Tyson Hilmer
Reputation: 771
Do failed CUDA memory allocations require a cudaDeviceReset call?
When a CUDA runtime API call for GPU memory allocation fails, is it necessary to reset the context using a call to cudaDeviceReset()?
Upvotes: 0
Views: 28
Answers (0)
Related Questions
- How can I flush GPU memory using CUDA (physical reset is unavailable)
- How to get the CUDA version?
- What is the canonical way to check for errors using the CUDA runtime API?
- Different CUDA versions shown by nvcc and NVIDIA-smi
- How to tell if tensorflow is using gpu acceleration from inside python shell?
- How do I select which GPU to run a job on?
- How do CUDA blocks/warps/threads map onto CUDA cores?
- Do I need provide Gpu context when creating unified memory?
- cuda context creation and resource association in runtime API applications
- How is CUDA memory managed?