Commit Graph

  • dd683850f4 Addressed both issues raised in review pull-request/356 Oleksandr Pavlyk 2026-05-13 11:42:40 -05:00
  • 6df6dc8d89 Enable building of NVBench on Windows (#362) main Oleksandr Pavlyk 2026-05-13 12:16:41 -05:00
  • f14055d5cc Change CMake's nvbench::main exported target to correspond to static library (#350) Oleksandr Pavlyk 2026-05-13 12:10:44 -05:00
  • e2763ca136 Refactor measurement cleanup guards for testability pull-request/358 Oleksandr Pavlyk 2026-05-13 11:09:12 -05:00
  • 0d64864ac3 Add testing/device/exception_safety.cu Oleksandr Pavlyk 2026-05-13 08:25:05 -05:00
  • ec98569a12 Renamed guard variables to reflect their purpose, apply arm-then-do to ops queueing kernels Oleksandr Pavlyk 2026-05-13 06:40:36 -05:00
  • 6798c95e23 Address remaining exception safety issue in measure_hot Oleksandr Pavlyk 2026-05-12 22:08:25 -05:00
  • 65908f767f Remove example/auto_throughput.py pull-request/347 Oleksandr Pavlyk 2026-05-04 08:32:32 -05:00
  • e07f87910a Add decorators for registering benchmarks and adding axis Oleksandr Pavlyk 2026-05-04 08:21:41 -05:00
  • a38bf890f0 Fix license header, add validation Oleksandr Pavlyk 2026-05-12 15:46:26 -05:00
  • 015f3caee2 Merge remote-tracking branch 'upstream/main' into add-bench-result Oleksandr Pavlyk 2026-05-12 15:23:39 -05:00
  • 151e463fb9 Address code rabbit review feedback Oleksandr Pavlyk 2026-05-12 15:22:57 -05:00
  • fae9dfca18 Fix for format_axis_value in json format script to handle None value Oleksandr Pavlyk 2026-05-12 15:06:31 -05:00
  • 1eba884fb3 Resolve ruff PLE0604 Oleksandr Pavlyk 2026-05-12 15:06:05 -05:00
  • 92e7c63f90 cpu_timer::start, cpu_timer::stop methods marked noexcept Oleksandr Pavlyk 2026-05-12 14:51:32 -05:00
  • 5f15bd69d7 Implement sample-count stopping criterion with parameter target-samples pull-request/341 Oleksandr Pavlyk 2026-04-30 17:33:35 -05:00
  • 9ea77bccaa Implement CLI option to control warmups for cold measurements (#339) Oleksandr Pavlyk 2026-05-12 14:30:08 -05:00
  • f885d4cb69 Add python/examples/benchmark_result_autotune.py Oleksandr Pavlyk 2026-05-12 14:12:24 -05:00
  • ebf9f9a087 Add .coderabbit.yaml following in footsteps of CCCL (#359) Oleksandr Pavlyk 2026-05-12 13:55:46 -05:00
  • 18632f9e30 Fix break due to file name change Oleksandr Pavlyk 2026-05-12 11:30:55 -05:00
  • 7dfbcad27c Create directories for output files (#360) Oleksandr Pavlyk 2026-05-12 10:26:28 -05:00
  • 00a5d1d33f Split tests in test_benchmark_result into smaller tests Oleksandr Pavlyk 2026-05-12 10:25:34 -05:00
  • 4a22923721 Add BenchmarkResult JSON results namespace and summary CLI Oleksandr Pavlyk 2026-05-12 09:58:30 -05:00
  • 231e9d7b1d Refactoring: BenchResult->BenchmarkResult Oleksandr Pavlyk 2026-05-11 17:36:12 -05:00
  • 390c919995 Use cleanup_guard in measure_cold_base::kernel_launch_timer Oleksandr Pavlyk 2026-05-11 07:01:36 -05:00
