[PATCH] D101844: [MicroBenchmarks] Add initial loop vectorization benchmarks.

[Florian Hahn via Phabricator via llvm-commits]

fhahn added a comment.

In D101844#2750869 <https://reviews.llvm.org/D101844#2750869>, @Meinersbur wrote:

> In D101844#2750168 <https://reviews.llvm.org/D101844#2750168>, @fhahn wrote:
>
>> Unfortunately I cannot share any detailed absolute numbers about runtime. Also, the total runtime will depend on how long it takes for google-benchmark to determine that the results are stable.
>
> I am mainly asking because in the past we mandated that the execution time per program remains below 10 seconds to limit the execution time of the test-suite, and not give a disproportional weight to single benchmarks. Of course the execution time varies depending on the platform, like the other Google benchmarks already in the test-suite, but this shouldn't stop giving a ballpark number of a typical execution times, as we did for others (e.g. D36582 <https://reviews.llvm.org/D36582>, D43319 <https://reviews.llvm.org/D43319>, D47675 <https://reviews.llvm.org/D47675>, D49503 <https://reviews.llvm.org/D49503>)

Ah I see. I think the total runtime of the binary is probably more than 10s and closer to a low single digit number of minutes at the most. Each applies the math function to 100000 elements, which still should be fairly quick per benchmark


Repository:
  rT test-suite

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D101844/new/

https://reviews.llvm.org/D101844