cuda-memcheck reporting Program hit cudaErrorNotReady (error 34) due to

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi All.

I have been using cuda-memcheck to check any cuda code that I write. However, running cuda-memcheck on my code report a cudaErrorNotReady error when I use cudaStreamQuery().

The call returns “cudaErrorNotReady” which is fine as it means that Gpu stream is still performing work. However cuda-memcheck picks it up as an error which seems like a bug.

Program hit cudaErrorNotReady (error 34) due to "device not ready" on CUDA API call to cudaStreamQuery. 
=========     Saved host backtrace up to driver entry point at error
=========     Host Frame:/usr/lib/i386-linux-gnu/libcuda.so.1 [0x377c43]

Has anyone else experienced the same issue?

Thanks,

Akmal

2

I think it is probably expected behavior. cuda-memcheck is designed to indicate any time a CUDA runtime API call returns a code that is not cudaSuccess

If you’d like to see a change in CUDA behavior, the recommended approach is to file a bug, the instructions are linked to a sticky post at the top of this sub-forum.