Hi all,
Whenever I try to move my tensors or model to the GPU using either the .cuda()
or .to('cuda')
method, the kernel just freezes and has to be terminated to be used again.
I’ve looked through several other related issues and they are either extremely old (circa 2017) or their solutions were that they had an incompatible version of cuda running - I think none were very useful.
Here is my environment details:
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 435.21 Driver Version: 435.21 CUDA Version: 10.1 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce RTX 207... Off | 00000000:01:00.0 Off | N/A |
| N/A 47C P8 5W / N/A | 303MiB / 7982MiB | 5% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1768 G /usr/lib/xorg/Xorg 135MiB |
| 0 2036 G /usr/bin/gnome-shell 116MiB |
| 0 2589 G ...uest-channel-token=11750413998548151078 49MiB |
+-----------------------------------------------------------------------------+
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Fri_Feb__8_19:08:17_PST_2019
Cuda compilation tools, release 10.1, V10.1.105
And I just followed the basic installation instructions on the website:
conda install pytorch torchvision cudatoolkit=10.1 -c pytorch
Any ideas? Thanks!