tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-29 04:35:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml-cpu: Faster IQ1 mul_mat_vec on AVX2 using BMI2 instructions (#12154)

Browse Source 
* 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
This commit is contained in:
Rémy O
2025-03-06 02:26:10 +01:00
committed by GitHub
parent 5e43f104cc
commit 07d1572347
8 changed files with 68 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
ggml/src/ggml-cpu/ggml-cpu.cpp
View File

@ -511,6 +511,9 @@ static ggml_backend_feature * ggml_backend_cpu_get_features(ggml_backend_reg_t r
if (ggml_cpu_has_fma()) {
features.push_back({ "FMA", "1" });
}
if (ggml_cpu_has_bmi2()) {
features.push_back({ "BMI2", "1" });
}
if (ggml_cpu_has_avx512()) {
features.push_back({ "AVX512", "1" });
}