]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: ff806a1)
Hexagon: Process M-tail rows on HMX instead of HVX (#22724)
authorTrivikram Reddy <redacted>
Tue, 5 May 2026 16:43:03 +0000 (11:43 -0500)
committerGitHub <redacted>
Tue, 5 May 2026 16:43:03 +0000 (09:43 -0700)
commitbbeb89d76c41bc250f16e4a6fefcc9b530d6e3f3
tree33664c6c6f867f7b9c426c375a6871e26e09fec7tree
parentff806a110de30ce5ef8de99a72c54ae07243bcdccommit | diff
Hexagon: Process M-tail rows on HMX instead of HVX (#22724)

* hex-mm: process m-tail rows on HMX instead of HVX

* hmx-mm: unroll and optimize padded activation loop

---------

Co-authored-by: Max Krasnyansky <redacted>
ggml/src/ggml-hexagon/htp/hmx-matmul-ops.c diff | blob | history
ggml/src/ggml-hexagon/htp/matmul-ops.c diff | blob | history
Packaging of ggml-org/llama.cpp