Skip to content
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.

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

Alignment Fixes for SIMD FFT #41

Merged
merged 2 commits into from
Apr 19, 2020
Merged

Conversation

bsupnik
Copy link

@bsupnik bsupnik commented Apr 19, 2020

This patch fixes unaligned memory for SIMD FFT by adding padding up to 16-bytes in sub-allocations of the various FFTs.
#40

Ben Supnik added 2 commits April 18, 2020 20:42
up to 16-byte boundaries.  This fixes crashes where the sub-buffers
for complex FFTs end up unaligned violating compiler expectations
@mborgerding mborgerding merged commit 07ccd81 into mborgerding:master Apr 19, 2020
@mborgerding
Copy link
Owner

I'm going to close this PR since your changes have been merged in a manual path.
The PR as it was a) failing to build under Travis CI and b) did not have a test that demonstrated issue #40.

I made a few changes along this line as of cf5813a

  • the existing source uses spaces, not tabs.
  • assert was used but the assert.h header was not being included anywhere.
  • IMO, assert is not a good way of error checking since some compilers bypass it entirely. It was used in KISS_FFT_ALIGN_CHECK which seemed like a debug artifact.
  • I added a very rudimentary testsse target which now gets run under Travis CI. All it checks is that the 256x256 plan does not crash. This should be expanded to do some level of correctness testing (at least forward/inverse consistency check)

Feel like tackling issue #24? It seems to lack a champion who is willing to work on it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants