Closed DrPeterVanNostrand closed 5 months ago
It looks like the only longer running test where it makes a difference is test_ignored_release_filecoin_proofs
where the test time goes from 30 to 40mins. But if it turns out that the total run time is a problem, I guess we could always split tests into separate jobs.
It looks like the only longer running test where it makes a difference is
test_ignored_release_filecoin_proofs
where the test time goes from 30 to 40mins. But if it turns out that the total run time is a problem, I guess we could always split tests into separate jobs.
Since it's actually testing the 'non-interactivity' aspect now where it wasn't previously, I think it's important to keep it and not sweat the automated testing time.
The original "4" was chosen to have a similar run time to the Interactive PoRep times
Oh ok, that was a good idea. I didn't realize that was how the number of challenges was chosen so I wasn't even thinking about test times.
not sweat the automated testing time
Ok, I'm going to merge.
Since it's actually testing the 'non-interactivity' aspect now where it wasn't previously,
I compared it with the numbers that did already had the Non-interactive PoRep tests. Non-interactive with 4 PoReps add about 5mins to the job, increasing it to 26 adds about another 10mins. Though I agree it's not a deal breaker.
Changes the minimum number NI-PoRep challenges for test sectors from 4 to
ceil(12.8 * min_porep_challenges)
(i.e. 26 challenges) and the NI-PoRep partition count for test sectors from 2 to 13; bumping the partition count allows NI-PoRep test sectors to use their current Groth16 keys (i.e. 2 PoRep challenges per circuit/partition).