Skip to content

Commit 097f253

Browse files
xctanxctan
committed
try to fix clang compilation errors
1 parent 3ac6896 commit 097f253

File tree

2 files changed

+36
-36
lines changed

2 files changed

+36
-36
lines changed

‎ggml/src/ggml-cpu/quants.c‎

Lines changed: 23 additions & 23 deletions
Original file line numberDiff line numberDiff line change
@@ -35,15 +35,15 @@ void quantize_row_q5_1(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, in
3535
quantize_row_q5_1_ref(x, y, k);
3636
}
3737

38-
GGML_CPU_NATIVE_IMPL(quantize_row_q8_0)
3938
void quantize_row_q8_0_generic(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k) {
4039
quantize_row_q8_0_ref(x, y, k);
4140
}
41+
GGML_CPU_NATIVE_IMPL(quantize_row_q8_0)
4242

43-
GGML_CPU_NATIVE_IMPL(quantize_row_q8_1)
4443
void quantize_row_q8_1_generic(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k) {
4544
quantize_row_q8_1_ref(x, y, k);
4645
}
46+
GGML_CPU_NATIVE_IMPL(quantize_row_q8_1)
4747

4848
//
4949
// 2-6 bit quantization in super-blocks
@@ -113,7 +113,6 @@ void quantize_row_q8_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, in
113113

114114
//===================================== Dot products =================================
115115

116-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q4_0_q8_0)
117116
void ggml_vec_dot_q4_0_q8_0_generic(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) {
118117
const int qk = QK8_0;
119118
const int nb = n / qk;
@@ -149,9 +148,9 @@ void ggml_vec_dot_q4_0_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, c
149148

150149
*s = sumf;
151150
}
151+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q4_0_q8_0)
152152

153153
// TODO: add WASM SIMD
154-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q4_1_q8_1)
155154
void ggml_vec_dot_q4_1_q8_1_generic(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) {
156155
const int qk = QK8_1;
157156
const int nb = n / qk;
@@ -187,8 +186,8 @@ void ggml_vec_dot_q4_1_q8_1_generic(int n, float * GGML_RESTRICT s, size_t bs, c
187186

188187
*s = sumf;
189188
}
189+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q4_1_q8_1)
190190

191-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q5_0_q8_0)
192191
void ggml_vec_dot_q5_0_q8_0_generic(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) {
193192
const int qk = QK8_0;
194193
const int nb = n / qk;
@@ -231,8 +230,8 @@ void ggml_vec_dot_q5_0_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, c
231230

232231
*s = sumf;
233232
}
233+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q5_0_q8_0)
234234

235-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q5_1_q8_1)
236235
void ggml_vec_dot_q5_1_q8_1_generic(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) {
237236
const int qk = QK8_1;
238237
const int nb = n / qk;
@@ -275,8 +274,8 @@ void ggml_vec_dot_q5_1_q8_1_generic(int n, float * GGML_RESTRICT s, size_t bs, c
275274

276275
*s = sumf;
277276
}
277+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q5_1_q8_1)
278278

279-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q8_0_q8_0)
280279
void ggml_vec_dot_q8_0_q8_0_generic(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) {
281280
const int qk = QK8_0;
282281
const int nb = n / qk;
@@ -306,8 +305,8 @@ void ggml_vec_dot_q8_0_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, c
306305

307306
*s = sumf;
308307
}
308+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q8_0_q8_0)
309309

310-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_tq1_0_q8_K)
311310
void ggml_vec_dot_tq1_0_q8_K_generic(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) {
312311
assert(nrc == 1);
313312
UNUSED(nrc);
@@ -359,8 +358,8 @@ void ggml_vec_dot_tq1_0_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
359358

360359
*s = sumf;
361360
}
361+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_tq1_0_q8_K)
362362

363-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_tq2_0_q8_K)
364363
void ggml_vec_dot_tq2_0_q8_K_generic(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) {
365364
assert(nrc == 1);
366365
UNUSED(nrc);
@@ -392,8 +391,8 @@ void ggml_vec_dot_tq2_0_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
392391

393392
*s = sumf;
394393
}
394+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_tq2_0_q8_K)
395395

