For most tests with a ref case, I believe the standard is to have the "bare" case name be the main case, in which comparisons with baselines (etc.) occur. However, this is reversed for the SSP test.
This lack of consistency is a problem for users of the test, and also for tools that post-process test results, such as baseline_gen_comp.
I propose that the SSP test be reworked to switch the meaning of the ref1 case and the "bare" case.