Feature #6614
Updated by Evgeny Novikov about 7 years ago
Sometimes scheduling can be done speculatively, e.g. we can try to solve more tasks with initially lower memory limits and re-solve them one or more times (there can be an option) if they meet memory limit. Since it depends on specifics of verification tasks to be solved (e.g. testing verification tasks require very few resources while complicated rule specifications likely require the most of them) it This is better to implement suitable for both native and cluster schedulers (VerifierCloud has this scheduling within VTG. itself) and thus generic scheduler infrastructure can take care of this.