2016-04-18 14:01 GMT+02:00 Tagir F. Valeev <amae...@gmail.com>: > Hello! > > It was just a quick test not in the clean environment, so you should > not draw any conclusions from the error numbers. It's quite expected > that for limit = 2000 the performance is the same as I have 4 CPU > machine and 2000 is greater than 128*4. On the other hand, 200 is less > than 128*4, so this case is also improved (though not so drastically > as less limits).
Thanks for the clarification Tagir. Makes sense now. I missed the "commonPoolParallelism + 1" bit ... Regards, Stefan > > With best regards, > Tagir Valeev. >
