How to Request changes in JupyterLab compute options
Flywheel allows sites to configure the choices available for compute in the Jupyter Notebook. This allows sites to customize the NVIDIA GPU types, CPU cores, or RAM available for use in Notebooks. If you would like to make a change to your site's configuration, you can submit a support request.
Note
Jupyter Notebooks is an optional Module provided by Flywheel for Machine Learning in Medical Imaging. It is only available for cloud deployed sites. If you would like more information about this feature, contact your Flywheel Account Executive either directly or through Flywheel support by submitting a ticket or emailing us at support@flywheel.io.
Default Configuration:
For Reference, the default setup at implementation has the following 4 slots (may differ slightly based on cloud vendor):
- "General Purpose" 4 CPU, 14GB RAM
- "Compute" 7 CPU, 27GB RAM
- "Compute-GPU" 7 CPU, 24GB RAM, 1GPU (an NVIDIA T4 w 16GB RAM)
- "Larger-Compute-GPU" 15CPU, 90GB RAM, 1GPU (an NVIDIA T4 w 16GB RAM)
This configuration provides both general purpose compute with no GPU for basic notebooks tasks all the way to a 15CPU server with large RAM and a GPU. Also, having mid-level "Compute" and "Compute-GPU" options allows users to test their notebooks without GPU and then switch to a GPU server to do the run, in order to save cost. The default GPU is an NVIDIA T4 (with 16GB RAM), which is typically a low-end, low cost option. A higher performance GPU can be requested such as a NVIDIA A100 with 80GB RAM.
Example of a higher GPU performance option for slot 4:
- "Larger-Compute-GPU" 24CPU, 100GB RAM, 1GPU (an NVIDIA A100 w 80GB RAM)
Tip
See Determining GPU Driver to learn how to determine existing GPUs in a JupyterLab Server instance.
Request info
Site Admins should create a Support Request and provide:
-
The slot (or slots) you would like to modify. The recommendation is to keep the first one or two slots for the 'smaller' compute choices without GPU, so that users always have an option for a low cost compute. Currently Flywheel supports 4 slots.
-
For each slot then provide:
- The type of GPU you would like to add. Please provide a primary and alternative option, since it may not be possible to get the first choice from the cloud vendor. Only one GPU is supported in each slot (no multi-GPUs on a server).
- The GPU RAM you would like, this will help us find an appropriate GPU to fit your needs.
- The CPU and number of cores (can be a SKU or family provided by the cloud vendor)
- The RAM desired (must be less than 100GB)
- The name of the slot, if you would like to change it.
- The Flywheel support team will then reach out to you with options available to confirm and then make the change.
Contact Support
Site Admins should contact Flywheel support by submitting a ticket or emailing us at support@flywheel.io.