]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 82bb485)
ggml : add aarch64 HWCAP fallbacks and fix fp16 variant detection (#25554)
authorKilian Hu <redacted>
Fri, 7 Aug 2026 11:07:10 +0000 (13:07 +0200)
committerGitHub <redacted>
Fri, 7 Aug 2026 11:07:10 +0000 (14:07 +0300)
commitcb26014d965036f0eacdab7387b222d28d36b9e6
tree48dc5046969a2f4b25532bcba506e7ae02b2e329tree
parent82bb48500a9a6b3374da2cf5811695369278e4d3commit | diff
ggml : add aarch64 HWCAP fallbacks and fix fp16 variant detection (#25554)

* ggml : add fallback definitions for missing aarch64 HWCAP bits

* ggml : require HWCAP_ASIMDHP for the aarch64 fp16 cpu variants

Also rename has_fp16_va to has_fp16, the field gates the whole FEAT_FP16
extension, scalar and vector half-precision arithmetic together.
ggml/src/ggml-cpu/arch/arm/cpu-feats.cpp diff | blob | history
Packaging of ggml-org/llama.cpp