root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-26 11:24:35 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

ggml : Fix compilation issues on ARM platform when building without fp16 (#10811)

Browse Source
This commit is contained in:
Karol Kontny 2024-12-13 01:04:19 +01:00 committed by GitHub
parent adffa6ffd5
commit d583cd03f6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 16 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
ggml/src/ggml-cpu/ggml-cpu.c
View File

@ -454,21 +454,21 @@ const struct ggml_type_traits_cpu * ggml_get_type_traits_cpu(enum ggml_type type
#define GGML_F32x4_ADD vaddq_f32 #define GGML_F32x4_ADD vaddq_f32
#define GGML_F32x4_MUL vmulq_f32 #define GGML_F32x4_MUL vmulq_f32
#define GGML_F32x4_REDUCE_ONE(x) vaddvq_f32(x) #define GGML_F32x4_REDUCE_ONE(x) vaddvq_f32(x)
#define GGML_F32x4_REDUCE(res, x) \ #define GGML_F32x4_REDUCE(res, x) \
{ \ { \
int offset = GGML_F32_ARR >> 1; \ int offset = GGML_F32_ARR >> 1; \
for (int i = 0; i < offset; ++i) { \ for (int i = 0; i < offset; ++i) { \
(x)[i] = vaddq_f32((x)[i], (x)[offset+i]); \ (x)[i] = vaddq_f32((x)[i], (x)[offset+i]); \
} \ } \
offset >>= 1; \ offset >>= 1; \
for (int i = 0; i < offset; ++i) { \ for (int i = 0; i < offset; ++i) { \
(x)[i] = vaddq_f32((x)[i], (x)[offset+i]); \ (x)[i] = vaddq_f32((x)[i], (x)[offset+i]); \
} \ } \
offset >>= 1; \ offset >>= 1; \
for (int i = 0; i < offset; ++i) { \ for (int i = 0; i < offset; ++i) { \
(x)[i] = vaddq_f32((x)[i], (x)[offset+i]); \ (x)[i] = vaddq_f32((x)[i], (x)[offset+i]); \
} \ } \
(res) = GGML_F32x4_REDUCE_ONE((x)[0]); \ (res) = (ggml_float) GGML_F32x4_REDUCE_ONE((x)[0]); \
} }
#define GGML_F32_VEC GGML_F32x4 #define GGML_F32_VEC GGML_F32x4
@ -2395,7 +2395,7 @@ static void ggml_init_arm_arch_features(void) {
uint32_t hwcap2 = getauxval(AT_HWCAP2); uint32_t hwcap2 = getauxval(AT_HWCAP2);
ggml_arm_arch_features.has_neon = !!(hwcap & HWCAP_ASIMD); ggml_arm_arch_features.has_neon = !!(hwcap & HWCAP_ASIMD);
ggml_arm_arch_features.has_dotprod = !!(hwcap && HWCAP_ASIMDDP); ggml_arm_arch_features.has_dotprod = !!(hwcap & HWCAP_ASIMDDP);
ggml_arm_arch_features.has_i8mm = !!(hwcap2 & HWCAP2_I8MM); ggml_arm_arch_features.has_i8mm = !!(hwcap2 & HWCAP2_I8MM);
ggml_arm_arch_features.has_sve = !!(hwcap & HWCAP_SVE); ggml_arm_arch_features.has_sve = !!(hwcap & HWCAP_SVE);