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

wip

This commit is contained in:
chraac 2026-02-07 13:12:04 +08:00
parent e307f68c3b
commit 274ade5ca2
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/ggml-hexagon/htp/flash-attn-ops.c
View File

@ -379,7 +379,7 @@ static void flash_attn_ext_f16_thread(struct htp_ops_context * octx, int ith, in
for (uint32_t iv = 0; ic + VLEN_FP32 <= current_block_size; ic += VLEN_FP32, ++iv) {
// 1. Compute scores
float __attribute__((aligned(VLEN))) scores_arr[VLEN_FP32];
for (int j = 0; j < VLEN_FP32; j += 2) {
for (uint32_t j = 0; j < VLEN_FP32; j += 2) {
const uint32_t cur_ic = ic + j;
const uint8_t * k_ptr = k_base + cur_ic * size_k_row_padded;
hvx_dot_f16_f16_aa_rx2(&scores_arr[j], q_ptr_vtcm, k_ptr, k_ptr + size_k_row_padded, DK, scale);
@ -432,7 +432,7 @@ static void flash_attn_ext_f16_thread(struct htp_ops_context * octx, int ith, in
float __attribute__((aligned(VLEN))) p_arr[VLEN_FP32];
*(HVX_Vector *) p_arr = P;
for (int j = 0; j < VLEN_FP32; j += 2) {
for (uint32_t j = 0; j < VLEN_FP32; j += 2) {
const uint32_t cur_ic = ic2 + j;
const uint8_t * v_ptr = v_base + cur_ic * size_v_row_padded;
hvx_mad_f32_f16_aa_rx2(VKQ32, v_ptr, v_ptr + size_v_row_padded, p_arr[j], p_arr[j + 1], DV);