ggml/src/ggml-cuda/convert.cuh


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

#include "common.cuh"

#define CUDA_DEQUANTIZE_BLOCK_SIZE 256

template<typename T>
using to_t_cuda_t = void (*)(const void * __restrict__ x, T * __restrict__ y, int64_t nrows, int64_t n_per_row, cudaStream_t stream);

typedef to_t_cuda_t<float> to_fp32_cuda_t;
typedef to_t_cuda_t<half> to_fp16_cuda_t;
typedef to_t_cuda_t<nv_bfloat16> to_bf16_cuda_t;

to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type);

to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type);

to_bf16_cuda_t ggml_get_to_bf16_cuda(ggml_type type);