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

step 1

Browse source
This commit is contained in:
jianyuzh 2023-12-29 17:25:40 +08:00 • committed by Meng, Hengyu
parent 0c00b4f654
commit ff83711055
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
ggml-sycl.cpp
View file

@ -122,7 +122,7 @@ static int g_ggml_sycl_debug=0;
//#define GGML_SYCL_DEBUG(...) (if(g_ggml_sycl_debug) printf(__VA_ARGS__))
#define GGML_SYCL_DEBUG(...) do{if(g_ggml_sycl_debug) printf(__VA_ARGS__);}while(0)
#define MIN_CC_DP4A 510 // minimum compute capability for __dp4a, an intrinsic for byte-wise dot products
#define MIN_CC_DP4A 610 // minimum compute capability for __dp4a, an intrinsic for byte-wise dot products
#define CC_VOLTA 700
#define CC_OFFSET_AMD 1000000
#define CC_RDNA2 (CC_OFFSET_AMD + 1030)
@ -246,7 +246,7 @@ static_assert(sizeof(sycl::half) == sizeof(ggml_fp16_t), "wrong fp16 size");
}
#endif // CUDART_VERSION >= 12000
[[noreturn]]
static void ggml_cuda_error(const char * stmt, const char * func, const char * file, const int line, const char * msg) {
fprintf(stderr, "CUDA error: %s: %s\n", stmt, msg);
fprintf(stderr, " in function %s at %s:%d\n", func, file, line);
@ -11043,7 +11043,7 @@ static void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1
GGML_SYCL_DEBUG("ggml_cuda_mul_mat ggml_cuda_op_mul_mat_vec_q path\n");
ggml_cuda_op_mul_mat(src0, src1, dst, ggml_cuda_op_mul_mat_vec_q, true);
} else {
GGML_SYCL_DEBUG("ggml_cuda_mul_mat ggml_cuda_op_dequantize_mul_mat_vec path\n");
// GGML_SYCL_DEBUG("ggml_cuda_mul_mat ggml_cuda_op_dequantize_mul_mat_vec path\n");
ggml_cuda_op_mul_mat(src0, src1, dst, ggml_cuda_op_dequantize_mul_mat_vec, false);
}
} else {
@ -11059,7 +11059,7 @@ static void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1
GGML_SYCL_DEBUG("ggml_cuda_mul_mat ggml_cuda_op_mul_mat_q path\n");
ggml_cuda_op_mul_mat(src0, src1, dst, ggml_cuda_op_mul_mat_q, true);
} else {
GGML_SYCL_DEBUG("ggml_cuda_mul_mat ggml_cuda_op_mul_mat_cublas path\n");
//GGML_SYCL_DEBUG("ggml_cuda_mul_mat ggml_cuda_op_mul_mat_cublas path\n");
ggml_cuda_op_mul_mat(src0, src1, dst, ggml_cuda_op_mul_mat_cublas, false);
}
}