]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: e948a7d)
Update CUDA graph on scale change plus clear nodes/params (#9550)
authoragray3 <redacted>
Sat, 21 Sep 2024 00:41:07 +0000 (01:41 +0100)
committerGitHub <redacted>
Sat, 21 Sep 2024 00:41:07 +0000 (02:41 +0200)
commit41f477879fd5ccc31211634292e8e293ec700e85
tree52ba924044f40647314d0a9f6d23957fd5ed8148tree
parente948a7da7af7f2dbfdcd695b3ba903ab12575f78commit | diff
Update CUDA graph on scale change plus clear nodes/params  (#9550)

* Avoid using saved CUDA graph if scale changes and reset nodes/params on update

Fixes https://github.com/ggerganov/llama.cpp/issues/9451

* clear before resize
ggml/src/ggml-cuda.cu diff | blob | history
ggml/src/ggml-cuda/common.cuh diff | blob | history
Packaging of ggml-org/llama.cpp