Graphistry 2.35.9: No-Code Graph Analysis for your CSVs, HA for multi-GPUs, Run on A100 / CUDA 11.0

Posted by Graphistry Staff on February 21, 2021

Graphistry 2.35.9 is an exciting release. Anyone on your team can now use Graphistry: As soon as someone has a CSV/XLS, they can drop it in and start exploring without any coding. Developers will enjoy our new Files API for easier and faster data management. On the performance side, we are continuing to push on GPU support, this time around automatic HA for the multi-GPU layer and A100 / CUDA 11.0 support. Read on for our highlights and dig deeper with the official change log. The release is already available on our enterprise portal, up on Graphistry Hub, and going through cloud marketplace approvals.

Get started now! Try on a CSV

Explore the relationships in any CSV/XLS with GPU graph analytics.. and no code!

The file uploader lets anyone quickly drop a data export from another system and explore it with the full power of Graphistry… without any coding. It is great both for helping our existing users go faster and for enabling more members of a team get going. Checkout our animated example of quickly looking through event logs from a security honeypot.

A100 and CUDA 11.0 Support

Graphistry now ships as both 10.2 CUDA containers and 11.0 CUDA containers. The 11.0 supports enables running Graphistry on the new powerful A100 GPUs. AWS and Azure Marketplace users have already been on 11.x drivers, so this automatically upgrades their libraries to 11.0. In coming posts, we’ll share examples of leveraging new GPU features like GPU Direct.

Additional highlights

Graphistry has some additional updates such as upgrading to Caddy 2 and adding improved HA for dask-cuda multi-GPU / bigger-than-memory support. Of note, the File API introduced in 2.34 is now exposed for use by your own apps. The CSV uploader and latest PyGraphistry versions show examples of the new capabilities it enables. You can check the official change log for further updates.

Happy graphing!

Get started now! Try on a CSV