CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml-cpu: dedup ggml_table_f32_f16 from simd-mappings.h

Browse Source 
we rely on the variable declaration in ggml-cpu.c instead

Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-06-26 00:09:14 +08:00
parent 5f2a09a8f6
commit f71b21d2f7
1 changed files with 8 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
ggml/src/ggml-cpu/simd-mappings.h
View File

@@ -18,6 +18,10 @@
#include <immintrin.h> #include <immintrin.h>
#endif #endif
#ifdef __cplusplus
extern "C" {
#endif
// //
// simd mappings // simd mappings
// //
@@ -137,14 +141,6 @@
} }
#endif #endif
// precomputed f32 table for f16 (256 KB)
// defined in ggml-cpu.c, initialized in ggml_cpu_init()
#ifdef __cplusplus
extern "C" float ggml_table_f32_f16[1 << 16];
#else
extern float ggml_table_f32_f16[1 << 16];
#endif
// On ARM NEON, it's quicker to directly convert x -> x instead of calling into ggml_lookup_fp16_to_fp32, // On ARM NEON, it's quicker to directly convert x -> x instead of calling into ggml_lookup_fp16_to_fp32,
// so we define GGML_CPU_FP16_TO_FP32 and GGML_CPU_FP32_TO_FP16 elsewhere for NEON. // so we define GGML_CPU_FP16_TO_FP32 and GGML_CPU_FP32_TO_FP16 elsewhere for NEON.
// This is also true for POWER9. // This is also true for POWER9.
@@ -1178,3 +1174,7 @@ static inline void __lzs_f16cx4_store(ggml_fp16_t * x, float32x4_t v_y) {
#define GGML_F32_ARR (GGML_F32_STEP/GGML_F32_EPR) #define GGML_F32_ARR (GGML_F32_STEP/GGML_F32_EPR)
#define GGML_F16_ARR (GGML_F16_STEP/GGML_F16_EPR) #define GGML_F16_ARR (GGML_F16_STEP/GGML_F16_EPR)
#endif #endif
#ifdef __cplusplus
}
#endif