vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

revert cuda changes as they are bugggy

Browse source
This commit is contained in:
Concedo 2023-08-09 22:36:41 +08:00
parent f8376c7e61
commit a07e6dd3ad
3 changed files with 392 additions and 681 deletions
Download patch file Download diff file Expand all files Collapse all files

5
Makefile
View file

@ -176,6 +176,11 @@ ifdef LLAMA_CUDA_KQUANTS_ITER
else else
NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2 NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2
endif endif
ifdef LLAMA_CUDA_MMQ_Y
NVCCFLAGS += -DGGML_CUDA_MMQ_Y=$(LLAMA_CUDA_MMQ_Y)
else
NVCCFLAGS += -DGGML_CUDA_MMQ_Y=64
endif # LLAMA_CUDA_MMQ_Y
#ifdef LLAMA_CUDA_CUBLAS #ifdef LLAMA_CUDA_CUBLAS
# NVCCFLAGS += -DGGML_CUDA_CUBLAS # NVCCFLAGS += -DGGML_CUDA_CUBLAS
#endif # LLAMA_CUDA_CUBLAS #endif # LLAMA_CUDA_CUBLAS

1066
ggml-cuda.cu
View file

File diff suppressed because it is too large Load diff

2
otherarch/ggml_v1.c
View file

@ -154,9 +154,11 @@ typedef double ggml_v1_float;
#undef bool #undef bool
#define bool _Bool #define bool _Bool
#else #else
#if !defined(__riscv)
#include <immintrin.h> #include <immintrin.h>
#endif #endif
#endif #endif
#endif
#ifdef __F16C__ #ifdef __F16C__