396-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q2_K_q8_K)
397396
void ggml_vec_dot_q2_K_q8_K_generic(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) {
398397
assert(nrc == 1);
399398
UNUSED(nrc);
@@ -445,8 +444,8 @@ void ggml_vec_dot_q2_K_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, c
445444
}
446445
*s = sumf;
447446
}
447+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q2_K_q8_K)
448448

449-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q3_K_q8_K)
450449
void ggml_vec_dot_q3_K_q8_K_generic(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) {
451450
assert(n % QK_K == 0);
452451
assert(nrc == 1);
@@ -525,8 +524,8 @@ void ggml_vec_dot_q3_K_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, c
525524
for (int l = 0; l < 8; ++l) sumf += sums[l];
526525
*s = sumf;
527526
}
527+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q3_K_q8_K)
528528

529-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q4_K_q8_K)
530529
void ggml_vec_dot_q4_K_q8_K_generic(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) {
531530
assert(n % QK_K == 0);
532531
assert(nrc == 1);
@@ -601,8 +600,8 @@ void ggml_vec_dot_q4_K_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, c
601600
for (int l = 0; l < 8; ++l) sumf += sums[l];
602601
*s = sumf;
603602
}
603+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q4_K_q8_K)
604604

605-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q5_K_q8_K)
606605
void ggml_vec_dot_q5_K_q8_K_generic(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) {
607606
assert(n % QK_K == 0);
608607
assert(nrc == 1);
@@ -682,8 +681,8 @@ void ggml_vec_dot_q5_K_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, c
682681
for (int l = 0; l < 8; ++l) sumf += sums[l];
683682
*s = sumf;
684683
}
684+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q5_K_q8_K)
685685

686-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q6_K_q8_K)
687686
void ggml_vec_dot_q6_K_q8_K_generic(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) {
688687
assert(n % QK_K == 0);
689688
assert(nrc == 1);
@@ -738,8 +737,8 @@ void ggml_vec_dot_q6_K_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, c
738737
for (int l = 0; l < 8; ++l) sumf += sums[l];
739738
*s = sumf;
740739
}
740+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_q6_K_q8_K)
741741

742-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq2_xxs_q8_K)
743742
void ggml_vec_dot_iq2_xxs_q8_K_generic(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) {
744743
assert(n % QK_K == 0);
745744
assert(nrc == 1);
@@ -781,8 +780,8 @@ void ggml_vec_dot_iq2_xxs_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs
781780
}
782781
*s = 0.125f * sumf;
783782
}
783+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq2_xxs_q8_K)
784784

785-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq2_xs_q8_K)
786785
void ggml_vec_dot_iq2_xs_q8_K_generic(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) {
787786
assert(n % QK_K == 0);
788787
assert(nrc == 1);
@@ -832,8 +831,8 @@ void ggml_vec_dot_iq2_xs_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
832831
}
833832
*s = 0.125f * sumf;
834833
}
834+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq2_xs_q8_K)
835835

836-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq2_s_q8_K)
837836
void ggml_vec_dot_iq2_s_q8_K_generic(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) {
838837
assert(n % QK_K == 0);
839838
assert(nrc == 1);
@@ -885,8 +884,8 @@ void ggml_vec_dot_iq2_s_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
885884

886885
*s = 0.125f * sumf;
887886
}
887+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq2_s_q8_K)
888888

889-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq3_xxs_q8_K)
890889
void ggml_vec_dot_iq3_xxs_q8_K_generic(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) {
891890
assert(n % QK_K == 0);
892891
assert(nrc == 1);
@@ -930,8 +929,8 @@ void ggml_vec_dot_iq3_xxs_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs
930929
}
931930
*s = 0.25f * sumf;
932931
}
932+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq3_xxs_q8_K)
933933

934-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq3_s_q8_K)
935934
void ggml_vec_dot_iq3_s_q8_K_generic(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) {
936935
assert(n % QK_K == 0);
937936
assert(nrc == 1);
@@ -987,8 +986,8 @@ void ggml_vec_dot_iq3_s_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
987986
}
988987
*s = sumf;
989988
}
989+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq3_s_q8_K)
990990

