Stream Work to Do

[vsoch]

• Investigate Parallel Cluster not showing up on plots
• Each analysis needs to be redone with the data grouped by how it was run.
• Add additional "type" that was run with threads / threading enabled / with SMT
• Make better life choices about plot colors 🎨
• GPU units should be GB/s not MB/s 📐
• Add slack writeup on CUDA_VISIBLE_DEVICES here

Add note to readme that with the CUDA_VISIBLE_DEVICES we likely have over-subscription for the GPUs

Update: I've dumped the slack history, and we are planning to re-run stream after we hear from Heidi