Field G. Van Zee
Field G. Van Zee
Hi @jakirkham. Thanks for your inquiry. We've done most of our QA on our next release (1.0), but one of our collaborators in industry stumbled upon a performance regression for...
Thanks for your understanding, @jakirkham.
> Thanks Field! 🙏 > > Appreciate the prompt and detailed update > > No worries. Understand how that goes. Was mostly trying to get a sense of roughly when...
I'm going to close this issue now that the new release is imminent. Thanks again for hanging in there with us, @jakirkham.
Hmm, don't have access to a Xeon Gold, but I can't reproduce that error on my x86_64 workstation.
@BhaskarNallani @kvaragan In case you haven't seen this yet... Looks like an issue with the AMD fork of BLIS.
@AngryLoki Thank you for taking the time to gather, visualize, and share these performance results! Don't worry; a proper `zen4` subconfiguration *will* be added to vanilla BLIS in the future....
@jeffhammond Let me know when you feel good about calling this PR done, and I'll merge.
> You mean if we choose a configure family like `arm64` or `x86_64` then different sub-families need different compile flag right? @winterland1989 Yes, but it actually goes beyond just the...
> @fgvanzee ready to look over. Still 181 files changed :) Thanks, will do. Can you remind me if the CI failures are still expected at this point?