CNT_CYCLES" - } -]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/cache.json deleted +0 -155
155 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/exception.json deleted +0 -47
47 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/instruction.json deleted +0 -143
143 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/memory.json deleted +0 -41
41 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/metrics.json deleted +0 -273
273 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/pipeline.json deleted +0 -23
@@ -1,23 +0,0 @@ -[ - { - "ArchStdEvent": "STALL_FRONTEND" - }, - { - "ArchStdEvent": "STALL_BACKEND" - }, - { - "ArchStdEvent": "STALL" - }, - { - "ArchStdEvent": "STALL_SLOT_BACKEND" - }, - { - "ArchStdEvent": "STALL_SLOT_FRONTEND" - }, - { - "ArchStdEvent": "STALL_SLOT" - }, - { - "ArchStdEvent": "STALL_BACKEND_MEM" - } -]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/spe.json deleted +0 -14
@@ -1,14 +0,0 @@ -[ - { - "ArchStdEvent": "SAMPLE_POP" - }, - { - "ArchStdEvent": "SAMPLE_FEED" - }, - { - "ArchStdEvent": "SAMPLE_FILTRATE" - }, - { - "ArchStdEvent": "SAMPLE_COLLISION" - } -]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2-v2/trace.json deleted +0 -29
@@ -1,29 +0,0 @@ -[ - { - "ArchStdEvent": "TRB_WRAP" - }, - { - "ArchStdEvent": "TRCEXTOUT0" - }, - { - "ArchStdEvent": "TRCEXTOUT1" - }, - { - "ArchStdEvent": "TRCEXTOUT2" - }, - { - "ArchStdEvent": "TRCEXTOUT3" - }, - { - "ArchStdEvent": "CTI_TRIGOUT4" - }, - { - "ArchStdEvent": "CTI_TRIGOUT5" - }, - { - "ArchStdEvent": "CTI_TRIGOUT6" - }, - { - "ArchStdEvent": "CTI_TRIGOUT7" - } -]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/bus.json new file +18 -0
@@ -0,0 +1,18 @@ +[ + { + "ArchStdEvent": "BUS_ACCESS", + "PublicDescription": "Counts memory transactions issued by the CPU to the external bus, including snoop requests and snoop responses. Each beat of data is counted individually." + }, + { + "ArchStdEvent": "BUS_CYCLES", + "PublicDescription": "Counts bus cycles in the CPU. Bus cycles represent a clock cycle in which a transaction could be sent or received on the interface from the CPU to the external bus. Since that interface is driven at the same clock speed as the CPU, this event is a duplicate of CPU_CYCLES." + }, + { + "ArchStdEvent": "BUS_ACCESS_RD", + "PublicDescription": "Counts memory read transactions seen on the external bus. Each beat of data is counted individually." + }, + { + "ArchStdEvent": "BUS_ACCESS_WR", + "PublicDescription": "Counts memory write transactions seen on the external bus. Each beat of data is counted individually." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/exception.json new file +62 -0
62 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/fp_operation.json new file +22 -0
@@ -0,0 +1,22 @@ +[ + { + "ArchStdEvent": "FP_HP_SPEC", + "PublicDescription": "Counts speculatively executed half precision floating point operations." + }, + { + "ArchStdEvent": "FP_SP_SPEC", + "PublicDescription": "Counts speculatively executed single precision floating point operations." + }, + { + "ArchStdEvent": "FP_DP_SPEC", + "PublicDescription": "Counts speculatively executed double precision floating point operations." + }, + { + "ArchStdEvent": "FP_SCALE_OPS_SPEC", + "PublicDescription": "Counts speculatively executed scalable single precision floating point operations." + }, + { + "ArchStdEvent": "FP_FIXED_OPS_SPEC", + "PublicDescription": "Counts speculatively executed non-scalable single precision floating point operations." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/general.json new file +10 -0
@@ -0,0 +1,10 @@ +[ + { + "ArchStdEvent": "CPU_CYCLES", + "PublicDescription": "Counts CPU clock cycles (not timer cycles). The clock measured by this event is defined as the physical clock driving the CPU logic." + }, + { + "ArchStdEvent": "CNT_CYCLES", + "PublicDescription": "Counts constant frequency cycles" + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l1d_cache.json new file +54 -0
54 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l1i_cache.json new file +14 -0
@@ -0,0 +1,14 @@ +[ + { + "ArchStdEvent": "L1I_CACHE_REFILL", + "PublicDescription": "Counts cache line refills in the level 1 instruction cache caused by a missed instruction fetch. Instruction fetches may include accessing multiple instructions, but the single cache line allocation is counted once." + }, + { + "ArchStdEvent": "L1I_CACHE", + "PublicDescription": "Counts instruction fetches which access the level 1 instruction cache. Instruction cache accesses caused by cache maintenance operations are not counted." + }, + { + "ArchStdEvent": "L1I_CACHE_LMISS", + "PublicDescription": "Counts cache line refills into the level 1 instruction cache, that incurred additional latency." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l2_cache.json new file +50 -0
