On Fri, Apr 08, 2022 at 03:36:30PM +0200, Miroslav Zagorac wrote: > Hello Willy, > > today I tested the operation speed of the opentracing module on branch 2.4 > and 2.5 several times and unfortunately I can't confirm what I wrote > yesterday (that the operation speed is almost doubled). > > The result I get today is almost the same speed. > > I don't know what the problem is, maybe I did something wrong yesterday > or Linux decided to do something different today (probably the first > one).
OK, this happens, do not worry. At least what matters is that there is no performance regression. > This means that perhaps the next paragraph should be removed from the > description of patch 0014: > > "In terms of code execution speed, this way of using context variables > is significantly faster than the previous one (when variables with names > were stored instead of hash). For example, in the worst case, when the > rate-limit is set to 100%, the operating speed is almost doubled (this > refers to the first result in the 'addons/ot/test/README-speed-ctx' > file)." Will do, thank you! Willy