  • 74de74c058 Make sure that throwing methods call noexcept ones instead of duplicating functionality Oleksandr Pavlyk 2026-05-09 15:38:56 -05:00
  • 651b6933e4 Improve exception safety of measure_hot Oleksandr Pavlyk 2026-05-08 18:28:54 -05:00
  • abb4e27add Improve exception safely in kernel_launch_timer Oleksandr Pavlyk 2026-05-08 18:25:45 -05:00
  • c227d9668a blocking_kernel.unblock_noexcept() noexcept method added Oleksandr Pavlyk 2026-05-08 18:22:51 -05:00
  • de8d0b7368 Implement mapping interface for BenchResult Oleksandr Pavlyk 2026-05-08 16:04:42 -05:00
  • 2604547eeb Improve Python BenchResult parsing and container APIs Oleksandr Pavlyk 2026-05-08 15:51:09 -05:00
  • e26fe6bda2 Initial implementation of cuda.bench.BenchResult class Oleksandr Pavlyk 2026-05-06 11:31:31 -05:00
  • dd1ffc94e8 Add Windows support pull-request/354 Marco Franzreb Salgado 2026-05-07 00:31:36 -07:00
  • d13a0fde32 Correct cuda cccl examples per change in api (#353) Oleksandr Pavlyk 2026-05-06 13:30:44 -05:00
  • 8d1b316765 Require at least 5 samples to begin estimating noise level pull-request/348 Oleksandr Pavlyk 2026-05-05 07:44:25 -05:00
  • e53a1a2654 Use median and IR/relative as cmp_time/ref_time and cmp_noise/ref_noise Oleksandr Pavlyk 2026-05-04 16:11:22 -05:00
  • ea592b6444 Tweaks for nvbench_compare Oleksandr Pavlyk 2026-05-04 15:58:42 -05:00
  • e292bb4eec Add statistics::compute_percentiles, use it in summaries of measure_cold Oleksandr Pavlyk 2026-05-01 15:23:26 -05:00
  • e9daaba0f9 Implement sample-count stopping criterion with parameter target-samples Oleksandr Pavlyk 2026-04-30 17:33:35 -05:00
  • bf0d2a807d Ensure that measure_cold::run_warmup instantiates blocking kernel Oleksandr Pavlyk 2026-04-30 13:32:02 -05:00
  • 81e27660b8 Implement warmup-runs count, supported as CLI Oleksandr Pavlyk 2026-04-30 12:26:31 -05:00
  • f392725015 Correct Python API signature of State.get_axis_values_as_strings (#346) Oleksandr Pavlyk 2026-05-04 08:40:29 -05:00
  • a81c2af2fc Remove defaults channel pull-request/323 Oleksandr Pavlyk 2026-03-24 16:40:14 -05:00
  • 02986519dd Add Github Actions workflow to build documentation Oleksandr Pavlyk 2026-03-24 15:43:40 -05:00
  • 2decce303d Add scaffolding to build C++/Python docs Oleksandr Pavlyk 2026-03-10 16:12:55 -05:00
  • a3364ca5c7 Port changes to the package from #323 (#337) Oleksandr Pavlyk 2026-04-22 08:28:15 -05:00
  • b0a46f44c2 Modularize color handling (#336) Oleksandr Pavlyk 2026-04-14 08:09:44 -05:00
  • 8d23e3e73c [pre-commit.ci] pre-commit autoupdate (#333) pre-commit-ci[bot] 2026-04-13 16:24:55 +00:00
  • e62c5b6f79 Correct description/hint entries for summaries with name "Noise" (#335) Oleksandr Pavlyk 2026-04-13 11:13:37 -05:00
  • 373970323f Merge pull request #331 from oleksandr-pavlyk/update-python-examples Nader Al Awar 2026-04-02 15:20:24 -04:00
