]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
llama : add thread safety test (llama/14035)
authorDiego Devesa <redacted>
Mon, 16 Jun 2025 15:11:43 +0000 (08:11 -0700)
committerGeorgi Gerganov <redacted>
Wed, 18 Jun 2025 07:21:15 +0000 (10:21 +0300)
commit7b2904659fc6ffa77bca0e50f705d079acd5cfd0
tree87c0f96022c6514cde0cb5be814a7d45f44c4a6e
parentfa57ab699e7db23704fb4b5c7ba80b41138d8e8f
llama : add thread safety test (llama/14035)

* llama : add thread safety test

* llamafile : remove global state

* llama : better LLAMA_SPLIT_MODE_NONE logic

when main_gpu < 0 GPU devices are not used

---------

Co-authored-by: Georgi Gerganov <redacted>
src/ggml-cpu/ggml-cpu-impl.h
src/ggml-cpu/ggml-cpu.c
src/ggml-cpu/llamafile/sgemm.cpp