(cross-post of https://forums.developer.nvidia.com/t/code-samples-for-gtc-s81772-don-t-leave-tensors-on-the-table-programming-and-optimizing-tensor-cores/376578, since the talk author is from the devtools team, so this might be the right forum)
https://www.nvidia.com/en-us/on-demand/session/gtc26-s81772/ was a great talk!
Is the “tensorCoreSamples.cu” file used to create example profiles available online anywhere? I couldn’t find it on github.
It would be great to see the full example code (especially for tma+wgmma pipelining) and play with profiling the kernels on my own.