@@ -37,7 +37,7 @@ static const uint64_t table_b2b_0[1 << 8] = { B8(00, 10) }; // ( b) << 4
37
37
static const uint64_t table_b2b_1 [1 << 8 ] = { B8 (10 , 00 ) }; // (!b) << 4
38
38
#endif
39
39
40
- void quantize_row_q8_0_native (const float * GGML_RESTRICT x , void * GGML_RESTRICT vy , int64_t k ) {
40
+ void quantize_row_q8_0 (const float * GGML_RESTRICT x , void * GGML_RESTRICT vy , int64_t k ) {
41
41
assert (QK8_0 == 32 );
42
42
assert (k % QK8_0 == 0 );
43
43
const int nb = k / QK8_0 ;
@@ -81,7 +81,7 @@ void quantize_row_q8_0_native(const float * GGML_RESTRICT x, void * GGML_RESTRIC
81
81
#endif
82
82
}
83
83
84
- void quantize_row_q8_1_native (const float * GGML_RESTRICT x , void * GGML_RESTRICT vy , int64_t k ) {
84
+ void quantize_row_q8_1 (const float * GGML_RESTRICT x , void * GGML_RESTRICT vy , int64_t k ) {
85
85
assert (k % QK8_1 == 0 );
86
86
const int nb = k / QK8_1 ;
87
87
@@ -134,7 +134,7 @@ static const int8_t kvalues_iq4nl[16] = {-127, -104, -83, -65, -49, -35, -22, -1
134
134
135
135
//===================================== Dot products =================================
136
136
137
- void ggml_vec_dot_q4_0_q8_0_native (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 ) {
137
+ void ggml_vec_dot_q4_0_q8_0 (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 ) {
138
138
const int qk = QK8_0 ;
139
139
const int nb = n / qk ;
140
140
@@ -427,7 +427,7 @@ void ggml_vec_dot_q4_0_q8_0_native(int n, float * GGML_RESTRICT s, size_t bs, co
427
427
* s = sumf ;
428
428
}
429
429
430
- void ggml_vec_dot_q4_1_q8_1_native (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 ) {
430
+ void ggml_vec_dot_q4_1_q8_1 (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 ) {
431
431
const int qk = QK8_1 ;
432
432
const int nb = n / qk ;
433
433
@@ -586,7 +586,7 @@ void ggml_vec_dot_q4_1_q8_1_native(int n, float * GGML_RESTRICT s, size_t bs, co
586
586
* s = sumf ;
587
587
}
588
588
589
- void ggml_vec_dot_q5_0_q8_0_native (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 ) {
589
+ void ggml_vec_dot_q5_0_q8_0 (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 ) {
590
590
const int qk = QK8_0 ;
591
591
const int nb = n / qk ;
592
592
@@ -698,7 +698,7 @@ void ggml_vec_dot_q5_0_q8_0_native(int n, float * GGML_RESTRICT s, size_t bs, co
698
698
* s = sumf ;
699
699
}
700
700
701
- void ggml_vec_dot_q5_1_q8_1_native (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 ) {
701
+ void ggml_vec_dot_q5_1_q8_1 (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 ) {
702
702
const int qk = QK8_1 ;
703
703
const int nb = n / qk ;
704
704
@@ -816,7 +816,7 @@ void ggml_vec_dot_q5_1_q8_1_native(int n, float * GGML_RESTRICT s, size_t bs, co
816
816
* s = sumf ;
817
817
}
818
818
819
- void ggml_vec_dot_q8_0_q8_0_native (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 ) {
819
+ void ggml_vec_dot_q8_0_q8_0 (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 ) {
820
820
const int qk = QK8_0 ;
821
821
const int nb = n / qk ;
822
822
@@ -1063,7 +1063,7 @@ void ggml_vec_dot_q8_0_q8_0_native(int n, float * GGML_RESTRICT s, size_t bs, co
1063
1063
* s = sumf ;
1064
1064
}
1065
1065
1066
- void ggml_vec_dot_tq1_0_q8_K_native (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 ) {
1066
+ void ggml_vec_dot_tq1_0_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 ) {
1067
1067
assert (nrc == 1 );
1068
1068
UNUSED (nrc );
1069
1069
UNUSED (bx );
@@ -1274,7 +1274,7 @@ void ggml_vec_dot_tq1_0_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, c
1274
1274
#endif
1275
1275
}
1276
1276
1277
- void ggml_vec_dot_tq2_0_q8_K_native (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 ) {
1277
+ void ggml_vec_dot_tq2_0_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 ) {
1278
1278
assert (nrc == 1 );
1279
1279
UNUSED (nrc );
1280
1280
UNUSED (bx );
@@ -1400,7 +1400,7 @@ void ggml_vec_dot_tq2_0_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, c
1400
1400
#endif
1401
1401
}
1402
1402
1403
- void ggml_vec_dot_q2_K_q8_K_native (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 ) {
1403
+ void ggml_vec_dot_q2_K_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 ) {
1404
1404
assert (nrc == 1 );
1405
1405
UNUSED (nrc );
1406
1406
UNUSED (bx );
@@ -1768,7 +1768,7 @@ void ggml_vec_dot_q2_K_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, co
1768
1768
#endif
1769
1769
}
1770
1770
1771
- void ggml_vec_dot_q3_K_q8_K_native (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 ) {
1771
+ void ggml_vec_dot_q3_K_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 ) {
1772
1772
assert (n % QK_K == 0 );
1773
1773
assert (nrc == 1 );
1774
1774
UNUSED (nrc );
@@ -2120,7 +2120,7 @@ void ggml_vec_dot_q3_K_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, co
2120
2120
2121
2121
}
2122
2122
2123
- void ggml_vec_dot_q4_K_q8_K_native (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 ) {
