The multithreading of the MSM builder was the culprit of sporadic failure of the test GoblinRecursionTests.Vanilla, so it was temporarily disabled. This lead to a small regression in the benchmark. If we ever get proof construction times down dramatically to where .2s matters, we should multithread this part of the code again.
The multithreading of the MSM builder was the culprit of sporadic failure of the test
GoblinRecursionTests.Vanilla
, so it was temporarily disabled. This lead to a small regression in the benchmark. If we ever get proof construction times down dramatically to where .2s matters, we should multithread this part of the code again.