📄 sgemm_common.h
字号:
/* * Copyright 1993-2008 NVIDIA Corporation. All rights reserved. * * NOTICE TO USER: * * This source code is subject to NVIDIA ownership rights under U.S. and * international Copyright laws. * * This software and the information contained herein is being provided * under the terms and conditions of a Source Code License Agreement. * * NVIDIA MAKES NO REPRESENTATION ABOUT THE SUITABILITY OF THIS SOURCE * CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" WITHOUT EXPRESS OR * IMPLIED WARRANTY OF ANY KIND. NVIDIA DISCLAIMS ALL WARRANTIES WITH * REGARD TO THIS SOURCE CODE, INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY, NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE. * IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY SPECIAL, INDIRECT, INCIDENTAL, * OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS * OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE * OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE * OR PERFORMANCE OF THIS SOURCE CODE. * * U.S. Government End Users. This source code is a "commercial item" as * that term is defined at 48 C.F.R. 2.101 (OCT 1995), consisting of * "commercial computer software" and "commercial computer software * documentation" as such terms are used in 48 C.F.R. 12.212 (SEPT 1995) * and is provided to the U.S. Government only as a commercial end item. * Consistent with 48 C.F.R.12.212 and 48 C.F.R. 227.7202-1 through * 227.7202-4 (JUNE 1995), all U.S. Government End Users acquire the * source code with only those rights set forth herein. */#include "cublasP.h"#if !defined(TILE_DIM)#error FILE_DIM not defined#endif__global__ void sgemm_main_gld_hw_na_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_na_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_ta_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_ta_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_na_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_na_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_ta_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_ta_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_hw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_gld_sw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_na_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_na_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_ta_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_ta_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_na_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_na_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_ta_nb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_ta_tb (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_hw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void sgemm_main_tex_sw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_na_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_na_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_ta_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_ta_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_hw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_na_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_na_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_ta_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_ta_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_tex_main_sw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_na_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_na_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_ta_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_ta_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_hw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_na_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_na_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_ta_nb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_ta_tb (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_sgemm_gld_main_sw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_na_nb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_na_tb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_ta_nb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_ta_tb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_tex_main_hw_ta_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_na_nb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_na_tb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_ta_nb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_ta_tb (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_na_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_na_tb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_ta_nb_fulltile (struct cublasSgemmParams parms);__global__ void fast_small_sgemm_gld_main_hw_ta_tb_fulltile (struct cublasSgemmParams parms);
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -