Hello Ludovic, On Sun, 21 Jun 2015 08:20:52 +0200 Ludovic Gasc <[email protected]> wrote:
> Hi Paul, > > It's pretty interesting. Thanks. > > I've a question: do you test your daemons with wrk instead of ab? > Could you test with the values: wrk -t8 -c256 -d1m ? I did not, for the same old reason - lack of time. uasyncio project started more than a year ago, and for all this time, I didn't formally announce it because I wasn't sure I did enough homework on it. So, I decided to just "throw it over the wall" to invite wider community review/criticism. And to answer your question, I selected Apache Bench because it's a well-known, easily accessible benchmarking tool. I used another tool, Boom (https://github.com/tarekziade/boom), but not for performance, rather correctness testing (ensuring that data received is actually what's expected): https://github.com/pfalcon/micropython-lib/blob/master/uasyncio/benchmark/test-boom-heavy.sh I'll add testing with wrk to my queue, but it may take a while (because there're bunch of other things I'm working on for MicroPython besides uasyncio). > I'm interested in to integrate uasyncio and picoweb in > FrameworkBenchmarks: https://www.techempower.com/benchmarks/ I read about them, but have to admit I didn't have a chance to review them in detail ;-(. But I re-did and included my tests comparing uasyncio & asyncio exactly to invite 3rd-party independent testing, so I'd appreciate if you could add it to your queue either. MicroPython+uasyncio should be relatively easy to install, one area where we may be lacking is documentation. But then to improve, we'd need independent feedback on that either. So, if/when you get to it, feel free to ask me any questions/give any feedback (by whatever way you like, e.g. via https://github.com/micropython/micropython/issues/new). Thanks, Paul > > Regards. > > Ludovic Gasc (GMLudo) > http://www.gmludo.eu/ -- Best regards, Paul mailto:[email protected]