2123
+ void ggml_vec_dot_q4_K_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 ) {
2124
2124
assert (n % QK_K == 0 );
2125
2125
assert (nrc == 1 );
2126
2126
UNUSED (nrc );
@@ -2342,7 +2342,7 @@ void ggml_vec_dot_q4_K_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, co
2342
2342
#endif
2343
2343
}
2344
2344
2345
- void ggml_vec_dot_q5_K_q8_K_native (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 ) {
2345
+ void ggml_vec_dot_q5_K_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 ) {
2346
2346
assert (n % QK_K == 0 );
2347
2347
assert (nrc == 1 );
2348
2348
UNUSED (nrc );
@@ -2494,7 +2494,7 @@ void ggml_vec_dot_q5_K_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, co
2494
2494
#endif
2495
2495
}
2496
2496
2497
- void ggml_vec_dot_q6_K_q8_K_native (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 ) {
2497
+ void ggml_vec_dot_q6_K_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 ) {
2498
2498
assert (n % QK_K == 0 );
2499
2499
#ifdef __ARM_FEATURE_MATMUL_INT8
2500
2500
assert ((nrc == 2 ) || (nrc == 1 ));
@@ -3027,7 +3027,7 @@ static const int8_t keven_signs_q2xs[1024] = {
3027
3027
};
3028
3028
#endif
3029
3029
3030
- void ggml_vec_dot_iq2_xxs_q8_K_native (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 ) {
3030
+ void ggml_vec_dot_iq2_xxs_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 ) {
3031
3031
assert (n % QK_K == 0 );
3032
3032
assert (nrc == 1 );
3033
3033
UNUSED (nrc );
@@ -3113,7 +3113,7 @@ void ggml_vec_dot_iq2_xxs_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs,
3113
3113
#endif
3114
3114
}
3115
3115
3116
- void ggml_vec_dot_iq2_xs_q8_K_native (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 ) {
3116
+ void ggml_vec_dot_iq2_xs_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 ) {
3117
3117
assert (n % QK_K == 0 );
3118
3118
assert (nrc == 1 );
3119
3119
UNUSED (nrc );
@@ -3219,7 +3219,7 @@ void ggml_vec_dot_iq2_xs_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs,
3219
3219
#endif
3220
3220
}
3221
3221
3222
- void ggml_vec_dot_iq2_s_q8_K_native (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 ) {
3222
+ void ggml_vec_dot_iq2_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 ) {
3223
3223
assert (n % QK_K == 0 );
3224
3224
assert (nrc == 1 );
3225
3225
UNUSED (nrc );
@@ -3351,7 +3351,7 @@ void ggml_vec_dot_iq2_s_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, c
3351
3351
3352
3352
}
3353
3353
3354
- void ggml_vec_dot_iq3_xxs_q8_K_native (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 ) {
3354
+ void ggml_vec_dot_iq3_xxs_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 ) {
3355
3355
assert (n % QK_K == 0 );
3356
3356
assert (nrc == 1 );
3357
3357
UNUSED (nrc );
@@ -3439,7 +3439,7 @@ void ggml_vec_dot_iq3_xxs_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs,
3439
3439
#endif
3440
3440
}
3441
3441
3442
- void ggml_vec_dot_iq3_s_q8_K_native (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 ) {
3442
+ void ggml_vec_dot_iq3_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 ) {
3443
3443
assert (n % QK_K == 0 );
3444
3444
assert (nrc == 1 );
3445
3445
UNUSED (nrc );
@@ -3587,7 +3587,7 @@ void ggml_vec_dot_iq3_s_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, c
3587
3587
#endif
3588
3588
}
3589
3589
3590
- void ggml_vec_dot_iq1_s_q8_K_native (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 ) {
3590
+ 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 ) {
3591
3591
assert (n % QK_K == 0 );
3592
3592
assert (nrc == 1 );
3593
3593
UNUSED (nrc );
@@ -3679,7 +3679,7 @@ void ggml_vec_dot_iq1_s_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, c
3679
3679
#endif
3680
3680
}
3681
3681
3682
- void ggml_vec_dot_iq1_m_q8_K_native (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 ) {
3682
+ void ggml_vec_dot_iq1_m_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 ) {
3683
3683
assert (n % QK_K == 0 );
3684
3684
assert (nrc == 1 );
3685
3685
UNUSED (nrc );
@@ -3814,7 +3814,7 @@ void ggml_vec_dot_iq1_m_q8_K_native(int n, float * GGML_RESTRICT s, size_t bs, c
3814
3814
#endif
3815
3815
}
3816
3816
3817
- void ggml_vec_dot_iq4_nl_q8_0_native (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 ) {
3817
+ void ggml_vec_dot_iq4_nl_q8_0 (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 ) {
3818
3818
assert (nrc == 1 );
3819
3819
UNUSED (nrc );
3820
3820
UNUSED (bx );
@@ -3874,7 +3874,7 @@ void ggml_vec_dot_iq4_nl_q8_0_native(int n, float * GGML_RESTRICT s, size_t bs,
3874
3874
* s = sumf ;
3875
3875
}
3876
3876
3877
- void ggml_vec_dot_iq4_xs_q8_K_native (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 ) {
3877
+ void ggml_vec_dot_iq4_xs_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 ) {
3878
3878
assert (nrc == 1 );
3879
3879
UNUSED (nrc );
3880
3880
UNUSED (bx );
0 commit comments