CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-13 10:57:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

metal : add support for non-padded FA KV (#16148)

Browse Source 
* metal : pad K, V and Mask when needed

* cont : simplify

* cuda : add TODO about KV padding requirement

* metal : add comments

* metal : remove mask padding requirement
This commit is contained in:
Georgi Gerganov
2025-10-07 08:23:30 +03:00
committed by GitHub
parent 1d6092fc72
commit 0a319bb75e
9 changed files with 460 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
ggml/src/ggml-cuda/fattn.cu
View File

@@ -208,6 +208,12 @@ static best_fattn_kernel ggml_cuda_get_best_fattn_kernel(const int device, const
const int cc = ggml_cuda_info().devices[device].cc;
// TODO: temporary until support is extended
// https://github.com/ggml-org/llama.cpp/pull/16148#issuecomment-3343525206
if (K->ne[1] % FATTN_KQ_STRIDE != 0) {
return BEST_FATTN_KERNEL_NONE;
}
switch (K->ne[0]) {
case 64:
case 128: