File size: 477 Bytes
a7165c8 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 |
// Copyright (c) 2024, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
#include "namespace_config.h"
#include "flash_fwd_launch_template.h"
namespace FLASH_NAMESPACE {
template<>
void run_mha_fwd_<cutlass::half_t, 32, true>(Flash_fwd_params ¶ms, cudaStream_t stream) {
run_mha_fwd_hdim32<cutlass::half_t, true>(params, stream);
}
} // namespace FLASH_NAMESPACE |