KEMBAR78
`<charconv>`: floating benchmark by AlexGuteniev · Pull Request #5700 · microsoft/STL · GitHub
Skip to content

<charconv>: floating benchmark #5700

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 12 commits into from
Oct 17, 2025
Merged

<charconv>: floating benchmark #5700

merged 12 commits into from
Oct 17, 2025

Conversation

@AlexGuteniev
Copy link
Contributor

@AlexGuteniev AlexGuteniev commented Aug 28, 2025 •
edited
Loading

Port benchmark from #1243 (comment) comment.

Changes:

  • Don't hard-code the number of repetitions, this is now determined by the framework
  • Use DoNotOptimize instead of a dummy variable
  • Measure only one conversion rather than whole vector. Primarily because Google Benchmark still shows these bigger times in ns, so they are inconveniently big when measuring them for the whole vector
  • Underscore in benchmark names, as they are now function names
  • N -> n, RT -> Rt, avoiding macro-like names
  • Randomly generated vectors do not persist between tests. I find this easier to maintain this way, despite some runtime cost, feel free to disagree
  • Avoided .size() in a loop. Perf habit: calling vector<T>::size() in a loop may inhibit certain optimizations this doesn't work, we skip adding sometimes, we have to look how many we have generated

Example output:

2025-08-28T09:43:39+03:00
Running out\bench\benchmark-charconv_floats.exe
Run on (12 X 2496 MHz CPU s)
CPU Caches:
  L1 Data 48 KiB (x6)
  L1 Instruction 32 KiB (x6)
  L2 Unified 1280 KiB (x6)
  L3 Unified 12288 KiB (x1)
-------------------------------------------------------------------------
Benchmark                               Time             CPU   Iterations
-------------------------------------------------------------------------
STL_float_plain_shortest             36.2 ns         36.1 ns     19478261
STL_double_plain_shortest            43.8 ns         44.5 ns     17920000
STL_float_scientific_shortest        34.8 ns         34.9 ns     22400000
STL_double_scientific_shortest       43.0 ns         39.1 ns     16000000
STL_float_fixed_shortest             43.0 ns         41.4 ns     16592593
STL_double_fixed_shortest             110 ns          110 ns      6400000
STL_float_general_shortest           35.4 ns         33.5 ns     22400000
STL_double_general_shortest          43.3 ns         43.3 ns     16592593
STL_float_hex_shortest               18.6 ns         18.4 ns     37333333
STL_double_hex_shortest              22.4 ns         22.0 ns     32000000
STL_float_scientific_8               59.7 ns         55.8 ns     11200000
STL_double_scientific_16             68.4 ns         68.4 ns     11200000
STL_float_fixed_6_lossy              40.6 ns         40.1 ns     17920000
STL_double_fixed_6_lossy             92.8 ns         94.2 ns      8960000
STL_float_general_9                  71.6 ns         69.8 ns      8960000
STL_double_general_17                81.5 ns         77.4 ns      7466667
STL_float_hex_6                      18.2 ns         18.0 ns     37333333
STL_double_hex_13                    21.2 ns         20.4 ns     34461538

@AlexGuteniev @StephanTLavavej
Port benchmark from microsoft#1243 comment
5f342c3 
Co-authored-by: StephanTLavavej <stl@nuwen.net>
@AlexGuteniev AlexGuteniev requested a review from a team as a code owner August 28, 2025 06:46
@github-project-automation github-project-automation bot moved this to Initial Review in STL Code Reviews Aug 28, 2025
AlexGuteniev
AlexGuteniev commented Aug 28, 2025
View reviewed changes
@StephanTLavavej StephanTLavavej added performance Must go faster test Related to test code labels Aug 28, 2025
@StephanTLavavej StephanTLavavej self-assigned this Aug 28, 2025
@StephanTLavavej StephanTLavavej added benchmark Warp nine! Nine point two! Nine point three! and removed performance Must go faster test Related to test code labels Sep 26, 2025
StephanTLavavej
StephanTLavavej reviewed Oct 16, 2025
View reviewed changes
@StephanTLavavej
Copy link
Member

Thanks, this is awesome! 😻 I pushed some simplifications, and verified that the benchmark results weren't degraded.

@StephanTLavavej StephanTLavavej removed their assignment Oct 16, 2025
@StephanTLavavej StephanTLavavej moved this from Initial Review to Ready To Merge in STL Code Reviews Oct 16, 2025
@StephanTLavavej
Copy link
Member

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.

@StephanTLavavej StephanTLavavej moved this from Ready To Merge to Merging in STL Code Reviews Oct 16, 2025
@StephanTLavavej StephanTLavavej merged commit 5f88083 into microsoft:main Oct 17, 2025
39 checks passed
@github-project-automation github-project-automation bot moved this from Merging to Done in STL Code Reviews Oct 17, 2025
@StephanTLavavej
Copy link
Member

Thanks for modernizing this benchmark and letting me relive the glory days of <charconv>! 😻 ⏱️ 🚀

@AlexGuteniev AlexGuteniev deleted the charconv-float-benchmark branch October 17, 2025 16:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@StephanTLavavej StephanTLavavej StephanTLavavej approved these changes

Assignees

No one assigned

Labels

benchmark Warp nine! Nine point two! Nine point three!

Projects

STL Code Reviews
Archived in project

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@AlexGuteniev @StephanTLavavej