Repository navigation
Feature/tame run times - #1
Merged
Merged
Conversation
CORE-AVX512 -> COMMON-AVX512 for consistency with Windows and for better performance on KNL
During the warm-up, measure time to execute one call, check if total number of running the batch exceeds 5 seconds. If so, reduce the actual batch_size used and rescale the result, extrapolating the time for the requested batch size. This combats excessive benchmark run-times when using stock Python.
This was caused by setting DFTI_COMPLEX_STORAGE to DFTI_COMPLEX_COMPLEX
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adjusted architecture flag when compiling native C benchmarks for FFT to use
COMMON-AVX512, rather thanCORE-AVX512. See reference.Default choice of 16 runs per batch with 24 repetitions of batch runs, combined with slow performance of FFT in stock NumPy/SciPy, results is excessive run-times for benchmarks of open source stack. Therefore, lowered default 24 repetitions to 6, and added code that would dynamically lower batch size if the batch run time exceeds 5, and rescale the timing to what it would have been should the request batch size was used.
@anton-malakhov @rscohn2