> On 7 September 2017 at 15:46, Dmitry Dolgov <9erthali...@gmail.com> wrote: > > It seems that for this particular workload it was about 20-25% slower.
Hmm...looks like I provided misleading data, sorry. The numbers from previous email are correct and I'm able to reproduce them, but surprisingly for me only for one particular situation when `log_statement=all; log_destination=stderr` and stderr is sent to a console with a tmux session running in it (and apparently it has some impact on the performance, but I'm not sure how exactly). In all other cases (when stderr is sent to a plain console, /dev/null, or we send logs to a file) numbers are different, and the difference between patch and master is quite less significant. average latency: clients patch master 10 0.321 0.286 20 0.669 0.602 30 1.016 0.942 40 1.358 1.280 50 1.727 1.637