why do we need cudaDeviceSynchronize(); in kernels with device-printf?

by

A kernel launch is asynchronous. This means it returns control to the CPU thread immediately after starting up the GPU process, before the kernel has finished executing.

So what is the next thing in the CPU thread here? Application exit.

At application exit, it’s ability to send output to the standard output is terminated by the OS.

Thus the output that is generated later by the kernel has nowhere to go, and you won’t see it.

On the other hand, if you use cudaDeviceSynchronize(), then the kernel is guaranteed to finish (and the output from the kernel will find a waiting standard output queue), before the application is allowed to exit.

More Related Contents:

  1. sending 3d array to CUDA kernel
  2. How do I choose grid and block dimensions for CUDA kernels?
  3. Reduce matrix rows with CUDA
  4. How is CUDA memory managed?
  5. How to measure the inner kernel time in NVIDIA CUDA?
  6. nvidia-smi Volatile GPU-Utilization explanation?
  7. How do I use Nvidia Multi-process Service (MPS) to run multiple non-MPI CUDA applications?
  8. CUDA how to get grid, block, thread size and parallalize non square matrix calculation
  9. Cuda kernel returning vectors
  10. Use of cudamalloc(). Why the double pointer?
  11. Cuda C – Linker error – undefined reference
  12. How can I access a graphics card’s output directly?
  13. Why is NVIDIA Pascal GPUs slow on running CUDA Kernels when using cudaMallocManaged
  14. Timing CUDA operations
  15. How does CUDA assign device IDs to GPUs?
  16. How do I check if an integer is even or odd? [closed]
  17. What is exactly the base pointer and stack pointer? To what do they point?
  18. Where is the C auto keyword used?
  19. What is the benefit of terminating if … else if constructs with an else clause?
  20. In C, why can’t an integer value be assigned to an int* the same way a string value can be assigned to a char*?
  21. Is there a tool to know whether a value has an exact binary representation as a floating point variable?
  22. Why do I require multiple EOF (CTRL+Z) characters?
  23. Why my kernel log is not showing the latest output?
  24. Why does this intentionally incorrect use of strcpy not fail horribly?
  25. What happens when I assign long int to int in C?
  26. Check if User Inputs a Letter or Number in C
  27. Count number of digits – which method is most efficient?
  28. Using getchar() on c gets the ‘Enter’ after input [duplicate]
  29. What exactly is -fno-builtin doing here?
  30. Pass va_list or pointer to va_list?

Leave a Comment