]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 14d3ba4)
sycl: set fattn_vec_nthreads to 256 for Battlemage (#25205)
authorTitaniumtown <redacted>
Tue, 14 Jul 2026 09:00:00 +0000 (05:00 -0400)
committerGitHub <redacted>
Tue, 14 Jul 2026 09:00:00 +0000 (12:00 +0300)
commitc1063ac9d75fe9238c82de94fc4b1f58b2cc6196
tree61da712d5d8685178b722881137b951ebbbe3c49tree
parent14d3ba45f3369e75a308212399cfada5d349883bcommit | diff
sycl: set fattn_vec_nthreads to 256 for Battlemage (#25205)

Currently detects lunarlake + battlemage / xe2 and
sets the value to 256.

Keeps default at 128, Intel's ARC Alchemist's prefered value.
ggml/src/ggml-sycl/fattn-vec.hpp diff | blob | history
Packaging of ggml-org/llama.cpp