Skip to content

Commit 16ff288

Browse files
khosravipashakhosravipasha
committed
fix /r/n spacing + arch-fallback
1 parent 062c7c6 commit 16ff288

11 files changed

Lines changed: 15 additions & 32 deletions

File tree

‎ggml/src/ggml-common.h‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -176,8 +176,8 @@ typedef sycl::half2 ggml_half2;
176176

177177
#define QK1_0 128
178178
typedef struct {
179-
ggml_half d; // delta
180-
uint8_t qs[QK1_0 / 8]; // bits / quants
179+
ggml_half d; // delta
180+
uint8_t qs[QK1_0 / 8]; // bits / quants
181181
} block_q1_0;
182182
static_assert(sizeof(block_q1_0) == sizeof(ggml_half) + QK1_0 / 8, "wrong q1_0 block size/padding");
183183

‎ggml/src/ggml-cpu/arch-fallback.h‎

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -16,6 +16,7 @@
1616
#define ggml_vec_dot_q8_0_q8_0_generic ggml_vec_dot_q8_0_q8_0
1717
#define ggml_vec_dot_mxfp4_q8_0_generic ggml_vec_dot_mxfp4_q8_0
1818
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
19+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
1920
#define ggml_vec_dot_tq1_0_q8_K_generic ggml_vec_dot_tq1_0_q8_K
2021
#define ggml_vec_dot_tq2_0_q8_K_generic ggml_vec_dot_tq2_0_q8_K
2122
#define ggml_vec_dot_q2_K_q8_K_generic ggml_vec_dot_q2_K_q8_K
@@ -82,6 +83,7 @@
8283
#elif defined(__x86_64__) || defined(__i386__) || defined(_M_IX86) || defined(_M_X64)
8384
// quants.c
8485
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
86+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
8587
// repack.cpp
8688
#define ggml_quantize_mat_q8_0_4x4_generic ggml_quantize_mat_q8_0_4x4
8789
#define ggml_quantize_mat_q8_K_4x4_generic ggml_quantize_mat_q8_K_4x4
@@ -112,6 +114,7 @@
112114
// quants.c
113115
#define quantize_row_q8_K_generic quantize_row_q8_K
114116
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
117+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
115118
#define ggml_vec_dot_tq1_0_q8_K_generic ggml_vec_dot_tq1_0_q8_K
116119
#define ggml_vec_dot_tq2_0_q8_K_generic ggml_vec_dot_tq2_0_q8_K
117120
#define ggml_vec_dot_iq1_m_q8_K_generic ggml_vec_dot_iq1_m_q8_K
@@ -160,6 +163,7 @@
160163
#define ggml_vec_dot_iq1_m_q8_K_generic ggml_vec_dot_iq1_m_q8_K
161164
#define ggml_vec_dot_mxfp4_q8_0_generic ggml_vec_dot_mxfp4_q8_0
162165
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
166+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
163167
// repack.cpp
164168
#define ggml_quantize_mat_q8_0_4x4_generic ggml_quantize_mat_q8_0_4x4
165169
#define ggml_quantize_mat_q8_0_4x8_generic ggml_quantize_mat_q8_0_4x8
@@ -200,6 +204,7 @@
200204
#elif defined(__riscv)
201205
// quants.c
202206
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
207+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
203208
// repack.cpp
204209
#define ggml_quantize_mat_q8_0_4x1_generic ggml_quantize_mat_q8_0_4x1
205210
#define ggml_quantize_mat_q8_0_4x4_generic ggml_quantize_mat_q8_0_4x4
@@ -240,6 +245,7 @@
240245
// quants.c
241246
#define quantize_row_q8_K_generic quantize_row_q8_K
242247
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
248+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
243249
#define ggml_vec_dot_tq1_0_q8_K_generic ggml_vec_dot_tq1_0_q8_K
244250
#define ggml_vec_dot_tq2_0_q8_K_generic ggml_vec_dot_tq2_0_q8_K
245251
#define ggml_vec_dot_q2_K_q8_K_generic ggml_vec_dot_q2_K_q8_K
@@ -303,6 +309,7 @@
303309
#define ggml_vec_dot_iq4_xs_q8_K_generic ggml_vec_dot_iq4_xs_q8_K
304310
#define ggml_vec_dot_mxfp4_q8_0_generic ggml_vec_dot_mxfp4_q8_0
305311
#define ggml_vec_dot_nvfp4_q8_0_generic ggml_vec_dot_nvfp4_q8_0
312+
#define ggml_vec_dot_q1_0_q8_0_generic ggml_vec_dot_q1_0_q8_0
306313
// repack.cpp
307314
#define ggml_quantize_mat_q8_0_4x4_generic ggml_quantize_mat_q8_0_4x4
308315
#define ggml_quantize_mat_q8_0_4x8_generic ggml_quantize_mat_q8_0_4x8

