Closed vinx13 closed 1 year ago
This PR syncs cutlass generator with upstream and enable fp32 SIMT kernels for SM75 (SM80 by default uses TF32 for FP32)
cc @masahi @yelite
This PR syncs cutlass generator with upstream and enable fp32 SIMT kernels for SM75 (SM80 by default uses TF32 for FP32)
cc @masahi @yelite