#pypy on 2021-08-23 — irc logs at libera.irclog.whitequark.org

2021-05-30 17:38 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 | insert pithy quote here

00:12 amccollum has quit [Ping timeout: 250 seconds]

00:25 fotis has joined #pypy

00:29 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/8812 [py3.8]

00:31 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-32/builds/7609 [py3.8]

00:33 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-s390x/builds/1587 [py3.8]

00:40 amccollum has joined #pypy

00:42 lritter has quit [Ping timeout: 240 seconds]

00:45 amccollum has quit [Ping timeout: 268 seconds]

00:59 fotis has quit [Ping timeout: 252 seconds]

01:00 <bbot2> Started: http://buildbot.pypy.org/builders/own-linux-x86-32/builds/7610 [py3.7]

01:00 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-s390x/builds/1588 [py3.7]

01:00 <bbot2> Started: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/8813 [py3.7]

01:00 fotis has joined #pypy

01:05 fotis has quit [Ping timeout: 250 seconds]

01:07 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/5668 [py3.8]

01:07 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/5669 [py3.7]

01:16 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/6463 [py3.8]

01:16 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/6464 [py3.7]

01:19 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-64/builds/357 [py3.8]

01:19 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-64/builds/358 [py3.7]

01:21 amccollum has joined #pypy

01:32 amccollum has quit [Ping timeout: 268 seconds]

01:32 fotis has joined #pypy

01:49 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/8813 [py3.7]

02:01 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-32/builds/7610 [py3.7]

02:01 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-s390x/builds/1588 [py3.7]

02:09 fotis has quit [Ping timeout: 252 seconds]

02:12 fotis has joined #pypy

02:14 gef_ has quit [Ping timeout: 248 seconds]

02:15 gef has quit [Ping timeout: 250 seconds]

02:17 fotis has quit [Ping timeout: 250 seconds]

02:18 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/6464 [py3.7]

02:20 amccollum has joined #pypy

02:27 amccollum has quit [Ping timeout: 248 seconds]

02:28 amccollum has joined #pypy

02:29 <bbot2> Failure: http://buildbot.pypy.org/builders/own-win-x86-64/builds/405 [py3.8]

02:29 <bbot2> Started: http://buildbot.pypy.org/builders/own-win-x86-64/builds/406 [py3.7]

02:38 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7818 [py3.8]

02:38 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7819 [py3.7]

02:40 amccollum has quit [Ping timeout: 252 seconds]

02:45 amccollum has joined #pypy

02:45 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/5669 [py3.7]

02:47 fotis has joined #pypy

02:50 amccollum has quit [Ping timeout: 258 seconds]

02:57 yuiza has joined #pypy

03:21 fotis has quit [Ping timeout: 268 seconds]

03:22 fotis has joined #pypy

03:24 amccollum has joined #pypy

03:27 fotis has quit [Ping timeout: 250 seconds]

03:30 amccollum has quit [Remote host closed the connection]

03:30 amccollum has joined #pypy

03:39 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7819 [py3.7]

03:55 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/1088 [py3.8]

03:55 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/1089 [py3.7]

03:58 fotis has joined #pypy

04:10 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-64/builds/358 [py3.7]

04:11 gef has joined #pypy

04:13 stkrdknmibalz has joined #pypy

04:14 gef_ has joined #pypy

04:33 fotis has quit [Ping timeout: 258 seconds]

04:34 fotis has joined #pypy

04:39 fotis has quit [Ping timeout: 252 seconds]

04:44 jkovac16 has joined #pypy

04:45 marmoute_ has joined #pypy

04:45 tumbleweed_ has joined #pypy

04:45 jerith_ has joined #pypy

04:45 marmoute has quit [*.net *.split]

04:45 tumbleweed has quit [*.net *.split]

04:45 jerith has quit [*.net *.split]

04:45 jerith_ is now known as jerith

04:45 gef_ has quit [Ping timeout: 248 seconds]

04:45 gef has quit [Ping timeout: 252 seconds]

04:46 jkovac1 has quit [Ping timeout: 248 seconds]

04:46 jkovac16 is now known as jkovac1

04:48 xcm_ has quit [*.net *.split]

04:48 jstoker has quit [*.net *.split]

04:48 nimaje has quit [*.net *.split]

04:48 agronholm has quit [*.net *.split]

