/benchmarking-public

A public mirror of our benchmarking runner repository

Primary LanguagePython

Faster CPython Benchmark Infrastructure

▶️ START A BENCHMARK RUN

For documentation about how this works, see the developer docs.

Running benchmarks from the GitHub web UI

Visit the benchmark action and click the "Run Workflow" button.

The available parameters are:

  • fork: The fork of CPython to benchmark. If benchmarking a pull request, this would normally be your GitHub username.
  • ref: The branch, tag or commit SHA to benchmark. If a SHA, it must be the full SHA, since finding it by a prefix is not supported.
  • machine: The machine to run on. One of linux-amd64 (default), windows-amd64, darwin-arm64 or all.
  • benchmark_base: If checked, the base of the selected branch will also be benchmarked. The base is determined by running git merge-base upstream/main $ref.
  • publish: If checked, the results will be published in the public ideas repo upon successful completion.

To watch the progress of the benchmark, select it from the benchmark action page. It may be canceled from there as well. To show only your benchmark workflows, select your GitHub ID from the "Actor" dropdown.

When the benchmarking is complete, the results are published to this repository and will appear in the master table. Each set of benchmarks will have:

  • The raw .json results from pyperformance.
  • Comparisons against important reference releases, as well as the merge base of the branch if benchmark_base was selected. These include
    • A markdown table produced by pyperf compare_to.
    • A set of "violin" plots showing the distribution of results for each benchmark.

The most convenient way to get results locally is to clone this repo and git pull from it.

Running benchmarks from the GitHub CLI

To automate benchmarking runs, it may be more convenient to use the GitHub CLI. Once you have gh installed and configured, you can run benchmarks by cloning this repository and then from inside it:

$ gh workflow run benchmark.yml -f fork=me -f ref=my_branch

Any of the parameters described above are available at the commandline using the -f key=value syntax.

Results

The following is only a summary of certain key revisions. There is also a complete list of results.

linux x86_64

date fork ref version hash vs. 3.10.4: vs. 3.11.0: vs. base:
2022-03-23 python main 3.10.4 9d38120 1.26x slower *
2022-12-06 python 1dd9be6584 3.10.9 1dd9be6 1.01x slower * 1.26x slower
2021-10-05 python 7c12e4835e 3.11.0a1 7c12e48 1.12x faster * 1.12x slower
2021-11-05 python e2b4e4bab9 3.11.0a2 e2b4e4b 1.15x faster * 1.09x slower
2021-12-08 python main 3.11.0a3 2e91dba 1.19x faster 1.06x slower *
2022-01-13 python main 3.11.0a4 9471106 1.21x faster 1.04x slower *
2022-02-03 python main 3.11.0a5 c4e4b91 1.21x faster 1.04x slower *
2022-03-07 python main 3.11.0a6 3ddfa55 1.19x faster 1.06x slower *
2022-04-05 python main 3.11.0a7 2e49bd0 1.23x faster 1.02x slower *
2022-05-06 python main 3.11.0b1 8d32a5c 1.28x faster 1.01x faster *
2022-05-30 python main 3.11.0b2 72f00f4 1.28x faster 1.02x faster *
2022-06-01 python main 3.11.0b3 eb0004c 1.29x faster 1.02x faster *
2022-07-11 python 5a7e1e0a92 3.11.0b4 5a7e1e0 1.27x faster * 1.01x faster
2022-07-25 python 0771d71eea 3.11.0b5 0771d71 1.27x faster * 1.01x faster
2022-08-05 python 41cb07120b 3.11.0rc1 41cb071 1.27x faster * 1.00x faster
2022-09-11 python ed7c3ff156 3.11.0rc2 ed7c3ff 1.26x faster * 1.00x slower
2022-10-24 python v3.11.0 3.11.0 deaf509 1.26x faster *
2022-12-06 python a7a450f84a 3.11.1 a7a450f 1.25x faster * 1.00x slower
2022-10-22 python main 3.12.0a1+ f58631b 1.30x faster * 1.03x faster *
2022-12-05 python e3a3863cb9 3.12.0a2+ e3a3863 1.31x faster * 1.03x faster
2022-12-21 python a7715ccfba 3.12.0a3+ a7715cc 1.32x faster * 1.03x faster

darwin arm64

date fork ref version hash vs. 3.10.4: vs. 3.11.0: vs. base:
2022-03-23 python v3.10.4 3.10.4 9d38120 1.22x slower
2022-12-06 python 1dd9be6584 3.10.9 1dd9be6 1.00x slower 1.22x slower
2021-11-05 python e2b4e4bab9 3.11.0a2 e2b4e4b 1.15x faster 1.05x slower
2021-12-08 python 2e91dba437 3.11.0a3 2e91dba 1.15x faster 1.06x slower
2022-01-13 python 9471106fd5 3.11.0a4 9471106 1.17x faster 1.04x slower
2022-02-03 python c4e4b91557 3.11.0a5 c4e4b91 1.09x faster 1.12x slower
2022-03-07 python 3ddfa55df4 3.11.0a6 3ddfa55 1.17x faster 1.04x slower
2022-04-05 python 2e49bd06c5 3.11.0a7 2e49bd0 1.22x faster 1.01x slower
2022-05-06 python 8d32a5c8c4 3.11.0b1 8d32a5c 1.21x faster 1.01x slower
2022-05-30 python 72f00f420a 3.11.0b2 72f00f4 1.21x faster 1.01x slower
2022-06-01 python eb0004c271 3.11.0b3 eb0004c 1.21x faster 1.01x slower
2022-07-11 python 5a7e1e0a92 3.11.0b4 5a7e1e0 1.23x faster 1.01x faster
2022-07-25 python 0771d71eea 3.11.0b5 0771d71 1.23x faster 1.01x faster
2022-08-05 python 41cb07120b 3.11.0rc1 41cb071 1.23x faster 1.01x faster
2022-09-11 python ed7c3ff156 3.11.0rc2 ed7c3ff 1.22x faster 1.00x faster
2022-10-24 python deaf509e8f 3.11.0 deaf509 1.22x faster
2022-12-06 python a7a450f84a 3.11.1 a7a450f 1.22x faster 1.00x faster
2022-10-24 python dfb5d272e6 3.12.0a0 dfb5d27 1.20x faster 1.02x slower
2022-11-13 python 367f552129 3.12.0a1+ 367f552 1.19x faster 1.03x slower
2022-12-05 python e3a3863cb9 3.12.0a2+ e3a3863 1.18x faster 1.04x slower
2022-12-16 python main 3.12.0a3+ 432117c 1.24x faster 1.02x faster

* indicates that the exact same versions of pyperformance was not used.