Skip to content

Commit

Permalink
ggml-cpu: fix typo in gemv/gemm iq4_nl_4_4 (ggerganov#10580)
Browse files Browse the repository at this point in the history
  • Loading branch information
FanShupei authored Nov 29, 2024
1 parent 0f77aae commit 4b3242b
Showing 1 changed file with 2 additions and 2 deletions.
4 changes: 2 additions & 2 deletions ggml/src/ggml-cpu/ggml-cpu-aarch64.c
Original file line number Diff line number Diff line change
Expand Up @@ -1020,7 +1020,7 @@ void ggml_gemv_iq4_nl_4x4_q8_0(int n, float * restrict s, size_t bs, const void
float * res_ptr = s;

for (int x = 0; x < nc / ncols_interleaved; x++) {
const block_q4_0x4 * b_ptr = (const block_q4_0x4 *) vx + (x * nb);
const block_iq4_nlx4 * b_ptr = (const block_iq4_nlx4 *) vx + (x * nb);

float32x4_t sumf = vdupq_n_f32(0);
for (int l = 0; l < nb; l++) {
Expand Down Expand Up @@ -3507,7 +3507,7 @@ void ggml_gemm_iq4_nl_4x4_q8_0(int n, float * restrict s, size_t bs, const void
for (int y = 0; y < nr / 4; y++) {
const block_q8_0x4 * a_ptr = (const block_q8_0x4 *) vy + (y * nb);
for (int x = 0; x < nc / ncols_interleaved; x++) {
const block_q4_0x4 * b_ptr = (const block_q4_0x4 *) vx + (x * nb);
const block_iq4_nlx4 * b_ptr = (const block_iq4_nlx4 *) vx + (x * nb);

float32x4_t sumf[4];
for (int m = 0; m < 4; m++) {
Expand Down

0 comments on commit 4b3242b

Please sign in to comment.