Can someone describe all the forms of parallelism present across the entire CMU latedays cluster?

1. SIMD (AVX2, AVX512 support)
2. Multithreading, Hyperthreading
3. Multiple Cores (Two, six-core Xeon e5-2620 v3 processors - 2.4 GHz and One, 60 cores Xeon Phi 5110p - 1 GHz)
4. Hybrid CPU+GPU - Tesla K40

To round that off, I'd say: 5. Implicit ILP in HW.

Also we have the coprocessor the Xeon Phi

Don't forget that the cluster itself provides parallelism across multiple computers/servers