50 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/l3_cache.json new file +22 -0
@@ -0,0 +1,22 @@ +[ + { + "ArchStdEvent": "L3D_CACHE_ALLOCATE", + "PublicDescription": "Counts level 3 cache line allocates that do not fetch data from outside the level 3 data or unified cache. For example, allocates due to streaming stores." + }, + { + "ArchStdEvent": "L3D_CACHE_REFILL", + "PublicDescription": "Counts level 3 accesses that receive data from outside the L3 cache." + }, + { + "ArchStdEvent": "L3D_CACHE", + "PublicDescription": "Counts level 3 cache accesses. level 3 cache is a unified cache for data and instruction accesses. Accesses are for misses in the lower level caches or translation resolutions due to accesses." + }, + { + "ArchStdEvent": "L3D_CACHE_RD", + "PublicDescription": "TBD" + }, + { + "ArchStdEvent": "L3D_CACHE_LMISS_RD", + "PublicDescription": "Counts any cache line refill into the level 3 cache from memory read operations that incurred additional latency." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/ll_cache.json new file +10 -0
@@ -0,0 +1,10 @@ +[ + { + "ArchStdEvent": "LL_CACHE_RD", + "PublicDescription": "Counts read transactions that were returned from outside the core cluster. This event counts when the system register CPUECTLR.EXTLLC bit is set. This event counts read transactions returned from outside the core if those transactions are either hit in the system level cache or missed in the SLC and are returned from any other external sources." + }, + { + "ArchStdEvent": "LL_CACHE_MISS_RD", + "PublicDescription": "Counts read transactions that were returned from outside the core cluster but missed in the system level cache. This event counts when the system register CPUECTLR.EXTLLC bit is set. This event counts read transactions returned from outside the core if those transactions are missed in the System level Cache. The data source of the transaction is indicated by a field in the CHI transaction returning to the CPU. This event does not count reads caused by cache maintenance operations." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/memory.json new file +46 -0
46 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/metrics.json new file +332 -0
332 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/retired.json new file +30 -0
@@ -0,0 +1,30 @@ +[ + { + "ArchStdEvent": "SW_INCR", + "PublicDescription": "Counts software writes to the PMSWINC_EL0 (software PMU increment) register. The PMSWINC_EL0 register is a manually updated counter for use by application software.\n\nThis event could be used to measure any user program event, such as accesses to a particular data structure (by writing to the PMSWINC_EL0 register each time the data structure is accessed).\n\nTo use the PMSWINC_EL0 register and event, developers must insert instructions that write to the PMSWINC_EL0 register into the source code.\n\nSince the SW_INCR event records writes to the PMSWINC_EL0 register, there is no need to do a read/increment/write sequence to the PMSWINC_EL0 register." + }, + { + "ArchStdEvent": "INST_RETIRED", + "PublicDescription": "Counts instructions that have been architecturally executed." + }, + { + "ArchStdEvent": "CID_WRITE_RETIRED", + "PublicDescription": "Counts architecturally executed writes to the CONTEXTIDR register, which usually contain the kernel PID and can be output with hardware trace." + }, + { + "ArchStdEvent": "TTBR_WRITE_RETIRED", + "PublicDescription": "Counts architectural writes to TTBR0/1_EL1. If virtualization host extensions are enabled (by setting the HCR_EL2.E2H bit to 1), then accesses to TTBR0/1_EL1 that are redirected to TTBR0/1_EL2, or accesses to TTBR0/1_EL12, are counted. TTBRn registers are typically updated when the kernel is swapping user-space threads or applications." + }, + { + "ArchStdEvent": "BR_RETIRED", + "PublicDescription": "Counts architecturally executed branches, whether the branch is taken or not. Instructions that explicitly write to the PC are also counted." + }, + { + "ArchStdEvent": "BR_MIS_PRED_RETIRED", + "PublicDescription": "Counts branches counted by BR_RETIRED which were mispredicted and caused a pipeline flush." + }, + { + "ArchStdEvent": "OP_RETIRED", + "PublicDescription": "Counts micro-operations that are architecturally executed. This is a count of number of micro-operations retired from the commit queue in a single cycle." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/spe.json new file +18 -0
