happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

cont : fix comment 

Co-authored-by: Aman Gupta <amangupta052@gmail.com>
This commit is contained in:
Georgi Gerganov 2026-03-10 12:27:11 +02:00 committed by GitHub
parent 39b6f5a760
commit 79541c0a93
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
ggml/src/ggml-cuda/ggml-cuda.cu
View File

@ -4999,7 +4999,8 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
#ifdef GGML_USE_MUSA
return false;
#else
// TODO: add non-KDA chunked support. for now enable chunked support for KDA only
KDA is faster using the AR kernel even when n_tokens >= 512.
//TODO: Add chunked kernel
return op->src[0]->ne[2] == 1 || op->src[3]->ne[0] == op->src[2]->ne[0];
#endif // GGML_USE_MUSA
case GGML_OP_FLASH_ATTN_EXT: