Graf commitů

752 Commity

Autor SHA1 Zpráva Datum
vedithal-amd bd7a1de879 Remove rocprofv1/v2 in favour of rocprofiler-sdk (#673)
* Set default rocprof interface as rocprofiler-sdk

* Remove rocrprofv1 and rocprofv2 interfaces

* Remove deprecation notice for rocprof v1/v2/v3 interfaces
  * Make rocprofiler-sdk the default interface and make rocprofv3 interface opt-in using ROCPROF=rocprofv3

* Add deprecation notice for rocprofv3
2025-09-24 10:37:01 -04:00
vedithal-amd 7df02745eb [rocprofiler-compute] Update gfx12 counter definitions (#1003)
* Update gfx12 counter definitions

* Add counter defintions for Navi 4

* Made changes per https://github.com/ROCm/rocm-systems/pull/238 and
  doubled checked register specification

* bugfix
2025-09-24 10:32:21 -04:00
vedithal-amd f5505b5989 Use ROCM_PATH for sdk library path (#1097) 2025-09-24 10:31:20 -04:00
jamessiddeley-amd 05315c5bb2 fixed function argument dir in apply_filters (#1100) 2025-09-24 09:50:58 -04:00
vedithal-amd 4962f237c2 Fix workload path (#1096) 2025-09-23 17:47:31 -04:00
xuchen-amd c3054c00b1 [rocprofiler-compute] Type annotation patch for analysis_db.py (#981) 2025-09-23 17:05:37 -04:00
systems-assistant[bot] 872f0aed0c Live attach/detach and its unit tests (#53) 2025-09-23 13:17:08 -04:00
xuchen-amd 68cd123b0f [rocprofiler-compute][TUI] improve for cross-platform uses (#1007) 2025-09-23 10:59:29 -04:00
cfallows-amd 9819e1cbfc Refactor roofline binary detection (#933)
* Simplify the roofline binary pickup process by determining which base distribution the system OS is based off of, and select the correct binary.
* Add more OS distribution support to roofline by modifying the detection parameters and adding an AZL binary
* Update changelog to include roofline support additions

---------

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
2025-09-22 12:04:20 -04:00
vedithal-amd fa31650298 [rocprofiler-compute] make pmc files deterministic (#1066)
* make pmc files deterministic
2025-09-22 10:48:35 -04:00
ywang103-amd 775ac73d25 change interval for host_trap in unit test to adapt to single kernel (#1064) 2025-09-19 17:21:02 -04:00
jamessiddeley-amd 18b4b84a9f update TCC_EA0_RDREQ_sum def (#1039) 2025-09-19 11:21:45 -04:00
xuchen-amd d41e115916 [rocprofiler-compute][webui] bug fix (#1047) 2025-09-18 10:48:01 -04:00
xuchen-amd 288bca17ea [rocprofiler-compute] refactor bug fixes (#994) 2025-09-16 14:20:33 -04:00
xuchen-amd a442766d26 [rocprofiler-compute] improve profile options (#999) 2025-09-15 18:21:45 -04:00
xuchen-amd eb46160a8f update proj toml (#974) 2025-09-12 16:24:44 -04:00
xuchen-amd 7ed6000e32 [rocprofiler-compute] Refactor to add type annotation and misc (#787) 2025-09-12 13:53:24 -04:00
jamessiddeley-amd 62843ed900 [rocprof-compute] Wrap negative values in L2 - Fabric interface detailed metrics (#833)
* update gfx942 soc_config 1700

* add MAX wrapper for Write and Atomic (32B)

* removed trailing whitespace and EA* fix

* added to CHANGELOG.md

* edited changelog
2025-09-11 14:51:58 -04:00
vedithal-amd 85a557673d Handle mutually exclusive report section filters (#710)
* Make --roof-only, --block and --set mutually exclusive from each other

* Update help output and documentation
  * Add sanitize function for checking profiler options

* Update filter blocks arguments when --set or --roof-only is provided

* Update filter_blocks in profiling_config.yaml based on --set option
  * Log Filtered Sections instead of Report Sections and Set Selection

* Move soc class function calls from rocprof compute base class to profiler base class

* Fix bug in panel level filtering using --filter-block option

* Remove roofline specific pmc files

* Move microbenchmark entry point from gfx specific soc class to base soc class

* Run microbenchmarks only if block 4 is selected or roof only is selected; skip for mi100
2025-09-09 17:48:20 -04:00
vedithal-amd 4238ddb185 Add unsupported counters for rocprofv1 (#889) 2025-09-09 09:42:08 -04:00
ggottipa-amd 5288d47275 Including copyright headers. (#879) 2025-09-09 11:02:47 +05:30
abchoudh-amd 682ae2d014 Streamline --list-metrics command line option in rocprof-compute (#310)
* Remove L2 channels from --list-metrics

