]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: e68c19b)
ggml-cuda: fix blackwell native builds (#18361)
authorAman Gupta <redacted>
Thu, 25 Dec 2025 14:12:11 +0000 (22:12 +0800)
committerGitHub <redacted>
Thu, 25 Dec 2025 14:12:11 +0000 (22:12 +0800)
commitb0fb0f0aee2c931cbc08e95aef68f615c6de90a2
treea924669824c116473836f9a3fb4a7e439d7904d5tree
parente68c19b0fdbb18d7a19217194c2795897e9c683ccommit | diff
ggml-cuda: fix blackwell native builds (#18361)

* ggml-cuda: fix blackwell native builds

Replace 12x in native architectures by 12xa

* replace for GGML_NATIVE=OFF too

* only replace for native

* remove 120f-virtual for default compilation

---------

Co-authored-by: Aman Gupta <aman>
ggml/src/ggml-cuda/CMakeLists.txt diff | blob | history
Packaging of ggml-org/llama.cpp