Johannes Gäßler da738a74f5 CUDA: reduce MMQ stream-k overhead (llama/22298) ...

* CUDA: reduce MMQ stream-k overhead

* use 32 bit integers for kbc

2026-04-30 11:29:19 +03:00

..

cmake

cmake : remove unused file (ggml/1419)

2026-02-08 09:29:10 +02:00

include

CUDA: manage NCCL communicators in context (llama/21891)

2026-04-30 11:29:09 +03:00

src

CUDA: reduce MMQ stream-k overhead (llama/22298)

2026-04-30 11:29:19 +03:00

.gitignore

…

CMakeLists.txt

HIP: flip GGML_HIP_GRAPHS to default on (llama/22254)

2026-04-30 11:29:16 +03:00