You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It is often a problem that CMS runs all files for each submission. In most OI styles problems, there is usually a subtask where very worse complexity solution can pass. Many participants usually code that first and submit. But since this is run on all the files, the later subtasks take up huge unnecessary judging time that it often jams up the server (low spec servers), especially in the beginning of the contest, sometimes even crash it (from experience)!
The Solution
For minimum in subtask scoring, stop evaluating more files once at least one of the file gets score 0. (This may also be related to feedback level full / restricted).
It may not be the default, but having an option for this will be very helpful!
Although it may be difficult to implement, since in CMS evaluation is done in parallel, in contrary to other online judges that stop on first wrong answer.
Alternative
Current way to overcome this problem is to just have small number of input files. Which can be a bad thing for some problems.
The text was updated successfully, but these errors were encountered:
The Problem
It is often a problem that CMS runs all files for each submission. In most OI styles problems, there is usually a subtask where very worse complexity solution can pass. Many participants usually code that first and submit. But since this is run on all the files, the later subtasks take up huge unnecessary judging time that it often jams up the server (low spec servers), especially in the beginning of the contest, sometimes even crash it (from experience)!
The Solution
For minimum in subtask scoring, stop evaluating more files once at least one of the file gets score 0. (This may also be related to feedback level full / restricted).
It may not be the default, but having an option for this will be very helpful!
Although it may be difficult to implement, since in CMS evaluation is done in parallel, in contrary to other online judges that stop on first wrong answer.
Alternative
Current way to overcome this problem is to just have small number of input files. Which can be a bad thing for some problems.
The text was updated successfully, but these errors were encountered: