]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
ggml-cpu: Faster IQ1 mul_mat_vec on AVX2 using BMI2 instructions (#12154)
authorRémy O <redacted>
Thu, 6 Mar 2025 01:26:10 +0000 (02:26 +0100)
committerGitHub <redacted>
Thu, 6 Mar 2025 01:26:10 +0000 (02:26 +0100)
commit07d15723470a0a5b15d8ccad1aff5b20354ffbe1
tree036285e5a3ee0834dd0fc23df1473d4aa4611754
parent5e43f104cca1a14874e980326a506b44fde022b8
ggml-cpu: Faster IQ1 mul_mat_vec on AVX2 using BMI2 instructions (#12154)

* ggml-cpu: Faster IQ1 mul_mat_vec on AVX2 using BMI2 instructions

* cmake: Add GGML_BMI2 build option

* ggml: enable BMI2 on relevant CPU variants

* ggml-cpu: include BMI2 in backend score

* ggml-cpu: register BMI2 in ggml_backend_cpu_get_features

* ggml-cpu: add __BMI2__ define when using MSVC
ggml/CMakeLists.txt
ggml/include/ggml-cpu.h
ggml/src/CMakeLists.txt
ggml/src/ggml-cpu/CMakeLists.txt
ggml/src/ggml-cpu/cpu-feats-x86.cpp
ggml/src/ggml-cpu/ggml-cpu-quants.c
ggml/src/ggml-cpu/ggml-cpu.c
ggml/src/ggml-cpu/ggml-cpu.cpp