Hello,
When I stopped the running by clich ctrl+z then run this command again:
python -m stretch.app.ai_pickup --use_llm
I still got this error as shown in screenshot:
Do i need every time to run: nvidia-smi then kill -9 [ ] ?
Hi, @Mallak_Alqaisi,
In this case, it’s better to stop the process using Ctrl+C instead of Ctrl+Z. Ctrl+C properly terminates the program and releases the GPU memory.
When using Ctrl+Z, the process is only paused (not stopped), so it continues holding onto GPU memory in the background. That’s why you run into the CUDA out-of-memory error when you try to run the command again.
If Ctrl+C doesn’t work and you end up using Ctrl+Z, then yes, you will need to manually stop the process by finding its PID (e.g., via nvidia-smi) and running:
kill -9 <PID>
Best,
Jason
This is what i got when i use ctrl+c and it didn’t works shown in the screenshot, Is there another way can i use to stop running instead of ctrl+z ?
If Ctrl+C doesn’t work, there isn’t really another way besides killing the process.
You can do:
kill %1
This will stop the latest job. If needed, you can force it with:
kill -9 %1