fast_sgemm.cu

来自「Nividia提供的CUDA的BLAS库源码」· CU 代码 · 共 741 行 · 第 1/2 页

CU
741
字号
#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           1#include "sgemm.h"}__global__ void fast_sgemm_tex_main_sw_ta_tb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           1#include "sgemm.h"}__global__ void fast_sgemm_tex_main_sw_na_nb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            0#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           1#include "sgemm.h"}__global__ void fast_sgemm_tex_main_sw_na_tb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            0#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           1#include "sgemm.h"}__global__ void fast_sgemm_tex_main_sw_ta_nb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           1#include "sgemm.h"}__global__ void fast_sgemm_tex_main_sw_ta_tb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           1#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_na_nb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 0#define TRANSA            0#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_na_tb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 0#define TRANSA            0#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_ta_nb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 0#define TRANSA            1#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_ta_tb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 0#define TRANSA            1#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_na_nb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 0#define TRANSA            0#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_na_tb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 0#define TRANSA            0#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_ta_nb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 0#define TRANSA            1#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_hw_ta_tb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 0#define TRANSA            1#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_na_nb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 1#define TRANSA            0#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_na_tb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 1#define TRANSA            0#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_ta_nb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_ta_tb (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   0#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_na_nb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            0#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_na_tb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            0#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_ta_nb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            0#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}__global__ void fast_sgemm_gld_main_sw_ta_tb_fulltile (struct cublasSgemmParams parms){#undef  FULL_TILES_ONLY#undef  USE_MIXED_STEPPER#undef  TRANSA#undef  TRANSB#undef  FAST_IMUL#undef  USE_TEX#define FULL_TILES_ONLY   1#define USE_MIXED_STEPPER 1#define TRANSA            1#define TRANSB            1#define FAST_IMUL         1#define USE_TEX           0#include "sgemm.h"}

⌨️ 快捷键说明

复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?