vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp

Browse source 
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Djip007 2024-12-07 12:51:22 +01:00 • committed by GitHub
parent b14b47132a
commit 7dc8a3e2d6
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp
View file

@ -4121,7 +4121,7 @@ static const tensor_traits<block_iq4_nl, 4, 4> iq4_nl_4x4_q8_0;
static const ggml::cpu::tensor_traits * ggml_aarch64_get_optimal_repack_type(const struct ggml_tensor * cur) {
if (cur->type == GGML_TYPE_Q4_0) {
if (ggml_cpu_has_avx2() || (ggml_cpu_has_sve() && ggml_cpu_has_matmul_int8() && ggml_cpu_get_sve_cnt() == QK8_0)) {
if (cur->ne[1] % 8==0) {
if (cur->ne[1] % 8 == 0) {
return &ggml::cpu::aarch64::q4_0_8x8_q8_0;
}
}