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: add ggml-impl.h future notes

Browse Source 
Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-06-21 21:06:57 +08:00
parent 46227c61c9
commit 72965ea8b0
1 changed files with 13 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
ggml/src/ggml-impl.h
View File

@@ -28,17 +28,6 @@
#include <immintrin.h> #include <immintrin.h>
#endif #endif
#if defined(GGML_VXE)
#include <vecintrin.h>
#include <ggml-cpu/ggml-cpu-impl.h>
#endif
#if defined(GGML_NNPA)
#ifndef __NNPA__
#define __NNPA__
#endif // __NNPA__
#endif // GGML_NNPA
#ifdef __cplusplus #ifdef __cplusplus
extern "C" { extern "C" {
#endif #endif
@@ -429,6 +418,19 @@ GGML_API void ggml_aligned_free(void * ptr, size_t size);
#define GGML_FP32_TO_FP16(x) GGML_COMPUTE_FP32_TO_FP16(x) #define GGML_FP32_TO_FP16(x) GGML_COMPUTE_FP32_TO_FP16(x)
#elif defined(__NNPA__) #elif defined(__NNPA__)
/*
* Note: This functionality is ready for use, but the compiler macros
* defined for the s390x platform are defined in ggml-cpu while
* this file is 1 step behind, in ggml-src. I currently have no
* idea how to fix this, so I am leaving it as is.
*
* CMake chain: ggml -> ggml-src -> ggml-cpu
* ^^^^^^^^ ^^^^^^^^
* | | ggml-cpu defines the macros
* | | needed for s390x detection.
* | this file is here, where the s390x
* | detection macros are not defined.
*/
#define GGML_COMPUTE_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x) #define GGML_COMPUTE_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x)
#define GGML_COMPUTE_FP32_TO_FP16(x) ggml_compute_fp32_to_fp16(x) #define GGML_COMPUTE_FP32_TO_FP16(x) ggml_compute_fp32_to_fp16(x)
@@ -474,12 +476,6 @@ GGML_API void ggml_aligned_free(void * ptr, size_t size);
} }
static inline float ggml_compute_fp16_to_fp32(ggml_fp16_t h) { static inline float ggml_compute_fp16_to_fp32(ggml_fp16_t h) {
#ifdef __NNPA__
printf("%s: __NNPA__ is defined.\n", __func__);
#else
printf("%s: __NNPA__ is not defined.\n", __func__);
#endif
const uint32_t w = (uint32_t) h << 16; const uint32_t w = (uint32_t) h << 16;
const uint32_t sign = w & UINT32_C(0x80000000); const uint32_t sign = w & UINT32_C(0x80000000);
const uint32_t two_w = w + w; const uint32_t two_w = w + w;