Issue with setgpu.py
Closed this issue · 3 comments
Hey friend, thanks for keeping this notebook up to date.
I've been using a similar notebook for almost a month without the setgpu.py trick.
Unfortunately I keep running on an issue when running blender with the setgpu.py
script. Here is what I get out on the terminal.
Blender 3.0.0 Alpha (hash 7d5e8d2bc149 built 2021-08-30 02:51:40)
Traceback (most recent call last):
File "/content/setgpu.py", line 6, in <module>
prefs.addons['cycles'].preferences.get_devices()
File "/content/blender-3.0.0-alpha+cycles-x.7d5e8d2bc149-linux.x86_64-release/3.0/scripts/addons/cycles/properties.py", line 1329, in get_devices
self.refresh_devices()
File "/content/blender-3.0.0-alpha+cycles-x.7d5e8d2bc149-linux.x86_64-release/3.0/scripts/addons/cycles/properties.py", line 1325, in refresh_devices
self.update_device_entries(_cycles.available_devices(device_type))
ValueError: Device "OPENCL" not known.
Read blend: /gdrive/MyDrive/Blender/primitives_02.blend
Notice the ValueError: Device "OPENCL" not known.
. Blender keeps running and I get a render out of the notebook, but it seems that is not using the GPU.
For more context, I'm using Google Colab Pro+ and I followed your advice to restart the notebook in order to get a P100
.
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02 Driver Version: 460.32.03 CUDA Version: 11.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 Tesla P100-PCIE... Off | 00000000:00:04.0 Off | 0 |
| N/A 39C P0 26W / 250W | 0MiB / 16280MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
Any advice? Thanks!
hey,
currently i do not have any idea about that, i keep this issue open if somebody came with a solution.
I use the free version of Coab, but did not face any problem even when i could access the P100.
Just assure that the Edit -> Preferences -> System be set on cuda.
and try the stable versions of colab (not 3.0.0)
Just tested Blender 3.0 with cycles X, use the following line without setgpu.py, that was really fast. but testing with K80, hope there is no problem with p100 either :
!sudo ./blender/blender -b '/content/sina2.blend' -noaudio -E 'CYCLES' -o '/content/sina' -s 28 -e 50 -a -- --cycles-device CUDA
Notebook Updated in 2022/03/10, and works properly.
so i close this issue.