Looks like Databricks has a serverless notebooks mode which is slightly different to the cluster based approach we use currently. We should make sure we have any differences documented.
https://docs.databricks.com/aws/en/compute/serverless/gpu
Forwarded from @jrbourbeau
Databricks has a new serverless compute product. Long-term that will end up replacing their "classic" Spark compute clusters. I tried a basic pip install cudf in a serverless notebook and it seemed to work out of the box

Looks like Databricks has a serverless notebooks mode which is slightly different to the cluster based approach we use currently. We should make sure we have any differences documented.
https://docs.databricks.com/aws/en/compute/serverless/gpu
Forwarded from @jrbourbeau