--list-metrics moved to general options

List metrics for the current architecture

Filter blocks for metrics

Removed test for --list-metrics in profile mode

Test the options don't throw error

Fixed --config-dir error

Test stdout for command line options

Provide path list for loading panel configs

Show L2 Cache (per) channel metrics

Changed command line option names

Can show two levels only

Removed filtering blocks

Moved blocks to original position

Removed filter block tests

Removed filtering

Formaating fix

Readability enhancement

Test formatting

Filter L2 channels without sysinfo

Show avilable metrics for current arch

Intermediate commit

Fixed tests

Added argument sanitization

Added list_metrics to ctest

merge iconflict resolution

Updated test marker

Updated changelog

Fixed formatting

* Updated docs
2025-09-08 20:21:46 +05:30
ywang103-amd c9b1ad72a5 scientific notion of memchart(CLI, TUI and GUI) (#764) 2025-09-08 01:30:30 -04:00
cfallows-amd 2e750763e7 Fix file naming for roofline PDF outputs when using --kernel (#803)
Move check for kernel filtering to add to roofline pdfs- was originally only labelling pdfs with the filtered kernel names from --kernel when --kernel-names was called, we want it at all times when kernels are filtered. 

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
2025-09-06 02:34:51 -04:00
vedithal-amd 77ed80f457 Fix Performance (GFLOPs) metric (#843) 2025-09-04 14:30:22 -04:00
vedithal-amd 181bdf9ca1 [rocprofiler-compute] Fix MI100 tests (#832)
* Fix MI100 tests

* Handle missing roofline in db_analysis.py
2025-09-03 11:09:12 -04:00
cfallows-amd c68ba44e72 Add single kernel filtering to roofline plots (#757)
* Add single kernel filtering for roofline
* Add --kernel to documentation
* Add kernel labels to roofline pdfs

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>

* Add test cases

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>

* Add autodetect for mode (profile or analyze) during roof validate and filter
Prevent --kernel from affecting roofline in gui mode- although this may be broken in develop branch anyways

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>

* Add note about roof-only usage checking for existing profiling files in the dir. If roof-only is not provided, rocprof-compute currently assumes it has to profile in full regardless. Will look into this another day.

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>

* Update CHANGELOG.md

Add line in resolved issues section to highlight that kernel filtering is now working for roofline plots

* Apply changes suggested by docs team

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>

* Update projects/rocprofiler-compute/CHANGELOG.md

Co-authored-by: Pratik Basyal <pratik.basyal@amd.com>

---------

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
Co-authored-by: Pratik Basyal <pratik.basyal@amd.com>
2025-08-27 13:41:07 -04:00
jamessiddeley-amd f0955f5a83 [rocprof-compute] added f4f6 description and VALU FLOPS split for empirical peaks (#739)
* added f4f6 description and VALU FLOPS split

* changed peak ammolite vars to local

* reverted to dict peak initialization

* ruff check format

* updated VALU descriptions

* updated VALU descriptions

* Update parser.py

* Update parser.py

Added gracefull NameError handling
Moved globals() update to init_metric_evaluation with ammolite__ vars and raw pmc_df

* update formatting
2025-08-27 12:46:41 -04:00
vedithal-amd 323d06c79c [rocprofiler-compute] Add database output format to analyze mode (#748)
Analysis data dump

* Add `--output-format` and `--output-name` option to analyze mode

* Remove `--output` and `-save-dfs` option to analyze mode

* Add documentation on `rocpd` output format and analysis database file

* Create sqlite3 database using object relation mapping (ORM) provided
  by sqlalchemy library

* Fix metrics config to remove metrics marked as `null`, fix `Unit` header, add
  missing `title`

* Add test cases to ensure analysis data dump work
2025-08-26 14:15:05 -04:00
xuchen-amd 5c8b34ddf5 [rocprofiler-compute][TUI] Add interactive metric description (#718) 2025-08-25 15:53:55 -04:00
vedithal-amd 9a02dae75f [rocprofiler-compute] [Bugfix] Fix analysis not working with rocpd (#704)
* fix rocpd roofline

* Improve rocpd test by using dynamic workload folder

* bugfix

* fix ruff format
2025-08-25 11:46:55 -04:00
Taylor Ding b5c8c8bcb1 Eval metrics performance optimizations (#435)
Post-analysis eval metrics performance optimizations.
2025-08-22 16:35:48 -04:00
jamessiddeley-amd 5deeea71df [rocprof-compute] Update Formatting (#671)
* updated rocprof-compute formatting

* fixed ammolite peak variables in parser.py

* format parser.py

* update formatting rocprof_compute_base
2025-08-22 12:22:17 -04:00
xuchen-amd ed877433f3 [rocprofiler-compute][TUI] Updates and refactor. (#703) 2025-08-22 11:02:32 -04:00
xuchen-amd 6bf5648c6d [rocprofiler-compute] Handle amd-smi rocm==6.4.2 json output issue. (#667) 2025-08-21 10:06:04 -04:00
xuchen-amd 0bf66a519c [rocprofiler-compute][TUI] Restructure Performance Metrics (#232) 2025-08-20 17:00:54 -04:00
vedithal-amd ed53090c25 Join SQC_* csv files as well (#669) 2025-08-20 15:33:48 -04:00
jamessiddeley-amd 5840940caa [rocprof-compute] Generalize Roofline (#325)
* per kernel analysis Roofline

