telamon icon indicating copy to clipboard operation
telamon copied to clipboard

Published 20 hours ago • verified publisher icon ulysseB

Improve the accuracy and cost of evaluations on the GPU

Open ulysseB opened this issue 7 years ago • 1 comments

We currently run each candidate 20 times on the GPU if it has an execution time within 3X of the current best candidate. Better accuracy and performance could be achieved by exiting after a few evaluations if the performance is too bad and evaluating more if the performance is close the best candidate (e.g <3%).

ulysseB avatar Jul 30 '18 12:07 ulysseB

In particular, the evaluation of the performance model is too slow.

ulysseB avatar Aug 22 '18 10:08 ulysseB