‎ggml/src/ggml-cpu/arch/loongarch/quants.c‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -2156,7 +2156,3 @@ void ggml_vec_dot_iq4_xs_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const v
21562156
ggml_vec_dot_iq4_xs_q8_K_generic(n, s, bs, vx, bx, vy, by, nrc);
21572157
#endif
21582158
}
2159-
2160-
void ggml_vec_dot_q1_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) {
2161-
ggml_vec_dot_q1_0_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
2162-
}

‎ggml/src/ggml-cpu/arch/powerpc/quants.c‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -2302,7 +2302,3 @@ void ggml_vec_dot_iq4_xs_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const v
23022302
ggml_vec_dot_iq4_xs_q8_K_generic(n, s, bs, vx, bx, vy, by, nrc);
23032303
#endif
23042304
}
2305-
2306-
void ggml_vec_dot_q1_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) {
2307-
ggml_vec_dot_q1_0_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
2308-
}

‎ggml/src/ggml-cpu/arch/riscv/quants.c‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -3621,7 +3621,3 @@ void ggml_vec_dot_mxfp4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const vo
36213621
ggml_vec_dot_mxfp4_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
36223622
#endif
36233623
}
3624-
3625-
void ggml_vec_dot_q1_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) {
3626-
ggml_vec_dot_q1_0_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
3627-
}

‎ggml/src/ggml-cpu/arch/s390/quants.c‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1463,7 +1463,3 @@ void ggml_vec_dot_iq4_xs_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const v
14631463
ggml_vec_dot_iq4_xs_q8_K_generic(n, s, bs, vx, bx, vy, by, nrc);
14641464
#endif
14651465
}
1466-
1467-
void ggml_vec_dot_q1_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) {
1468-
ggml_vec_dot_q1_0_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
1469-
}

‎ggml/src/ggml-cpu/arch/wasm/quants.c‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1218,7 +1218,3 @@ void ggml_vec_dot_q6_K_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const voi
12181218
ggml_vec_dot_q6_K_q8_K_generic(n, s, bs, vx, bx, vy, by, nrc);
12191219
#endif
12201220
}
1221-
1222-
void ggml_vec_dot_q1_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) {
1223-
ggml_vec_dot_q1_0_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
1224-
}

‎ggml/src/ggml-cpu/arch/x86/quants.c‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1077,10 +1077,6 @@ void ggml_vec_dot_q8_0_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const voi
10771077
*s = sumf;
10781078
}
10791079

1080-
void ggml_vec_dot_q1_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) {
1081-
ggml_vec_dot_q1_0_q8_0_generic(n, s, bs, vx, bx, vy, by, nrc);
1082-
}
1083-
10841080
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) {
10851081
assert(nrc == 1);
10861082
UNUSED(nrc);

‎ggml/src/ggml.c‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -7661,8 +7661,8 @@ size_t ggml_quantize_chunk(
76617661
size_t result = 0;
76627662

76637663
switch (type) {
7664-
case GGML_TYPE_Q1_0: result = quantize_q1_0(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;
7665-
case GGML_TYPE_Q4_0: result = quantize_q4_0(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;
7664+
case GGML_TYPE_Q1_0: result = quantize_q1_0(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;
7665+
case GGML_TYPE_Q4_0: result = quantize_q4_0(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;
76667666
case GGML_TYPE_Q4_1: result = quantize_q4_1(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;
76677667
case GGML_TYPE_Q5_0: result = quantize_q5_0(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;
76687668
case GGML_TYPE_Q5_1: result = quantize_q5_1(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break;

‎gguf-py/gguf/constants.py‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -3966,7 +3966,7 @@ class LlamaFileType(IntEnum):
39663966
MOSTLY_TQ2_0 = 37 # except 1d tensors
39673967
MOSTLY_MXFP4_MOE = 38 # except 1d tensors
39683968
MOSTLY_NVFP4 = 39 # except 1d tensors
3969-
MOSTLY_Q1_0 = 40 # except 1d tensors
3969+
MOSTLY_Q1_0 = 40 # except 1d tensors
39703970

39713971
GUESSED = 1024 # not specified in the model file
39723972

@@ -4075,8 +4075,8 @@ class VisionProjectorType:
40754075
GGMLQuantizationType.TQ1_0: (256, 2 + 4 * 13),
40764076
GGMLQuantizationType.TQ2_0: (256, 2 + 64),
40774077
GGMLQuantizationType.MXFP4: (32, 1 + 16),
4078-
GGMLQuantizationType.NVFP4: (64, 4 + 32),
4079-
GGMLQuantizationType.Q1_0: (128, 2 + 16),
4078+
GGMLQuantizationType.NVFP4: (64, 4 + 32),
4079+
GGMLQuantizationType.Q1_0: (128, 2 + 16),
40804080
}
40814081

40824082

0 commit comments

Comments
 (0)