0.0性能分析工具
https://developer.nvidia.com/performance-analysis-tools
0.1 CUDA-GDB
https://docs.nvidia.com/cuda/cuda-gdb/index.html
https://blog.csdn.net/u010794523/article/details/38657227
https://blog.csdn.net/sinat_28750977/article/details/69062708
http://book.51cto.com/art/201301/376309.htm
1.下载nsight
https://www.nvidia.com/object/nsight.html
https://developer.nvidia.com/nsight-visual-studio-edition
https://developer.nvidia.com/gameworksdownload#?dn=nsight-visual-studio-edition-5-6-0
- cuda toolkit 自带nvprof 评价性能,主要三点。
- occupancy
nvprof –metrics achieved_occupancy ./a.out - gld_throughput
nvprof –metrics gld_throughput - gdl_efficiency
nvprof –metrics gld_efficiency