CUDA 编程相关；tensorflow GPU 编程；关键知识点记录；CUDA 编译过程；NVCC

本文章主要是记录，cuda 编程过程中遇到的相关概念，名字解释和问题；主要是是用来备忘：

• cuda PTX ：并行线程执行（Parallel Thread eXecution，PTX）代码是编译后的GPU代码的一种中间形式，它可以再次编译为原生的GPU微码。
• 下面几条链接，是对nvcc 编译过程中，编译参数的解释：-gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61；本编译参数适用于 1080ti显卡；此种模式为cuda fatbinary模式，用于优化cudabin的执行；
• https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#virtual-architecture-feature-list 
• https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#gpu-feature-list 
• nvcc 编译过程理解：https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#gpu-compilation  
• 

• 上图是：Two-Staged Compilation with Virtual and Real Architectures，详细情况，请看上面链接；

• 

• 上图是：Just-in-Time Compilation of Device Code  

• https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#fatbinaries （CUDA fatbinaries 模式）
• https://www.cnblogs.com/zjutzz/p/10272424.html （nvcc 与 C++ 混合编译Demo, 较为优秀）

## cuda 11.6.0 安装总结 2022年11月10日20:10:38

===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-11.6/

Please make sure that
 -   PATH includes /usr/local/cuda-11.6/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-11.6/lib64, or, add /usr/local/cuda-11.6/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-11.6/bin
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 510.00 is required for CUDA 11.6 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
    sudo <CudaInstaller>.run --silent --driver

Logfile is /var/log/cuda-installer.log

保持更新，更多内容，请关注 cnblogs.com/xuyaowen;