Diagnostix can help you determine whether your testset covers the most common bugs. When you submit your testset (as a .tar file, containing the sources of your test programmes), Diagnostix will run each test programme against a number of faulty compilers.
When the output of a faulty compiler differs from the output the reference compiler would give, you have discovered that bug. The result is a score saying how many bugs your testset has found.
Of course there are many more possible bugs than the ones that we have built into Diagnostix, so the fact that your testset found all the Diagnostix bugs, does not mean it will find all bugs in your own compiler! We have only included the bugs that were most common in previous years.