Comment on page
Addressing Slow Code
Workspace runs on performant servers in the cloud. If you are on Workspace Starter, i.e. Workspace for free, your workspace gets 0.5 vCPUs and 4GB of RAM. If you are on Workspace Premium, your workspace gets 2 vCPU and 16GB of RAM. For more information on the difference between the different plans, check Pricing.
If the work you're doing is resource intensive, you may experience your code being slow to execute. This article lists the most common cases of code running slowly with suggestions on how you can address them.
- Try to reduce the amount of data you are plotting, for example, by aggregating over a certain dimension or just taking a sample of the data.
- Some plot types are notoriously resource intensive to generate, for example, swarm plots. Consider another plot type.
- Your cell might contain an infinite loop, check for while loops that might have a condition that's never false.
- The database you are querying might be slow to respond. In that case, consider restarting the database or making your database server more powerful.
- You may be querying a lot of rows, resulting a large data transfer to your notebook. In that case, you can try a couple of things:
- Make your query more specific so you fetch only the data you need.
- Limit the number of rows returned by your query until you're certain the rows contain the data you need, and only then perform a query without a limit.
- If you want to aggregate the result of the query, consider doing the aggregation in SQL rather than in Python, so a lot of computation already happens on the database.
- Some machine learning models benefit from GPUs when training. DataCamp Workspace does not provide GPU machines yet. As an alternative, you can train the model on your own computer and then upload the trained weights to your workspace.
- If you're on Workspace Starter and your workload can be paralllellized, consider upgrading to Workspace Premium to use more vCPUs for the training.
Last modified 6mo ago