2 questions 1 bug report

[kolpav]

• Do ML engineers ever change machines or they just retire?
• Can I just make stuff up in whitepapers? Theres always a possibility that you pip env kunda your machine wrong...

At first I was going to just dump my console here and leave as I wasted enough time on this but then I realized I don't know anyone who actually does ML for a living and that machine setup has to be a problem and I got curious. Even when I make one program detect and actually render using cuda its still not enough for the other one.

nvcc

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_May__3_18:49:52_PDT_2022
Cuda compilation tools, release 11.7, V11.7.64
Build cuda_11.7.r11.7/compiler.31294372_0

nvidia-smi

Tue Dec  6 15:41:52 2022       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 520.56.06    Driver Version: 520.56.06    CUDA Version: 11.8     |
|-------------------------------+----------------------+----------------------+
| 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  NVIDIA GeForce ...  Off  | 00000000:01:00.0  On |                  N/A |
|  0%   37C    P8    26W / 250W |    473MiB /  8192MiB |     19%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+

blender --debug-cycles (aka everything is as good as linux cuda install can be)

I1206 15:08:02.463270 162736 device.cpp:32] CUEW initialization succeeded
I1206 15:08:02.463402 162736 device.cpp:34] Found precompiled kernels
I1206 15:08:02.519917 162736 device.cpp:182] Device has compute preemption or is not used for display.
I1206 15:08:02.520030 162736 device.cpp:185] Added device "NVIDIA GeForce RTX 3070" with id "CUDA_NVIDIA GeForce RTX 3070_0000:01:00".

whisper call with cuda as a device 💥

Attempting to deserialize object on a CUDA '
RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.