mirror of
https://github.com/deepseek-ai/DeepEP
synced 2025-06-26 18:28:11 +00:00
Support BF16 for low-latency kernels
This commit is contained in:
@@ -137,7 +137,7 @@ void dispatch(void* packed_recv_x, float* packed_recv_x_scales,
|
||||
const void* x, const int64_t* topk_idx,
|
||||
int* next_clean, int num_next_clean_int,
|
||||
int num_tokens, int hidden, int num_max_dispatch_tokens_per_rank,
|
||||
int num_topk, int num_experts, int rank, int num_ranks,
|
||||
int num_topk, int num_experts, int rank, int num_ranks, bool use_fp8,
|
||||
void* workspace, cudaStream_t stream, int phases);
|
||||
|
||||
void combine(void* combined_x,
|
||||
|
||||
Reference in New Issue
Block a user