@@ -13,7 +13,9 @@ You can open the notebook in Google Colab from here, and enable GPU :
## Run the notebook on DTU Gbar
You can run the notebook on Gbar on an interactive GPU node. First get to a terminal on the cluster either ssh, ThinLinc client, or browser-based ThinLinc [thinlinc.gbar.dtu.dk](thinlinc.gbar.dtu.dk). Open the terminal (command line).
You can run the notebook on Gbar on an interactive GPU node.
First get to a terminal on the cluster, either with ssh, ThinLinc client, or a browser-based ThinLinc [thinlinc.gbar.dtu.dk](thinlinc.gbar.dtu.dk). Open the terminal (command line).
Log onto an interactive GPU node such as `voltash` by running the following command on the command line:
...
...
@@ -25,13 +27,13 @@ To set up a Python environment, you can use the script `env02506.sh` which we pr
The first time you run this, the script will create a new python environment called `env02506` and activate it.
Then you should install the packages needed which include:
`pip install torch torchvision`
`pip install Pillow`
Then you should install the packages:
`pip install notebook`
```
pip install torch torchvision
pip install Pillow
pip install notebook
```
Now you are good to go. You can navigate to the folder that you wish to store the code, download the notebook from the link above, and open a jupyter notebook by typing the following in the command line:
...
...
@@ -40,7 +42,11 @@ Now you are good to go. You can navigate to the folder that you wish to store th
## Use QIM platform
We are working on establishing an easier way of using Gbar functionality which does the steps described above automatically. This is still experimental.
Go to [https://platform.qim.dk/](https://platform.qim.dk/). Log on. Under Tools start Jupyter launcher. Type in your DTU credentials (it still needs to be done twice). Under basic config QIM environment choose `qim3d` (it has the largest number of packages installed). Under advanced config choose HPC que called `gpuqim`. Tick the checkbox Reset SSH tunnel. Click the button `Start jupyter server`.
First go to QIM platform [https://platform.qim.dk/](https://platform.qim.dk/) and log on using your DTU credentials.
Under Tools menu, choose Jupyter launcher. In the new window yype in your DTU credentials (needs to be done twice). Under basic config QIM environment choose `qim3d` (it has the largest number of packages installed). Under advanced config choose HPC queue called `gpuqim`. Tick the checkbox Reset SSH tunnel. Click the button `Start jupyter server`.
The platform shold launch your job and a button with `Open Jupyter` should appear, leading you to a jupyter server running in your home directory on the Gbar.
