Closed mratsim closed 3 years ago
This uses the new primitives mentioned in #92 for cached group checks.
Now we check at deserialization:
Instead of checking that at verification time. Since public keys are reused multiple time this is a significant performance improvement.
New perf
In #97 we had
So we cumulated a total of (4025212-4588537)/(4588537) = 12.2% perf improvement compared to the current BLST in Nimbus.
Closes #42 as well
okay so I broke the PR ... I didn't close it ... nor did I merge it
This uses the new primitives mentioned in #92 for cached group checks.
Now we check at deserialization:
Instead of checking that at verification time. Since public keys are reused multiple time this is a significant performance improvement.
New perf
In #97 we had
So we cumulated a total of (4025212-4588537)/(4588537) = 12.2% perf improvement compared to the current BLST in Nimbus.
Closes #42 as well