Elliot's TODOs

[ronawho]

Higher Priority:

• [x] Gather performance results for UPC, MPI, and Chapel for all problem sizes and commit results (done in https://github.com/npadmana/DistributedFFT/pull/57)
• [x] Figure out next steps for running NPB-FT as part of Chapel's performance suite -- https://github.com/npadmana/DistributedFFT/issues/42
• [x] Create/update upstream issues for providing low-level comm interfaces and for improving performance of the useCommPrimitives=false case (done in https://github.com/npadmana/DistributedFFT/issues/59)

Medium Priority:

• [ ] Figure out performance differences between intel and gcc/llvm backends -- https://github.com/npadmana/DistributedFFT/issues/25

Lower Priority:

• [x] Figure out Swan vs Crystal timing differences -- https://github.com/npadmana/DistributedFFT/issues/39
• [x] Gather some InfiniBand numbers -- https://github.com/npadmana/DistributedFFT/issues/49

Misc:

• [x] Try using an explicit inner coforall instead of multiple foralls to verify task-creation time is negligible (I have verified task creation time is minimal in full SIM code.)
• [x] Each trial has to set up the plans (currently run with FFTW_MEASURE.) Could we set up once and cache/reuse plans (done in https://github.com/npadmana/DistributedFFT/pull/56)

[npadmana]

I think the last of the miscellaneous issues is now done as well! :+1: