]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 8c146a8)
vulkan: use flops instead of weight tensor size for submission heuristic (#25005)
authorRuben Ortlam <redacted>
Mon, 29 Jun 2026 13:24:44 +0000 (15:24 +0200)
committerGitHub <redacted>
Mon, 29 Jun 2026 13:24:44 +0000 (15:24 +0200)
commit25a1d63f4346b472e508c6dbd9ab2ed1d81ace2e
treef246b7754ea2df3b1f1ccdf723a9855b18565529tree
parent8c146a8366304c871efc26057cc90370ccf58dadcommit | diff
vulkan: use flops instead of weight tensor size for submission heuristic (#25005)

* vulkan: extract flops calculation into function

* use flops instead of matmul src0 tensor size for submission threshold

* use unsigned ints
ggml/src/ggml-vulkan/ggml-vulkan.cpp diff | blob | history
Packaging of ggml-org/llama.cpp