Experiment output should separate slowdowns from broken runs

[sampsyo]

Currently, the "bad" category contains:

• Runs with high error (low-quality output).
• Runs that ended in an error (i.e., crashes).
• Runs with no speedup.

The latter category is worth investigating separately, as pointed out by @luluyi661, since it indicates useful results about the program. And the "bad" name is especially confusing.

[shichengfeng0]

Hi Adrian,

Just want to mention my thoughts to the group as you asked.

If we could simplify the set up process so that it would be weigh more easier for beginner to learn more about accept! For example, simple command line in terminal or some zipped files so that all we have to do is just double click:>.

Thanks, Chengfeng

2014-05-05 16:01 GMT-07:00 Adrian Sampson notifications@github.com:

Currently, the "bad" category contains:

• Runs with high error (low-quality output).
• Runs that ended in an error (i.e., crashes).
• Runs with no speedup.

The latter category is worth investigating separately, as pointed out by @luluyi661 https://github.com/luluyi661, since it indicates useful results about the program. And the "bad" name is especially confusing.

— Reply to this email directly or view it on GitHubhttps://github.com/uwsampa/accept/issues/13 .