Searched refs:force_split_kernel (Results 1 – 1 of 1) sorted by relevance
244 void run_mha_fwd(Flash_fwd_params ¶ms, cudaStream_t stream, bool force_split_kernel=false) { in run_mha_fwd() argument247 … if (params.num_splits <= 1 && !force_split_kernel) { // If we don't set it num_splits == 0 in run_mha_fwd()