]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 4cb7e0b)
ggml-cpu: add always_inline to tinyBLAS_PPC accumulator saves (#20791)
authorshalinib-ibm <redacted>
Fri, 20 Mar 2026 23:11:45 +0000 (04:41 +0530)
committerGitHub <redacted>
Fri, 20 Mar 2026 23:11:45 +0000 (07:11 +0800)
commite6ec21e62f45a247d3a3d2cb8f164141853b1a9b
tree8300b1a997be7a1b03b180c18a065c6d3367529atree
parent4cb7e0bd61e7e1101e8ab10db5dee70c5717a386commit | diff
ggml-cpu: add always_inline to tinyBLAS_PPC accumulator saves (#20791)

Explicitly mark save_acc and add_save_Acc with always_inline
in tinyBLAS_PPC. This ensures the compiler keeps MMA accumulator
disassembly within kernel's register context, preventing un-necessary
stask spills.

Signed-off-by: Shalini Salomi Bodapati <redacted>
ggml/src/ggml-cpu/llamafile/sgemm.cpp diff | blob | history
Packaging of ggml-org/llama.cpp