I was thinking if programmers can’t do situps for the entire length of their test suite, they and their code is out of shape.
There are plenty of legitimate reasons a test suite can take a long time to run (compiler test suites come to mind), so I wouldn’t put too much stock in this sentiment.
Yeah that’s a fair point. I feel this should still apply to the test suite for the section I’m working on.
If programmers want to get in shape, does that mean they should write slower tests?
The talk is how they speed up a compiler with a GPU and APL. The presenter take their optimization (parallelization of tree transformation with matrix operations) as an excuse to explain that we need to do simpler code, faster code and code that is more economical (requires less knowledge).
Great talk. Now I need to learn APL ;-)
See also https://scholarworks.iu.edu/dspace/handle/2022/24749 and https://github.com/Co-dfns/Co-dfns