04:48 agronholm_ has joined #pypy

04:49 jstoker_ has joined #pypy

04:49 xcm_ has joined #pypy

04:49 nimaje has joined #pypy

04:49 xcm_ has joined #pypy

04:49 xcm_ has quit [Signing in (xcm_)]

04:49 jstoker_ is now known as jstoker

05:05 fotis has joined #pypy

05:14 amccollum has quit [Remote host closed the connection]

05:15 amccollum has joined #pypy

05:16 amccollum has quit [Remote host closed the connection]

05:16 amccollum has joined #pypy

05:18 <bbot2> Failure: http://buildbot.pypy.org/builders/own-win-x86-64/builds/406 [py3.7]

05:35 amccollum has quit []

05:39 fotis has quit [Ping timeout: 240 seconds]

06:08 fotis has joined #pypy

06:17 <mattip> py3.8 tests now stand at {'Ran': 9102, 'failures': 227, 'errors': 40, 'skipped': 1483}

06:23 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-aarch64/builds/992 [py3.8]

06:23 <bbot2> Started: http://buildbot.pypy.org/builders/own-linux-aarch64/builds/993 [py3.7]

06:28 glyph has quit [Quit: End of line.]

06:28 glyph has joined #pypy

06:31 mathieu1 has joined #pypy

06:38 mathieu1 has quit [Ping timeout: 240 seconds]

06:38 mathieu1 has joined #pypy

06:42 fotis has quit [Ping timeout: 250 seconds]

06:42 fotis has joined #pypy

06:45 gef has joined #pypy

06:45 gef_ has joined #pypy

06:48 yuiza has quit [Remote host closed the connection]

06:49 fotis has quit [Ping timeout: 248 seconds]

07:00 fotis has joined #pypy

07:06 fotis has quit [Ping timeout: 252 seconds]

07:06 fotis has joined #pypy

07:12 fotis has quit [Ping timeout: 248 seconds]

07:12 <mattip> 507 packages have passed the windows migration on conda-forge, and conda-forge's pytorch feedstock is close to working

07:14 fotis has joined #pypy

07:15 <LarstiQ> feedstock?

07:17 gef has quit [Ping timeout: 240 seconds]

07:17 gef_ has quit [Ping timeout: 248 seconds]

07:20 fotis has quit [Ping timeout: 252 seconds]

07:22 fotis has joined #pypy

07:28 fotis has quit [Ping timeout: 240 seconds]

07:29 <mattip> that's what they call "wheels"

07:31 <mattip> https://github.com/conda-forge/pytorch-cpu-feedstock/pull/66

07:31 Atque has quit [Ping timeout: 240 seconds]

07:57 <LarstiQ> ah

07:59 gef has joined #pypy

08:01 gef_ has joined #pypy

08:14 marmoute_ is now known as marmoute

08:22 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7820 [Batuhan Taskaya: force build, jit-fewer-bridges]

08:31 <bbot2> Started: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/3362 [Carl Friedrich Bolz-Tereick: force build, jit-fewer-bridges]

08:44 Julian has joined #pypy

08:45 <cfbolz> mattip: isidentical and me just finished a version of the jit that should be able to stop making endless bridges

09:05 amirouche has quit [Quit: WeeChat 2.8]

09:05 amirouche has joined #pypy

09:07 <LarstiQ> cfbolz: finished? The tweet I just read sounded like you were starting, that's quick :)

09:08 <cfbolz> LarstiQ: it's the first attempt

09:08 <cfbolz> will need super careful tuning

09:08 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7820 [Batuhan Taskaya: force build, jit-fewer-bridges]

09:08 <cfbolz> we don't want to make existing benchmarks slower (which mostly aren't affected by the performance problem, except maybe for sphinx)

09:12 amirouche is now known as babelbot

09:16 babelbot is now known as amirouche

09:17 <LarstiQ> cfbolz: right

09:17 <LarstiQ> is it even possible to choose a cutoff point that works for every piece of code?

09:18 <cfbolz> LarstiQ: probably not

09:18 <cfbolz> But there are various such trade offs

09:18 <mattip> nice progress

09:19 <cfbolz> Batuhan is really good, was super efficient to work with him

09:19 <mattip> :)

09:24 fotis has joined #pypy

09:29 fotis has quit [Ping timeout: 248 seconds]

09:29 yuiza has joined #pypy

09:30 fotis has joined #pypy

09:34 fotis has quit [Ping timeout: 240 seconds]

09:39 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-aarch64/builds/1089 [py3.7]

09:58 <yuiza> Hi, where can I access benchmark programs for PyPy 3?