* added per-kernel eval_metric calculation with display

* fixed typo

* updated tty.py show_all()

* formatting

* fixed ctest failures and updated equations

* formatting

* updated metric descriptoins

* review tweaks

* update docs

* added roofline gui analysis

* updated GUI docs

* updated print statement

* comment tweaks and ran ruff formatting
2025-08-20 09:58:08 -04:00
xuchen-amd 483315fd0a Fix amd_smi lookup logic to account for amd-smi>=26.0.0 (#320) 2025-08-13 10:21:45 -04:00
vedithal-amd 63f31859e4 Fix instruction cache metrics (#276)
* Fix IL1 Latency metric

* Add SQC_ICACHE_INFLIGHT_LEVEL counter and corresponding accumulation
  counter to counter definitions

* Add missing coll_level

* Fix descriptions and tests
2025-08-12 14:03:11 -04:00
xuchen-amd 9a4983feac Fix fstring formatting. (#315) 2025-08-12 11:54:58 -04:00
xuchen-amd 0929e6ecc1 [rocprofiler-compute] Update amd smi lookup (#223) 2025-08-11 17:37:33 -04:00
systems-assistant[bot] 58d2a016ce Format source code to PEP8 using Ruff (#36)
* added ruff docs

* style: Run ruff and black before yapf pass

* yapf -r -i (23 fixes)

* fixed conf.py and ran ruff format .

* fixed conf.py 2

* formatted argparser.py

* formatted src/rocprof_compute_analyze

* formatted src/rocprof_compute_profile

* formatted soc_base.py

* formatted rocprof_compute_tui

* formatted gui_components

* formatted src/utils

* formatted tests/

* format extra files

* cleanup

* fix test_utils.py

* fixed typos

* Update pyproject.toml

* Update README.md

* Update test_utils.py

---------

Signed-off-by: jamessiddeley-amd <James.Siddeley@amd.com>
Co-authored-by: James Siddeley <James.Siddeley@amd.com>
Co-authored-by: systems-assistant[bot] <systems-assistant[bot]@users.noreply.github.com>
2025-08-08 15:32:30 -04:00
systems-assistant[bot] d3f9ab25eb Use own counter definition (#91)
* Use own counter definition
  * Do not depend on rocprofiler-sdk counter definition

* Add missing counter definitions for MI100, MI200, MI300, MI350 series
  * Counters added based on register specification
  * This prevents some missing metrics

* Enable SQC_DCACHE_INFLIGHT_LEVEL counter and associated metrics

* Enable TCP_TCP_LATENCY counter and associated counter for all GPUs
  except MI300

* Update TCC_EA_* counters for MI100 to TCC_EA0_*
  * Update MI100 metrics which depend on TCC_EA0_* counters

* Enable accumulation counters for MI100

* Improve rocprof list avail usage to get a better idea of supported
  counters

* Update CHANGELOG

* Move accumulation counters to counter definition

---------

Co-authored-by: Vignesh Edithal <Vignesh.Edithal@amd.com>
2025-08-08 14:39:10 -04:00
cfallows-amd d8d739f57c Re-add rhel8 roofline binaries (#122)
RHEL8 now being supported until EOS, rebuilding rhel8 bins for rocm7 (previously built for rocm7 rhel9).

Remove roofline-rhel9-rocm7, replace with new roofline-rhel8-rocm7.
Update check for roof bin.
Update any doc mentions of rhel min support version back to rhel8.

---------

Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
2025-08-07 12:24:02 -04:00
systems-assistant[bot] 89c74ac3d3 Update Unit of Bandwidth metrics to Gbps (#96)
* Add Utilization to metric name for Bandwidth related metrics whose Unit
  is Percent

* Update Unit of Bandwidth metrics to Gbps
    * Update metric Formula to use total duration as denominator instead of normalization unit.
    * Update metric Description
    * Update metric Unit

* Update CHANGELOG
2025-08-06 18:39:50 -04:00
xuchen-amd 34dd26fb07 Enable single pass counter collection (#833)
[ROCm/rocprofiler-compute commit: 6a77d241ed]
2025-08-06 10:35:05 -04:00
vedithal-amd 30c32c0bd1 Fix L2 cache bandwidth metrics for MI350 (#843)
* Fix L2 cache bandwidth metrics for MI350

* Address review comments

[ROCm/rocprofiler-compute commit: b349e406ed]
2025-08-01 14:00:46 -04:00
Fei Zheng 7a4d187f40 update license files (#849)
[ROCm/rocprofiler-compute commit: 0d6e172679]
2025-08-01 10:14:39 -06:00
xuchen-amd b6a7d72212 TUI: Kernel Selection (#769)
[ROCm/rocprofiler-compute commit: 9ddee8de2e]
2025-08-01 11:31:43 -04:00