DeepSeek/FlashMLA
1
0
Fork 0
mirror of https://github.com/deepseek-ai/FlashMLA synced 2025-06-26 18:15:54 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

update gmem

Browse Source
This commit is contained in:
chenhongmin.will
2025-02-25 09:40:56 +08:00
parent d833dbd711
commit b67a18f850
5 changed files with 29 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
csrc/flash_mla.h
View File

@@ -47,7 +47,7 @@ static constexpr int TileSchedulerMetaDataSize = 8;
////////////////////////////////////////////////////////////////////////////////////////////////////
template<typename T, int Headdim>
template<typename T, typename To, int Headdim>
void run_mha_fwd_splitkv_mla(Flash_fwd_mla_params &params, cudaStream_t stream);
struct Mla_metadata_params {