[BUG] CuteDSL example hits IMA for large tensors due to strides in Int32

[tridao]

Tensors in Cute DSL uses strides in int32 by default. This causes IMA for large tensors. Is there a way to force strides to be int64?

Steps/Code to reproduce bug

python examples/python/CuTeDSL/ampere/elementwise_add.py --M=65384 --N=65384

RuntimeError: CUDA error: an illegal memory access was encountered

cc: @thakkarV

[thakkarV]

Yeah this is a known limitation for now. Future releases will come with support for 64b strides.

[fengxie]

@thakkarV is right. To simplify the first release, we don't support 64bit or mixed integer type in CuTe algebra. This limitation will be removed in future release. Thanks for catching. Updating documentation as well.

A side notes: python int and float are default to int32 and float32 for now for the same consideration. It's under consideration to revisit this part as well.

[github-actions[bot]]

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.

[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.

[ktaebum]

Hi, is there any update on this? I am also encountering the same issue.

[brandon-yujie-sun]

Thanks for asking. i64 stride support will be included in the upcoming 4.3 dev release very soon.