]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: a69f846)
llama : change cpu_buft_list order: ACCEL -> GPU host -> CPU extra -> CPU (#12632)
authorDjip007 <redacted>
Sat, 29 Mar 2025 13:07:37 +0000 (14:07 +0100)
committerGitHub <redacted>
Sat, 29 Mar 2025 13:07:37 +0000 (14:07 +0100)
commit0bb2919335d00ff0bc79d5015da95c422de51f03
tree9f56800e2c73e1428c38e6ab06dde9604030b946tree
parenta69f8463510a70fe0c85701b8c7ede283a0d1a7dcommit | diff
llama : change cpu_buft_list order: ACCEL -> GPU host -> CPU extra -> CPU (#12632)

this allow to use GPU host when possible over CPU repack.
this have the same effect to resolve this issues (#12498) without
completely disable CPU extra buffer.

Co-authored-by: philou <redacted>
src/llama-model.cpp diff | blob | history
Packaging of ggml-org/llama.cpp