  • 39730efbc3 Update requirements to reflect packages used by examples Oleksandr Pavlyk 2026-04-02 10:37:17 -05:00
  • 9f75642387 Add patch to cutlass.base_dsl.dsl.BaseDSL to work-around a bug Oleksandr Pavlyk 2026-04-02 10:29:31 -05:00
  • 488173a242 Add --no-color flag to nvbench_compare.py which can be used for github issues and PRs python-0.2.1 python-0.2.1-release Nader Al Awar 2026-04-01 18:27:54 -04:00
  • 7a68e53df0 Rename flag from markdown to no-color Nader Al Awar 2026-04-01 17:01:29 -05:00
  • 7e5e784855 Add --markdown flag to nvbench_compare.py which can be use for github issues/prs Nader Al Awar 2026-04-01 14:53:13 -05:00
  • 93bc59d05c Renamed CUTLASS example to reflect that it uses CuteDSL Oleksandr Pavlyk 2026-04-01 08:24:29 -05:00
  • e4cfddeb87 Rewrote cutlass_gemm example to use CuteDSL Oleksandr Pavlyk 2026-04-01 08:23:41 -05:00
  • 3f284b4004 Renamed cccl_* examples Oleksandr Pavlyk 2026-04-01 08:20:20 -05:00
  • 5bdb30f4b6 Update to cccl_parallel_segmented_reduce example per changes in API Oleksandr Pavlyk 2026-04-01 08:18:15 -05:00
  • d8739fc208 Update to cccl_cooperative_block_reduce example Oleksandr Pavlyk 2026-04-01 08:17:52 -05:00
  • 974eb5ee0f Replace use of cupy.cuda.ExternalStream with cupy.cuda.Stream.from_external Oleksandr Pavlyk 2026-04-01 08:17:12 -05:00
  • 7c60edcc0a cuda.core.experimental -> cuda.core Oleksandr Pavlyk 2026-04-01 08:16:04 -05:00
  • 836a6c12f4 Merge pull request #326 from oleksandr-pavlyk/fix-sfinae-incomplete Oleksandr Pavlyk 2026-03-24 16:02:28 -05:00
  • 317dc6824e Mark NVBench headers as SYSTEM for consuming targets + FIX (#330) Oleksandr Pavlyk 2026-03-23 14:10:41 -05:00
  • 9a91b9ef0c Reworked cupti_profiler to use Host + Range Profiler APIs end-to-end (#327) Oleksandr Pavlyk 2026-03-23 10:51:16 -05:00
  • 1d823c6975 Merge pull request #328 from oleksandr-pavlyk/set-type-axes-names-in-auto-throughput-example Oleksandr Pavlyk 2026-03-20 18:44:03 -05:00
  • 56cdaed0af Merge pull request #299 from NVIDIA/pre-commit-ci-update-config Oleksandr Pavlyk 2026-03-20 16:15:20 -05:00
  • a6e570083d Merge pull request #329 from oleksandr-pavlyk/fix-fmt-target-name-in-tests Oleksandr Pavlyk 2026-03-20 08:49:05 -05:00
  • 4c278b08b3 Link against fmt::fmt target, not fmt. Consistent with nvbench/CMakeLists.txt Oleksandr Pavlyk 2026-03-19 14:53:06 -05:00
  • 49636c70b3 Set type-axes name to ItemsPerThread to replace auto-generated T Oleksandr Pavlyk 2026-03-19 14:35:46 -05:00
  • 728212f9f1 Merge pull request #315 from bernhardmgruber/plot_diff_script Bernhard Manfred Gruber 2026-02-28 01:38:27 +01:00
  • 4164909c52 Feedback Bernhard Manfred Gruber 2026-02-28 01:19:18 +01:00
  • 5387d2005b Merge pull request #322 from oleksandr-pavlyk/feature/save-frequencies Oleksandr Pavlyk 2026-02-27 13:30:11 -06:00
