vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

uint -> uint32_t

Browse source
This commit is contained in:
Johannes Gäßler 2024-06-30 12:05:44 +02:00
parent ec15f4d520
commit 0480dab44a
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
ggml/src/ggml-cuda/vecdotq.cuh
View file

@ -846,7 +846,7 @@ static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1(
const int q2 = get_int_b2(bq2->qs, iqs);
const uint8_t * aux8 = (const uint8_t *) &q2;
const uint aux32 = get_int_b2(bq2->qs, iqs + 1);
const uint32_t aux32 = get_int_b2(bq2->qs, iqs + 1);
int sumi = 0;
#pragma unroll
@ -976,7 +976,7 @@ static __device__ __forceinline__ float vec_dot_iq3_xxs_q8_1(
const int2 q3_packed = make_int2(get_int_b2(bq3->qs, iqs), get_int_b2(bq3->qs, iqs+1));
const uint8_t * q3 = (const uint8_t *) &q3_packed;
const uint aux32 = get_int_b2(bq3->qs, QK_K/16 + iqs/2);
const uint32_t aux32 = get_int_b2(bq3->qs, QK_K/16 + iqs/2);
int sumi = 0;
#pragma unroll