@@ -3854,7 +3854,7 @@ static inline __m128i get_scale_shuffle(int i) {
3854
3854
}
3855
3855
#endif
3856
3856
3857
- void ggml_vec_dot_q4_0_q8_0 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bx , const void * restrict vy , size_t by , int nrc ) {
3857
+ void ggml_vec_dot_q4_0_q8_0 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bbx , const void * restrict vy , size_t bby , int nrc ) {
3858
3858
const int qk = QK8_0 ;
3859
3859
const int nb = n / qk ;
3860
3860
@@ -3865,8 +3865,8 @@ void ggml_vec_dot_q4_0_q8_0(int n, float * restrict s, size_t bs, const void * r
3865
3865
assert (nrc == 1 );
3866
3866
#endif
3867
3867
UNUSED (nrc );
3868
- UNUSED (bx );
3869
- UNUSED (by );
3868
+ UNUSED (bbx );
3869
+ UNUSED (bby );
3870
3870
UNUSED (bs );
3871
3871
3872
3872
const block_q4_0 * restrict x = vx ;
@@ -4221,7 +4221,7 @@ void ggml_vec_dot_q4_0_q8_0(int n, float * restrict s, size_t bs, const void * r
4221
4221
#endif
4222
4222
}
4223
4223
4224
- void ggml_vec_dot_q4_1_q8_1 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bx , const void * restrict vy , size_t by , int nrc ) {
4224
+ void ggml_vec_dot_q4_1_q8_1 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bbx , const void * restrict vy , size_t bby , int nrc ) {
4225
4225
const int qk = QK8_1 ;
4226
4226
const int nb = n / qk ;
4227
4227
@@ -4232,8 +4232,8 @@ void ggml_vec_dot_q4_1_q8_1(int n, float * restrict s, size_t bs, const void * r
4232
4232
assert (nrc == 1 );
4233
4233
#endif
4234
4234
UNUSED (nrc );
4235
- UNUSED (bx );
4236
- UNUSED (by );
4235
+ UNUSED (bbx );
4236
+ UNUSED (bby );
4237
4237
UNUSED (bs );
4238
4238
4239
4239
const block_q4_1 * restrict x = vx ;
@@ -4439,16 +4439,16 @@ void ggml_vec_dot_q4_1_q8_1(int n, float * restrict s, size_t bs, const void * r
4439
4439
#endif
4440
4440
}
4441
4441
4442
- void ggml_vec_dot_q5_0_q8_0 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bx , const void * restrict vy , size_t by , int nrc ) {
4442
+ void ggml_vec_dot_q5_0_q8_0 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bbx , const void * restrict vy , size_t bby , int nrc ) {
4443
4443
const int qk = QK8_0 ;
4444
4444
const int nb = n / qk ;
4445
4445
4446
4446
assert (n % qk == 0 );
4447
4447
assert (qk == QK5_0 );
4448
4448
assert (nrc == 1 );
4449
4449
UNUSED (nrc );
4450
- UNUSED (bx );
4451
- UNUSED (by );
4450
+ UNUSED (bbx );
4451
+ UNUSED (bby );
4452
4452
UNUSED (bs );
4453
4453
4454
4454
const block_q5_0 * restrict x = vx ;
@@ -4730,16 +4730,16 @@ void ggml_vec_dot_q5_0_q8_0(int n, float * restrict s, size_t bs, const void * r
4730
4730
#endif
4731
4731
}
4732
4732
4733
- void ggml_vec_dot_q5_1_q8_1 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bx , const void * restrict vy , size_t by , int nrc ) {
4733
+ void ggml_vec_dot_q5_1_q8_1 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bbx , const void * restrict vy , size_t bby , int nrc ) {
4734
4734
const int qk = QK8_1 ;
4735
4735
const int nb = n / qk ;
4736
4736
4737
4737
assert (n % qk == 0 );
4738
4738
assert (qk == QK5_1 );
4739
4739
assert (nrc == 1 );
4740
4740
UNUSED (nrc );
4741
- UNUSED (bx );
4742
- UNUSED (by );
4741
+ UNUSED (bbx );
4742
+ UNUSED (bby );
4743
4743
UNUSED (bs );
4744
4744
4745
4745
const block_q5_1 * restrict x = vx ;
@@ -5034,7 +5034,7 @@ void ggml_vec_dot_q5_1_q8_1(int n, float * restrict s, size_t bs, const void * r
5034
5034
#endif
5035
5035
}
5036
5036
5037
- void ggml_vec_dot_q8_0_q8_0 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bx , const void * restrict vy , size_t by , int nrc ) {
5037
+ void ggml_vec_dot_q8_0_q8_0 (int n , float * restrict s , size_t bs , const void * restrict vx , size_t bbx , const void * restrict vy , size_t bby , int nrc ) {
5038
5038
const int qk = QK8_0 ;
5039
5039
const int nb = n / qk ;
5040
5040
@@ -5045,8 +5045,8 @@ void ggml_vec_dot_q8_0_q8_0(int n, float * restrict s, size_t bs, const void * r
5045
5045
assert (nrc == 1 );
5046
5046
#endif
5047
5047
UNUSED (nrc );
5048
- UNUSED (bx );
5049
- UNUSED (by );
5048
+ UNUSED (bbx );
5049
+ UNUSED (bby );
5050
5050
UNUSED (bs );
5051
5051
5052
5052
const block_q8_0 * restrict x = vx ;
@@ -9290,8 +9290,13 @@ static inline __m256i mul_add_epi8(const __m256i x, const __m256i y) {
9290
9290
}
9291
9291
#endif
9292
9292
9293
- void ggml_vec_dot_iq1_s_q8_K ( const int n , float * restrict s , const void * restrict vx , const void * restrict vy ) {
9293
+ void ggml_vec_dot_iq1_s_q8_K ( int n , float * GGML_RESTRICT s , size_t bs , const void * GGML_RESTRICT vx , size_t bx , const void * GGML_RESTRICT vy , size_t by , int nrc ) {
9294
9294
assert (n % QK_K == 0 );
9295
+ assert (nrc == 1 );
9296
+ UNUSED (nrc );
9297
+ UNUSED (bx );
9298
+ UNUSED (by );
9299
+ UNUSED (bs );
9295
9300
9296
9301
const block_iq1_s * restrict x = vx ;
9297
9302
const block_q8_K * restrict y = vy ;
0 commit comments