]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 7f251fd)
TP: fix 0-sized tensor slices, AllReduce fallback (#21808)
authorJohannes Gäßler <redacted>
Mon, 20 Apr 2026 16:09:39 +0000 (18:09 +0200)
committerGitHub <redacted>
Mon, 20 Apr 2026 16:09:39 +0000 (18:09 +0200)
commitfb19f94c715c466230c72d2a32822f8a9e113708
treea922d058137a47f5c031153ea8ac4b90433c7f8btree
parent7f251fdbce614a50141005dc70ce3787b7777a8ecommit | diff
TP: fix 0-sized tensor slices, AllReduce fallback (#21808)

* TP: fix 0-sized tensor slices, AllReduce fallback

* fix layer structure <-> GPU count aliasing

* add missing std::fill

* fix CUDA device set, max ggml ctx size
ggml/src/ggml-backend-meta.cpp diff | blob | history
ggml/src/ggml-cuda/ggml-cuda.cu diff | blob | history
src/llama-model.cpp diff | blob | history
Packaging of ggml-org/llama.cpp