Add a mode that load nvidia.ko only so that we display on intel and do cuda deep learning in NV GPU
Bug #1824282 reported by
Yuan-Chen Cheng
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
nvidia-prime (Ubuntu) |
Fix Released
|
Undecided
|
Alberto Milone |
Bug Description
Per current NV and intel GPU driver, if we load nvidia.ko only, we can do display on intel GPU and leave the NV to deep learning for a graphic workstation.
I proposed to add a mode might call "intel+nv_cuda" for this mode.
Test procedure:
1. install the new package on a platform with both intel and NV graphic card.
2. $ sudo prime-select intel+nv_cuda; sudo reboot
3. $ nvidia-smi # it will show the nvidia GPU status and we will found that no GUI process use any NV GPU memory.
Changed in nvidia-prime (Ubuntu): | |
assignee: | nobody → Alberto Milone (albertomilone) |
Changed in nvidia-prime (Ubuntu): | |
status: | New → Triaged |
To post a comment you must log in.
the last commit in
https:/ /code.launchpad .net/~ycheng- twn/junk/ +git/nvidia- prime/+ ref/ubuntu/ devel
have the code that I mentioned.