@@ -0,0 +1,18 @@ +[ + { + "ArchStdEvent": "SAMPLE_POP", + "PublicDescription": "Counts statistical profiling sample population, the count of all operations that could be sampled but may or may not be chosen for sampling." + }, + { + "ArchStdEvent": "SAMPLE_FEED", + "PublicDescription": "Counts statistical profiling samples taken for sampling." + }, + { + "ArchStdEvent": "SAMPLE_FILTRATE", + "PublicDescription": "Counts statistical profiling samples taken which are not removed by filtering." + }, + { + "ArchStdEvent": "SAMPLE_COLLISION", + "PublicDescription": "Counts statistical profiling samples that have collided with a previous sample and so therefore not taken." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/spec_operation.json new file +110 -0
110 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/stall.json new file +30 -0
@@ -0,0 +1,30 @@ +[ + { + "ArchStdEvent": "STALL_FRONTEND", + "PublicDescription": "Counts cycles when frontend could not send any micro-operations to the rename stage because of frontend resource stalls caused by fetch memory latency or branch prediction flow stalls. All the frontend slots were empty during the cycle when this event counts." + }, + { + "ArchStdEvent": "STALL_BACKEND", + "PublicDescription": "Counts cycles whenever the rename unit is unable to send any micro-operations to the backend of the pipeline because of backend resource constraints. Backend resource constraints can include issue stage fullness, execution stage fullness, or other internal pipeline resource fullness. All the backend slots were empty during the cycle when this event counts." + }, + { + "ArchStdEvent": "STALL", + "PublicDescription": "Counts cycles when no operations are sent to the rename unit from the frontend or from the rename unit to the backend for any reason (either frontend or backend stall)." + }, + { + "ArchStdEvent": "STALL_SLOT_BACKEND", + "PublicDescription": "Counts slots per cycle in which no operations are sent from the rename unit to the backend due to backend resource constraints." + }, + { + "ArchStdEvent": "STALL_SLOT_FRONTEND", + "PublicDescription": "Counts slots per cycle in which no operations are sent to the rename unit from the frontend due to frontend resource constraints." + }, + { + "ArchStdEvent": "STALL_SLOT", + "PublicDescription": "Counts slots per cycle in which no operations are sent to the rename unit from the frontend or from the rename unit to the backend for any reason (either frontend or backend stall)." + }, + { + "ArchStdEvent": "STALL_BACKEND_MEM", + "PublicDescription": "Counts cycles when the backend is stalled because there is a pending demand load request in progress in the last level core cache." + } +]
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/sve.json new file +50 -0
50 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/tlb.json new file +66 -0
66 lines not on this page · raw message
tools/perf/pmu-events/arch/arm64/arm/neoverse-n2r0p0/trace.json new file +38 -0
@@ -0,0 +1,38 @@ +[ + { + "ArchStdEvent": "TRB_WRAP", + "PublicDescription": "This event is generated each time the current write pointer is wrapped to the base pointer." + }, + { + "ArchStdEvent": "TRCEXTOUT0", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 0." + }, + { + "ArchStdEvent": "TRCEXTOUT1", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 1." + }, + { + "ArchStdEvent": "TRCEXTOUT2", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 2." + }, + { + "ArchStdEvent": "TRCEXTOUT3", + "PublicDescription": "This event is generated each time an event is signaled by ETE external event 3." + }, + { + "ArchStdEvent": "CTI_TRIGOUT4", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 4." + }, + { + "ArchStdEvent": "CTI_TRIGOUT5", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 5." + }, + { + "ArchStdEvent": "CTI_TRIGOUT6", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 6." + }, + { + "ArchStdEvent": "CTI_TRIGOUT7", + "PublicDescription": "This event is generated each time an event is signaled on CTI output trigger 7." + } +]
tools/perf/pmu-events/arch/arm64/mapfile.csv +1 -1
@@ -34,7 +34,7 @@ 0x00000000410fd460,v1,arm/cortex-a510,core 0x00000000410fd470,v1,arm/cortex-a710,core 0x00000000410fd480,v1,arm/cortex-x2,core -0x00000000410fd490,v1,arm/neoverse-n2-v2,core +0x00000000410fd490,v1,arm/neoverse-n2r0p0,core 0x00000000410fd493,v1,arm/neoverse-n2r0p3-v2,core 0x00000000410fd4f0,v1,arm/neoverse-n2r0p3-v2,core 0x00000000420f5160,v1,cavium/thunderx2,core
Signature · 1 line
2.34.1