1
0

flash_fwd_hdim64_bf16_gqa2_sm90.cu 411 B

123456789
  1. // Copyright (c) 2024, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
  2. // Splitting the different head dimensions to different files to speed up compilation.
  3. #include "flash_fwd_launch_template.h"
  4. template<>
  5. void run_mha_fwd_gqa_<cutlass::bfloat16_t, 64, 2>(Flash_fwd_params &params, cudaStream_t stream) {
  6. run_mha_fwd_hdim64_gqa<cutlass::bfloat16_t, 2>(params, stream);
  7. }