read.md

• https://github.com/Tony-Tan/CUDA_Freshman

• https://github.com/Bruce-Lee-LY/cuda_hook

• https://muzhan.blog.csdn.net/article/details/120157690

• https://blog.csdn.net/inthat/article/details/119931358

• https://blog.csdn.net/leviopku/category_11340416.html

• https://zhuanlan.zhihu.com/p/622838762

• https://gitlab.com/nvidia/headers/cuda-individual/cudart

• https://blog.csdn.net/kunhe0512/article/details/128951128

• https://developer.nvidia.com/zh-cn/blog/cuda-virtual-address-management-cn/

• https://zhuanlan.zhihu.com/p/646091910

• https://segmentfault.com/a/1190000018460795

• CUDA 10.2 引入虚拟显存

• https://www.cnblogs.com/deny/p/16305744.html

const char *nvmlErrorString(nvmlReturn_t result) {
  const char *(*_entry)(nvmlReturn_t) = NVML_FIND_ENTRY(nvml_library_entry, nvmlErrorString);
  return _entry(result);
}


CUresult cuLaunchKernel_ptsz(CUfunction f, unsigned int gridDimX, unsigned int gridDimY, unsigned int gridDimZ, unsigned int blockDimX, unsigned int blockDimY, unsigned int blockDimZ, unsigned int sharedMemBytes, CUstream hStream, void **kernelParams, void **extra) { 
     rate_limiter(gridDimX * gridDimY * gridDimZ, blockDimX * blockDimY * blockDimZ);
}

readelf -Ws /usr/bin/nvidia-smi