991-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq1_s_q8_K)
992991
void ggml_vec_dot_iq1_s_q8_K_generic(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) {
993992
assert(n % QK_K == 0);
994993
assert(nrc == 1);
@@ -1031,8 +1030,8 @@ void ggml_vec_dot_iq1_s_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
10311030

10321031
*s = sumf;
10331032
}
1033+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq1_s_q8_K)
10341034

1035-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq1_m_q8_K)
10361035
void ggml_vec_dot_iq1_m_q8_K_generic(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) {
10371036
assert(n % QK_K == 0);
10381037
assert(nrc == 1);
@@ -1093,8 +1092,8 @@ void ggml_vec_dot_iq1_m_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
10931092

10941093
*s = sumf;
10951094
}
1095+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq1_m_q8_K)
10961096

1097-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq4_nl_q8_0)
10981097
void ggml_vec_dot_iq4_nl_q8_0_generic(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) {
10991098
assert(nrc == 1);
11001099
UNUSED(nrc);
@@ -1123,8 +1122,8 @@ void ggml_vec_dot_iq4_nl_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
11231122
}
11241123
*s = sumf;
11251124
}
1125+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq4_nl_q8_0)
11261126

1127-
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq4_xs_q8_K)
11281127
void ggml_vec_dot_iq4_xs_q8_K_generic(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) {
11291128
assert(nrc == 1);
11301129
UNUSED(nrc);
@@ -1170,6 +1169,7 @@ void ggml_vec_dot_iq4_xs_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
11701169
}
11711170
*s = sumf;
11721171
}
1172+
GGML_CPU_NATIVE_IMPL(ggml_vec_dot_iq4_xs_q8_K)
11731173

11741174
// ============================ 4-bit non-linear quants
11751175

‎ggml/src/ggml-cpu/repack.cpp‎

Lines changed: 13 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -48,7 +48,6 @@ static const int8_t kvalues_iq4nl[16] = {-127, -104, -83, -65, -49, -35, -22, -1
4848

4949
extern "C" {
5050

51-
GGML_CPU_NATIVE_IMPL(ggml_quantize_mat_q8_0_4x4)
5251
void ggml_quantize_mat_q8_0_4x4_generic(const float * GGML_RESTRICT x, void * GGML_RESTRICT vy, int64_t k) {
5352
assert(QK8_0 == 32);
5453
assert(k % QK8_0 == 0);
@@ -86,8 +85,8 @@ void ggml_quantize_mat_q8_0_4x4_generic(const float * GGML_RESTRICT x, void * GG
8685
}
8786
}
8887
}
88+
GGML_CPU_NATIVE_IMPL(ggml_quantize_mat_q8_0_4x4)
8989

90-
GGML_CPU_NATIVE_IMPL(ggml_quantize_mat_q8_0_4x8)
9190
void ggml_quantize_mat_q8_0_4x8_generic(const float * GGML_RESTRICT x, void * GGML_RESTRICT vy, int64_t k) {
9291
assert(QK8_0 == 32);
9392
assert(k % QK8_0 == 0);
@@ -125,8 +124,8 @@ void ggml_quantize_mat_q8_0_4x8_generic(const float * GGML_RESTRICT x, void * GG
125124
}
126125
}
127126
}
127+
GGML_CPU_NATIVE_IMPL(ggml_quantize_mat_q8_0_4x8)
128128

129-
GGML_CPU_NATIVE_IMPL(ggml_quantize_mat_q8_K_4x8)
130129
void ggml_quantize_mat_q8_K_4x8_generic(const float * GGML_RESTRICT x, void * GGML_RESTRICT vy, int64_t k) {
131130
assert(QK_K == 256);
132131
assert(k % QK_K == 0);
@@ -177,6 +176,7 @@ void ggml_quantize_mat_q8_K_4x8_generic(const float * GGML_RESTRICT x, void * GG
177176
}
178177
}
179178
}
179+
GGML_CPU_NATIVE_IMPL(ggml_quantize_mat_q8_K_4x8)
180180

181181
} // extern "C"
182182

@@ -203,7 +203,6 @@ template <> void ggml_quantize_mat_t<8, GGML_TYPE_Q8_K>(const float * GGML_RESTR
203203