09:59 Julian has quit [Ping timeout: 252 seconds]

10:01 Julian has joined #pypy

10:10 <bbot2> Failure: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/3362 [Carl Friedrich Bolz-Tereick: force build, jit-fewer-bridges]

10:10 <bbot2> Started: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/3363 [py3.7]

10:14 Atque has joined #pypy

10:34 <mattip> yuiza: is https://foss.heptapod.net/pypy/benchmarks what you are looking for?

11:18 Julian has quit [Ping timeout: 252 seconds]

11:30 fotis has joined #pypy

11:36 fotis has quit [Ping timeout: 248 seconds]

11:40 <bbot2> Success: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/3363 [py3.7]

11:54 <yuiza> mattip: exactly. I found many benchmarks namely py3_*.py from the branch https://foss.heptapod.net/pypy/benchmarks/-/tree/topic/python3_benchmarks/bitbucket-pr-5

11:55 <yuiza> mattip: thank you!

11:57 fotis has joined #pypy

12:08 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-aarch64/builds/993 [py3.7]

12:17 fotis has quit [Ping timeout: 250 seconds]

12:18 fotis has joined #pypy

12:20 stkrdknmibalz has quit [Ping timeout: 240 seconds]

12:23 fotis has quit [Ping timeout: 240 seconds]

12:32 <mattip> yuiza: we have kept these benchmarks around at the version they are so that the results on speed.pypy.org remain stable

12:32 <mattip> for more updated versions, perhaps the pyperf suite has some interesting things

12:33 <mattip> https://pypi.org/project/pytest-benchmark/

12:33 <mattip> although some of those are tiny micro-benchmarks

12:34 <mattip> and then there is also https://pybenchmarks.org/

12:34 <mattip> which we discussed here a few days ago

12:53 fotis has joined #pypy

13:12 Dejan has joined #pypy

13:24 <yuiza> mattip: Oh, many thanks. I measure the performance of threaded code generation in PyPy (you can find the code at the threaded-code-generation branch in PyPy repo). It is the reason why I ask here. Threaded code generation is still work-in-progress, so I currently use the branch py3.7-hack-measure-bytecode-dispatch.

13:28 <mattip> the idea is to run the JIT in another thread?

13:28 <Dejan> that does make sense

13:35 <yuiza> mattip: Thread in threaded code doesn't mean a thread. It is a technique to generate code that essentially contains call instructions to subroutines. I and cfbolz are creating a method-based JIT on PyPy.

13:36 <yuiza> This is the position paper describing our idea https://arxiv.org/abs/2106.12496

13:37 <yuiza> We aim to use threaded code for a baseline method-based JIT. After creating it, we create a fully optimized method-based JIT based on the baseline JIT.

13:45 lritter has joined #pypy

13:52 <mattip> ahh, cool. So the "thread" is as in the wikipedia definition

13:52 <mattip> https://en.wikipedia.org/wiki/Thread_(computing)

13:53 <mattip> but not necessarily executed concurrently with other such pieces of code

13:57 fotis has quit [Ping timeout: 252 seconds]

13:57 <yuiza> mattip: The word is a little bit confusing. Yes, not concurrently executed. It's just a compilation technique.

13:59 <Corbin> The term makes me think of Forth compilation strategies.

14:00 fotis has joined #pypy

14:03 <yuiza> Corbin: Forth..!? It's very old programming language. Actually threaded code might be used in Forth. https://dl.acm.org/doi/abs/10.1145/146559.146561

14:06 yuiza_ has joined #pypy

14:06 fotis has quit [Ping timeout: 240 seconds]

14:10 <Dejan> I've just finished reading the paper

14:11 <Dejan> looking forward to benchmark results :)

14:18 Dejan has quit [Quit: Leaving]

14:31 Ninpo has quit [Quit: WeeChat 2.5]

14:35 Ninpo has joined #pypy

14:36 fotis has joined #pypy

14:40 Ninpo has quit [Quit: WeeChat 2.8]

14:42 Ninpo has joined #pypy

14:55 Dejan has joined #pypy

15:13 fotis has quit [Ping timeout: 250 seconds]