mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2025-01-03 15:24:35 +00:00
25 lines
979 B
Plaintext
25 lines
979 B
Plaintext
#pragma once
|
|
|
|
#include "common.cuh"
|
|
#include "mmq.cuh"
|
|
|
|
#include <cstdint>
|
|
|
|
#define CUDA_QUANTIZE_BLOCK_SIZE 256
|
|
#define CUDA_QUANTIZE_BLOCK_SIZE_MMQ 128
|
|
|
|
static_assert(MATRIX_ROW_PADDING % CUDA_QUANTIZE_BLOCK_SIZE == 0, "Risk of out-of-bounds access.");
|
|
static_assert(MATRIX_ROW_PADDING % (4*CUDA_QUANTIZE_BLOCK_SIZE_MMQ) == 0, "Risk of out-of-bounds access.");
|
|
|
|
typedef void (*quantize_cuda_t)(
|
|
const float * x, void * vy, const int64_t kx0, const int64_t kx1, const int64_t channels, const int64_t kx0_padded,
|
|
const ggml_type type_x, cudaStream_t stream);
|
|
|
|
void quantize_row_q8_1_cuda(
|
|
const float * x, void * vy, const int64_t kx0, const int64_t kx1, const int64_t channels, const int64_t kx0_padded,
|
|
const ggml_type type_x, cudaStream_t stream);
|
|
|
|
void quantize_mmq_q8_1_cuda(
|
|
const float * x, void * vy, const int64_t kx0, const int64_t kx1, const int64_t channels, const int64_t kx0_padded,
|
|
const ggml_type type_x, cudaStream_t stream);
|