Dear Chameleon users,

 

Hey! It's your friendly neighborhood Chameleon changelog, back from a long vacation. We'll be sending monthly changelogs out all year to keep you informed on the latest and greatest developments on the testbed. Here's what we've put together for you this month:

 

General availability for CHI@NU! The CHI@NU site has been available for Early Users on a “by request” basis for the last six months or so while we have been working out methods of adding an associate site to Chameleon easily. We are happy to announce that the Chameleon Associate Site at Northwestern is now fully integrated into the rest of the testbed and can be used by any Chameleon project or user. Last year we highlighted the Mellanox ConnectX-5 100GE network cards available on nodes there. The CHI@NU site offers more opportunity for advanced and high-performance networking experiments, and can additionally be stitched to the existing CHI@TACC and CHI@UC sites for dedicated wide-area experiments. Interested in becoming an associate site yourself? Contact us or learn more about CHI-in-a-Box, our packaged deployment of Chameleon that you can deploy yourself, here!

 

New Cascade Lake hardware is now available at TACC. Late last year, we announced the preliminary installation of one rack of Intel Cascade Lake CPUs. We are happy to announce that three more racks, for a grand total of 56 nodes, are now available at CHI@TACC for your experimentation. These new "Cascade Lake R" compute nodes include the Xeon Gold 6240R Cascade Lake Scalable Performance Refresh CPUs with support for 96 threads! As usual, you can find all of the hardware details in the Chameleon hardware catalogue!

 

Orchestrate your experiments with Chameleon's official Python library. Particularly enterprising users may have discovered that, while there has been a Python library called "python-chi" for some time, its API was still under development and we never had proper documentation for it. We're happy to announce that python-chi is now ready for prime time, complete with full documentation and some helpful usage examples to get you started. With the library, you can make reservations, launch new bare metal instances, and manage network abstractions for networking experiments. Let us know how you like it, and please file issues for any bugs you happen to encounter.

 

CUDA gets turned up to 11. We've released a new suite of base images for CentOS and Ubuntu with the latest CUDA 11 release for all of your GPU experimentation. These images should be launched on our "gpu_*" class nodes, like the V100s we recently added to CHI@UC.

 

See all publications using Chameleon! We're entering phase three of the testbed and have collected a list of the publications that have cited the testbed over the years, to give you a sense of how everybody is using Chameleon to do interesting research. As a reminder, please cite Chameleon when you are publishing! Refer to our official citation guidance.

 

We're hiring! The Nimbus team is looking for both a junior and senior cloud developer to help us in Phase 3. Check out our jobs page for more info on the positions, and spread the word!

 

Until next time!