cfbolz changed the topic of #pypy to: #pypy PyPy, the flexible snake https://pypy.org | IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end and https://libera.irclog.whitequark.org/pypy | the pypy angle is to shrug and copy the implementation of CPython as closely as possible, and staying out of design decisions
pbsds3 has quit [Quit: The Lounge - https://thelounge.chat]
itamarst has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #pypy
[Arfrever] has quit [Read error: Connection reset by peer]
_whitelogger has joined #pypy
[Arfrever] has joined #pypy
Dejan has joined #pypy
itamarst has joined #pypy
jcea has joined #pypy
<itamarst> after how many calls to a function does JITing kick in?
<itamarst> (if I want to warm up a benchmark so it's benchmarking the JITed code)
<cfbolz> itamarst: loops are traced after 1039 iterations, functions after 1619 enters
<itamarst> but... looks like it's adjustable?
<itamarst> would that make sense to do orwould it distort results
<Alex_Gaynor> Probably it would distort it relative to real world usage
<Alex_Gaynor> If you find params that help in the real world, you likely want to use them in your benchmark though
<itamarst> :/
<itamarst> this is gonna be fun
<itamarst> codspeed.io uses cachegrind, which makes everything super slow
<itamarst> so 2700 iterations is gonna take a while
<cfbolz> itamarst: ah, interesting. the are doing this for reproducibility?
<itamarst> yeah
<itamarst> on the one hand it's very useful for that. on the other hand you lose real-world performance impacts of e.g. ILP
<itamarst> and for CPython lacking JIT, I don't think that's a problem
<itamarst> PyPy... dunno
<cfbolz> itamarst: are you contracting for them? or like part of the company or something?
<itamarst> contracting for a project that is using twisted that I am speeding up and therefore writing benchmarks and therefore setup codspeed
<itamarst> i.e. yak shaving
<cfbolz> heh, fair enough
<cfbolz> itamarst: that project is using pypy though? or you're thinking about benchmarking more generally in the context of the project
<itamarst> twisted and therefore the project based on it benefit heavily from PyPy because of way too many function calls
<itamarst> and so insofar as I'm doing performance work benchmarks are useful, and benchmarks running on PyPy are useful
<cfbolz> 👍
Dejan has quit [Quit: Leaving]
infernix has quit [Ping timeout: 260 seconds]
dustinm has quit [Quit: Leaving]
dustinm has joined #pypy
infernix has joined #pypy