]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 9190e8e)
ggml : speed-up ggml_vec_dot_q4_1() ARM_NEON + 32-bit ARM support (#900)
authorGeorgi Gerganov <redacted>
Thu, 13 Apr 2023 15:32:36 +0000 (18:32 +0300)
committerGitHub <redacted>
Thu, 13 Apr 2023 15:32:36 +0000 (18:32 +0300)
commitd990e3fffc5b0f5448e90a16c79a4f2675100af0
treeb9380725b8510b2fdac4b4a26b031869ae71e1bctree
parent9190e8eac8bdc108c40d2d7505e9b45fa773251fcommit | diff
ggml : speed-up ggml_vec_dot_q4_1() ARM_NEON + 32-bit ARM support (#900)

* ggml : speed-up q4_1 ARM_NEON by ~5%

* ggml : implement vaddvq when missing

* ggml : implement vminvq and vmaxvq when missing

* ggml : implement vzip when missing

* ggml : fix comment

* ggml : try to use correct ifdef
ggml.c diff | blob | history
Packaging of ggml-org/llama.cpp