TODO Figure out how to run futhark-autotune on 217a17

For the ppop paper, this is how they tuned their programs:

FUTHARK_INCREMENTAL_FLATTENING=1 futhark/tools/futhark-autotune --futhark-bench=bin/futhark-bench --compiler=bin/futhark-opencl  --stop-after 2400  --test-limit 10000000 --bail-threshold=5000 benchmarks/OptionPricing.fut --save-json tunings/OptionPricing.json

But what are futhark-bench and futhark-opencl? It seems like they just correspond to futhark bench and futhark opencl, respectively.

It seems like they used their own benchmarks directory instead of futhark-benchmarks.

TODO Find out what benchmark programs and data-sets we should run auto-tuning benchmarks on

DONE Figure out what conference cosmin is thinking of submitting to

And adapt our paper to their requirements. Also make note of the page count.

This is it: http://tfp2021.org/

Draft papers and papers submitted for formal review are either submitted as extended abstracts (4 to 10 pages in length) or as full papers (20 pages).

Here's the result: ~/src/futhark-papers/ (magit-rev 88d82118)