GPUDirect for video could not work on Tesla V100

AI & Data Science Deep Learning (Training & Inference) cuDNN
1

安装NVIDIA GPU Direct for Video SDK 安装包，采用fifoCUDA示例程序，调用dvpInitCUDAContext(DVP_DEVICE_FLAGS_SHARE_APP_CONTEXT)函数返回-1，
示例程序的打印信息：
CUDA fifo example v0.3
DVP_ERROR: Initialization failed
Fail on line 877
DVP Failed with status FFFFFFFF
DVP_ERROR: Initialization failed
Fail on line 885
DVP Failed with status FFFFFFFF
WARNING!!! Testing data enabled, so may be slow perf.
Assertion failed: 0, file extsdk.cpp, line 997

4

here are potential solutions to the issue you’re facing with the NVIDIA GPU Direct for Video SDK and the fifoCUDA sample program, specifically regarding the dvpInitCUDAContext function returning -1:

  1. Check System Requirements:
  • Ensure your system meets all the required hardware and software specifications for the NVIDIA GPU Direct for Video SDK and the fifoCUDA sample program.
  1. Verify Installation:
  • Confirm that the installation of the NVIDIA GPU Direct for Video SDK and the fifoCUDA sample program was completed correctly without any errors.
  1. Debugging Process:
  • Utilize debugging tools to analyze error logs and the application flow, which may help to identify the cause of the issue with the dvpInitCUDAContext function.
  1. Update Drivers and Software:
  • Ensure all GPU drivers and the CUDA Toolkit are up to date. Compatibility issues or bugs in outdated versions could lead to errors.