[FEA] Structured sparsity for convolution

[jonpry]

GEMM currently supports 50% structured sparsity on ampere as in example 15_ampere_sparse_tensorop_gemm. Is there a way to use this GEMM to power a conv2d? I have not been able to find any implementations of sparse tensorop conv2d except for those built in to TensorRT. CuDNN does not support sparse weights, having this feature in cutlass would be very useful.

[mnicely]

Thanks for the request. This is something we've been thinking about but don't have the bandwidth to work on at the moment. Could you tell me more about your use cases?

[jonpry]

We have our own compiler system and don't use TensorRT. For individual 2d convolutions, cudnn/cutlass usually give us the best results. Structured sparsity is a fairly easy thing for us to add from the training side, but without some sort of inference solution we can't really make use of it.

[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.