]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: d12f781)
CUDA: revert part of the RDNA1 optimizations (#8309)
authorDaniele <redacted>
Fri, 5 Jul 2024 07:06:09 +0000 (07:06 +0000)
committerGitHub <redacted>
Fri, 5 Jul 2024 07:06:09 +0000 (09:06 +0200)
commit0a423800ffe4e5da3d83527ef3473da88cd78146
tree39b31d72638b2afbe82839af8b3663f3cb79fed9tree
parentd12f781074b92589a72a36ffabb583933f7b9dc0commit | diff
CUDA: revert part of the RDNA1 optimizations (#8309)

The change on the launch_bounds was causing a small performance drop in perplexity of 25 t/s
ggml/src/ggml-cuda/mmq.cuh diff | blob | history
Packaging of ggml-org/llama.cpp