Can't find any numbers in the linked thread with the patches. Surely some preliminary benchmarking must have been performed that could tell us something about the real world potential of the change?
How would you benchmark something like this? Run multiple processes concurrently and then sort by total run time? Or measure individual process wait time?
I guess both make sense, and a lot of other things (synthetical benchmarks, microbenchmarks, real-world benchmarks, best/average/worst case latency comparison, best/average/worst case throughput comparison...)