exo
exo copied to clipboard
Not working on WSL (debian) on Windows 11
Just wondering if this is expect to work on Windows?
I wanted a NVIDIA GPU and it came in a gaming PC (470 Super)
I quickly realised this project wouldn't work natively on Windows so I tried it on WSL using debian:
Linux W11 5.15.167.4-microsoft-standard-WSL2 #1 SMP Tue Nov 5 00:21:55 UTC 2024 x86_64 GNU/Linux
After a LOT of time trying to get the NVIDIA cuda packages installed I eventually managed:
python --version
Python 3.12.2
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Tue_Oct_29_23:50:19_PDT_2024
Cuda compilation tools, release 12.6, V12.6.85
Build cuda_12.6.r12.6/compiler.35059454_0
nvidia-smi
Fri Dec 6 19:43:15 2024
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 560.35.05 Driver Version: 556.12 CUDA Version: 12.5 |
|-----------------------------------------+------------------------+----------------------+
| 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 RTX 4070 ... On | 00000000:01:00.0 On | N/A |
| 30% 21C P8 7W / 220W | 569MiB / 12282MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 193 G /Xwayland N/A |
+-----------------------------------------------------------------------------------------+
Resulting weights from Softmax:
0.0000000 0.0000000 0.0000000 1.0000000 0.0000000 0.0000714 0.0000000 0.0000000 0.0000000 0.0000000
Loading image data/five_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000008 0.0000000 0.0000002 0.0000000 1.0000000 0.0000154 0.0000000 0.0000012 0.0000006
Result of classification: 1 3 5
Test passed!
I found an issue with the NVIDIA driver as the line in exo/topology/device_capabilities.py:
I changed:
#gpu_raw_name = pynvml.nvmlDeviceGetName(handle).upper()
gpu_raw_name = "NVIDIA GEFORCE RTX 4070 SUPER"
Then exo runs and initially looks good just with this single node. However, with the first small model a simple 'hi' reslted in a generated response of
HowQuestion of you you you you you you you you you you you you you you you you you you you you you you you you you you
and apalling performance before exo crashed.
Any ideas