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 2025-12-19 11:15:35 +08:00
parent afaeb543bc
commit 7ef467ce20
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/matmul-ops.c
View File

@ -924,8 +924,8 @@ static void vec_dot_f16_f32(const int n, float * restrict s, const void * restri
const HVX_UVector * restrict vx = (const HVX_UVector * restrict) x;
const HVX_UVectorPair * restrict vy = (const HVX_UVectorPair * restrict) y;
uint32_t nv0 = n / 64; // num full fp16 hvx vectors
uint32_t nv1 = n % 64; // leftover elements
uint32_t nv0 = n / VLEN_FP16; // num full fp16 hvx vectors
uint32_t nv1 = n % VLEN_FP16; // leftover elements
// for some reason we need volatile here so that the compiler doesn't try anything funky
volatile HVX_Vector rsum = Q6_V_vsplat_R(0);