Due to late-discovered performance regressions, the PR #1779 had to be reverted in full (#2022). Nevertheless, from the scheduler design standpoint the patch is certainly an improvement. We need therefore to root cause the observed regressions and understand what is the best way to address those.
Due to late-discovered performance regressions, the PR #1779 had to be reverted in full (#2022). Nevertheless, from the scheduler design standpoint the patch is certainly an improvement. We need therefore to root cause the observed regressions and understand what is the best way to address those.