]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: c4abcb2)
ggml-cpu: handle 3d tensors in repack mat_mul (#17241)
authorAlberto Cabrera Pérez <redacted>
Thu, 13 Nov 2025 20:53:00 +0000 (20:53 +0000)
committerGitHub <redacted>
Thu, 13 Nov 2025 20:53:00 +0000 (12:53 -0800)
commitbecc4816dd6e601d2e0beb7b9c7e6767c8688b12
treeb47b713018373412835013a3f8afc839e8c52bbdtree
parentc4abcb2457217198efdd67d02675f5fddb7071c2commit | diff
ggml-cpu: handle 3d tensors in repack mat_mul (#17241)

* ggml-cpu: handle 3d tensors in repack mul_mat

* Removed unnecessary branch, removed need for <algorithm>

* Fixed dst_ptr pointer in chunk + clang_format

* GGML_ASSERT to check wdata within bounds

* Accidental ggml.h inclusion

* Improved GGML_ASSERT on wdata boundaries

* Address performance regression in Qwen and llama.cpp due to chunking
ggml/src/ggml-cpu/repack.cpp diff | blob | history
Packaging of ggml-org/llama.cpp