ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-06 06:20:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

q8_k_r8: this change for NEON got lost?

Browse Source
This commit is contained in:
Kawrakow
2024-12-16 13:31:21 +01:00
parent e811de75e9
commit 3283fbeb82
1 changed files with 2 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
ggml/src/iqk/iqk_mul_mat.cpp
View File

@@ -9353,12 +9353,8 @@ void mul_mat_q8_k_r8_q8_k(int n, const void * vx, size_t bx, const DataInfo& inf
auto d4h = vcvt_f32_f16(vld1_f16((const float16_t *)iq8[ibl].d+4));
int32x4_t isum[2*nrc_y] = {};
for (int ib = 0; ib < QK_K/16; ++ib) {
auto q1 = vld1q_u8_x4(iq8[ibl].qs + 128*ib + 0);
auto q2 = vld1q_u8_x4(iq8[ibl].qs + 128*ib + 64);
for (int k = 0; k < 4; ++k) {
q1.val[k] = veorq_u8(q1.val[k], vdupq_n_u8(0x80));
q2.val[k] = veorq_u8(q2.val[k], vdupq_n_u8(0x80));
}
auto q1 = vld1q_s8_x4(iq8[ibl].qs + 128*ib + 0);
auto q2 = vld1q_s8_x4(iq8[ibl].qs + 128*ib + 64);
for (int iy = 0; iy < nrc_y; ++iy) {
auto y = vld1q_s8(q8.y[iy][ibl].qs+16*ib);
isum[2*iy+0] = vdotq_laneq_s32(isum[2*iy+0], q1.val[0], y, 0);