  • c9705de4a4 Reserve enough space clock-rates for min samples, if specified Oleksandr Pavlyk 2026-02-27 12:49:35 -06:00
  • 0abc8ec82b Extend nvbench_compare.py with --plot, axis/benchmark filtering, and dark mode Bernhard Manfred Gruber 2026-02-05 12:01:38 +01:00
  • ba7150e447 Merge pull request #314 from bernhardmgruber/plot_script Oleksandr Pavlyk 2026-02-26 12:59:16 -06:00
  • 800f640c20 Apply reviewer feedback Bernhard Manfred Gruber 2026-02-26 19:23:51 +01:00
  • 998ab125ce Don't override m_check_throttling if throttling threshold is non-positive Oleksandr Pavlyk 2026-02-20 16:34:53 -06:00
  • 731e0c2c30 Swapped data members m_sm_clock_rates and m_sm_clock_rate_accumulator Oleksandr Pavlyk 2026-02-19 15:33:57 -06:00
  • 4da9f431c0 Templatize write_out_values for different storage formats Oleksandr Pavlyk 2026-02-19 15:32:00 -06:00
  • 988420b5b1 Use write_out_values utility to save frequencies Oleksandr Pavlyk 2026-02-13 10:19:06 -06:00
  • 40b2f4ece2 Better place to stop freq timer? Georgy Evtushenko 2025-05-03 19:54:39 -07:00
  • a487a38895 Dump frequencies Georgy Evtushenko 2025-04-22 14:37:16 -07:00
  • d3a0bec4a8 Feedback from review Bernhard Manfred Gruber 2026-02-05 14:13:16 +01:00
  • 28ed32bb47 Implement dark mode using style sheets Bernhard Manfred Gruber 2026-02-05 14:00:33 +01:00
  • ec9759037d I have no idea what I am doing Bernhard Manfred Gruber 2026-02-05 11:15:27 +01:00
  • ccde9fc4d4 More Bernhard Manfred Gruber 2026-02-05 10:56:36 +01:00
  • 0be190b407 Add a script to plot benchmark results Bernhard Manfred Gruber 2026-02-05 10:36:52 +01:00
  • f77d001206 Fix pull-request/263 Bernhard Manfred Gruber 2025-08-18 16:52:54 +02:00
  • 8541265963 Also compare batch measurements in nvbench_compare.py Bernhard Manfred Gruber 2025-08-14 19:16:59 +02:00
  • dc59f98ecd Remove cupti from cuda-bench dependencies (#311) python-0.2.0 python-0.2.0-release Nader Al Awar 2026-02-03 14:16:26 -06:00
  • 90ad8bcbc7 Merge pull request #296 from bernhardmgruber/compare_sub_results Bernhard Manfred Gruber 2026-02-03 20:02:34 +01:00
  • c6ef87575c Allow partial comparison in nvbench_compare.py Bernhard Manfred Gruber 2025-12-10 11:06:55 +01:00
  • d75fc74162 Merge branch 'main' into remove-cupti-python Nader Al Awar 2026-02-03 08:58:41 -06:00
  • 867d5d4276 Merge pull request #294 from oleksandr-pavlyk/add-docstrings Oleksandr Pavlyk 2026-02-03 08:51:55 -06:00
  • 8a128ed7d9 Merge pull request #309 from oleksandr-pavlyk/support-skipping-batched-runs Oleksandr Pavlyk 2026-02-02 17:57:45 -06:00
  • 4fa4296810 Remove cuda.pathfinder function Nader Al Awar 2026-02-02 16:43:45 -06:00
  • f2d5730104 Disable CUPTI in cmake file Nader Al Awar 2026-02-02 16:03:15 -06:00
  • 6df5fc8c67 Remove cupti from cuda-bench dependencies Nader Al Awar 2026-02-02 15:37:13 -06:00
  • a33a454a2d Make skip_hot_measurement method const Oleksandr Pavlyk 2026-02-02 14:42:07 -06:00