Enable AMD Matrix Core intrinsic for MetaSchedule

Hi everyone, I’d like to initiate a discussion about enabling AMD Matrix Core intrinsic for MetaScheduler. As we know, MetaScheduler has supported NVIDIA Tensor Core intrinsic and ARM CPU intrinsic. Enabling AMD Matrix Core can further expand the scope of application of TVM. Is there anyone in the community working on this? I’m willing to try adding this feature.


Thanks @malixian. As far as I know, no one is currently addressing this. It would be gratifying to see volunteers contribute to this remarkable feature.

This feature is supported at [https://github.com/apache/tvm/pull/15939], hope to get some feedback.

1 Like