cutlass [QST] Any theory about the "layout algebra"?

What is your question?

Hi there, thank you for the work on CUTLASS3.0/CuTe. The "layout algebra" in 3.0 is much more elegant and easier to use than iterators.

I guess there should be a (formal) theory supporting the "layout algebra". Otherwise, how do you known a complex combination of basic operators will yield what you want?

Also, I noticed that there were many issues asking about the calculation of layouts. You kindly answered them case by case. But I think such questions will be endless. as users do not know the underlying computing principles.

In fact, I have been reading the source code of CuTe recently, hoping to reverse engineer the underlying theory. But I found it difficult.

So, if there is a (formal) theory, can you provide some information about it?

Thank you!

Oct 08 '23 02:10 eatingtomatoes

Yes, publications are pending. We'll have a lot more information to release soon.

Oct 08 '23 05:10 ccecka

Great! I'm looking forward to that!

Oct 08 '23 06:10 eatingtomatoes

Yes, publications are pending. We'll have a lot more information to release soon.

Thanks for sharing, is the “layout algebra” related to “tensor algebra"?

Oct 08 '23 13:10 YangWang92

Only in that it's application is for tensor algebra.

Oct 19 '23 19:10 ccecka

By the way, I'm curious about the acronym CuTe. Does it stand for CUDA Tensor Expression? Thanks!

Oct 30 '23 05:10 YangWang92

Hi @ccecka! A month has passed, but the release does not come. Do you have a concrete schedule?

I am struggling with these code every day. Save me out, Please！

Nov 02 '23 01:11 eatingtomatoes

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

Dec 02 '23 01:12 github-actions[bot]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.

Mar 05 '24 16:03 github-actions[bot]