happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

ggml-cpu/x86: fix unused changemask warning in repack (#20692)

This commit is contained in:
Shaw Nguyen 2026-03-18 23:45:06 +07:00 committed by GitHub
parent 4efd326e71
commit 78d550b541
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cpu/arch/x86/repack.cpp
View File

@ -531,7 +531,6 @@ static void gemv_q4_b32_8x8_q8_0_lut_avx(int n, float * GGML_RESTRICT s, size_t
UNUSED(bs);
__m128i changemask = _mm_set_epi8(15, 14, 7, 6, 13, 12, 5, 4, 11, 10, 3, 2, 9, 8, 1, 0);
__m256i finalpermutemask = _mm256_set_epi32(7, 5, 3, 1, 6, 4, 2, 0);
// Permute mask used for easier vector processing at later stages
@ -580,6 +579,7 @@ static void gemv_q4_b32_8x8_q8_0_lut_avx(int n, float * GGML_RESTRICT s, size_t
if constexpr (
std::is_same_v<block_tx8, block_q4_0x8> ||
std::is_same_v<block_tx8, block_iq4_nlx8>) {
const __m128i changemask = _mm_set_epi8(15, 14, 7, 6, 13, 12, 5, 4, 11, 10, 3, 2, 9, 8, 1, 0);
col_scale_f32 = GGML_F32Cx8_REARRANGE_LOAD(b_ptr[b].d, changemask);
} else if constexpr (std::is_same_v<block_tx8, block_mxfp4x8>) {
// Load 8 E8M0 exponents and convert to float via LUT