204204
extern "C" {
205205

206-
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_0_4x4_q8_0)
207206
void ggml_gemv_q4_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
208207
const int qk = QK8_0;
209208
const int nb = n / qk;
@@ -247,8 +246,8 @@ void ggml_gemv_q4_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
247246
for (int j = 0; j < ncols_interleaved; j++) s[x * ncols_interleaved + j] = sumf[j];
248247
}
249248
}
249+
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_0_4x4_q8_0)
250250

251-
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_0_4x8_q8_0)
252251
void ggml_gemv_q4_0_4x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
253252
const int qk = QK8_0;
254253
const int nb = n / qk;
@@ -292,8 +291,8 @@ void ggml_gemv_q4_0_4x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
292291
for (int j = 0; j < ncols_interleaved; j++) s[x * ncols_interleaved + j] = sumf[j];
293292
}
294293
}
294+
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_0_4x8_q8_0)
295295

296-
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_0_8x8_q8_0)
297296
void ggml_gemv_q4_0_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
298297
const int qk = QK8_0;
299298
const int nb = n / qk;
@@ -339,8 +338,8 @@ void ggml_gemv_q4_0_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
339338
}
340339
}
341340
}
341+
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_0_8x8_q8_0)
342342

343-
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_K_8x8_q8_K)
344343
void ggml_gemv_q4_K_8x8_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
345344
const int qk = QK_K;
346345
const int nb = n / qk;
@@ -418,8 +417,8 @@ void ggml_gemv_q4_K_8x8_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
418417
}
419418
}
420419
}
420+
GGML_CPU_NATIVE_IMPL(ggml_gemv_q4_K_8x8_q8_K)
421421

422-
GGML_CPU_NATIVE_IMPL(ggml_gemv_iq4_nl_4x4_q8_0)
423422
void ggml_gemv_iq4_nl_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
424423
const int qk = QK8_0;
425424
const int nb = n / qk;
@@ -465,8 +464,8 @@ void ggml_gemv_iq4_nl_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs
465464
}
466465
}
467466
}
467+
GGML_CPU_NATIVE_IMPL(ggml_gemv_iq4_nl_4x4_q8_0)
468468

469-
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_0_4x4_q8_0)
470469
void ggml_gemm_q4_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
471470
const int qk = QK8_0;
472471
const int nb = n / qk;
@@ -522,8 +521,8 @@ void ggml_gemm_q4_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
522521
}
523522
}
524523
}
524+
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_0_4x4_q8_0)
525525

526-
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_0_4x8_q8_0)
527526
void ggml_gemm_q4_0_4x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
528527
const int qk = QK8_0;
529528
const int nb = n / qk;
@@ -577,8 +576,8 @@ void ggml_gemm_q4_0_4x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
577576
}
578577
}
579578
}
579+
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_0_4x8_q8_0)
580580

581-
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_0_8x8_q8_0)
582581
void ggml_gemm_q4_0_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
583582
const int qk = QK8_0;
584583
const int nb = n / qk;
@@ -632,8 +631,8 @@ void ggml_gemm_q4_0_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs,
632631
}
633632
}
634633
}
634+
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_0_8x8_q8_0)
635635

636-
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_K_8x8_q8_K)
637636
void ggml_gemm_q4_K_8x8_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
638637
const int qk = QK_K;
639638
const int nb = n / qk;
@@ -722,8 +721,8 @@ void ggml_gemm_q4_K_8x8_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
722721
}
723722
}
724723
}
724+
GGML_CPU_NATIVE_IMPL(ggml_gemm_q4_K_8x8_q8_K)
725725

726-
GGML_CPU_NATIVE_IMPL(ggml_gemm_iq4_nl_4x4_q8_0)
727726
void ggml_gemm_iq4_nl_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
728727
const int qk = QK8_0;
729728
const int nb = n / qk;
@@ -779,6 +778,7 @@ void ggml_gemm_iq4_nl_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs
779778
}
780779
}
781780
}
781+
GGML_CPU_NATIVE_IMPL(ggml_gemm_iq4_nl_4x4_q8_0)
782782

783783
} // extern "C"
784784

0 commit comments

Comments
 (0)