github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
01:16 MasterDuke left 01:28 lucasb left 01:42 zakharyas joined 02:42 evalable6 left 02:47 evalable6 joined
timotimo freelist => [[9 100] [9 99] [9 98] [9 97] [9 96] [9 95] [9 94] [9 93] [9 92] [9 91] [9 90] [9 89] [9 88] [9 87] [9 86] [9 85] [9 84] [9 83] [9 82] [9 81] [9 80] [9 79] [9 78] [9 77] [9 76] [9 75] [9 74] [9 73] [9 72] [9 71] [9 70] [9 69] [9 68] [9 67] [9 66] [9 65] [9 64] [9 63] [9 62] [9 61] [9 60] [9 59] [9 58] [9 57] [9 56] [9 55] [9 54] [9 53] [9 52] [9 51] [9 50] [9 49] [9 48] [9 47] [9 46] [9 45] [9 44] 02:54
[9 43] [9 42] [9 41] [9 40] [9 39] [9 38] [9 37] [9 36] [9 35] [9 34] [9 33] [9 32] [9 31] [9 30] [9 29] [9 28] [9 27] [9 26] [9 25] [9 24] [9 23]]
03:47 evalable6 left 03:51 evalable6 joined 04:40 zakharyas left 04:41 zakharyas joined 06:41 sena_kun joined 06:43 sena_kun left 07:04 sena_kun joined, sena_kun left 08:27 sena_kun joined 08:30 domidumont joined 09:46 zakharyas left 12:10 robertle joined 12:11 lucasb joined
Geth MoarVM: cygx++ created pull request #1161:
Pull out top-level API into standalone header		 13:05
16:41 MasterDuke joined 16:44 robertle left 16:51 pamplemousse joined 16:56 sena_kun left
MasterDuke timotimo, et al.: github.com/libtom/libtommath/pull/330 17:34
17:50 robertle joined
timotimo MasterDuke: so the torad values are for "the old implementation" and the todec ones are of "the new implementation"? 18:04
i would have loved one or two speedup x for all the blocks 18:05
it's really hard to see, especially since most values don't line up
looks like their travis is giving a few red Xes 18:06
looks like all you need to do for the very first one is run astyle against your code to have it formatted for you?! 18:07
i dunno
looks like you'll want to rename "index" since string.h already declares it
anyway, you can read the travis logs yourself, i'm sure me pointing out each red X isn't worth much 18:08
18:38 sena_kun joined
MasterDuke huh, some of the existing code wouldn't survive astyle. oh well, guess i can run it 18:55
18:57 domidumont left 19:01 pamplemousse left 19:02 pamplemousse joined
MasterDuke timotimo: and yes re old vs new values. the new version is faster at ~2**2000. if you want to help make it non-recursive maybe it could be made faster at lower values... 19:08
19:13 Kaiepi left, Kaiepi joined 19:27 Ven``_ joined 19:38 Altai-man_ joined 19:40 sena_kun left 19:43 pamplemousse left 19:44 Ven``_ left 19:45 Ven``_ joined 19:54 Altai-man_ left, sena_kun joined 19:55 sena_kun left
timotimo i now have a little dataset of how usage of the FSA changes over time (in a short timeframe, so "high resolution") 21:40
MasterDuke what can you do with that?
timotimo i want an animation 21:41
MasterDuke for decorative purposes? or think it will teach something? 21:42
timotimo i hope to learn something
MasterDuke nice 21:43
timotimo i also want to do the same thing for the gen2, because it is structured similarly
MasterDuke anything in particular you think you'll learn? 21:50
timotimo i would hope to find out why some things grow over a long time 21:55
like the whateverable bots
MasterDuke oh, that would be good to know 21:58
AlexDaniel there was something… 21:59
AlexDaniel looks at github.com/perl6/whateverable/issues/24
timotimo: see github.com/rakudo/rakudo/issues/1501
there was also a long explanation by jnthn somewhere 22:01
from what I remember, it's not leaking memory but it tends to settle on an amount that's a bit too high
timotimo right
AlexDaniel at the time it was “OK”, but I think we should be aiming at a much smaller footprint 22:02
MasterDuke the last comment by bbkr says there's still a (smaller than before) leak
22:03 Ven``_ left
AlexDaniel the memory issue came up during perlcon in one of the discussions 22:04
in which I honestly said that they should be expecting half a gig of ram usage per process
to me it seems like performance is not a big roadblock for adoption anymore, but memory usage probably is 22:06
so timotimo++ 22:07
timotimo i feel brain-paralyzed
MasterDuke well, i'd argue that parsing/regex/grammar performance is still so bad as to likely be a blocker for the kinds of things people use perls for 22:08
AlexDaniel MasterDuke: that's true, but in many other use cases it makes no difference at all
say building services with perl6
MasterDuke but otherwise performance as improved dramatically, way more that i thought possible in such a short time
AlexDaniel I don't mind perl5 being the queen of ascii crunching 22:10
timotimo personally, i'm very anti-monarchies, but continue
Kaiepi is half a gig the minimum or just on average? 22:11
AlexDaniel shareable6: uptime
shareable6 AlexDaniel, 4 days, 8 hours, 36 minutes, and 37 seconds, 566.769531MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
MasterDuke i wouldn't mind terribly if perl5 is faster on average. i mind how much faster it is
timotimo this is a bot that does almost nothing, right?
AlexDaniel quotable6: uptime 22:12
quotable6 AlexDaniel, 1 day, 20 hours, 24 minutes, and 58 seconds, 227.265625MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
AlexDaniel Kaiepi: it's something you should expect your memory usage to settle on
so for short-lived processes maybe just 250 MiB
Kaiepi ah, ok 22:13
AlexDaniel timotimo: yeah, it doesn't do anything unless somebody runs Blin
timotimo: I don't know if anybody did 22:14
during the last 4 days
timotimo we need some observability :)
AlexDaniel bisectable6: uptime
bisectable6 AlexDaniel, 3 weeks, 4 days, 10 hours, 14 minutes, and 26 seconds, 431.3125MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
AlexDaniel bisectable is not used that often, but look that's 3 weeks
unicodable6: uptime 22:15
tellable6 AlexDaniel, I'll pass your message to unicodable6
timotimo hahahaha
Kaiepi i've never seen one of the bots i run go higher than 400MB apart from back when there were pretty big memory leaks in sockets code, but it's not stable enough to keep more than a few days uptime yet
AlexDaniel unicodable6_: uptime
unicodable6_ AlexDaniel, 3 weeks, 4 days, 10 hours, 15 minutes, and 14 seconds, 405.726563MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
tellable6 2019-08-23T22:15:04Z #moarvm <AlexDaniel> unicodable6: uptime
AlexDaniel Kaiepi: what do you mean not stable enough? Does it segfault? 22:16
Kaiepi no, it's something really difficult to unit test properly
nothing to do with moarvm itself 22:17
AlexDaniel oh, let's try that again 22:18
unicodable6: uptime
huh…
timotimo okay i really need to run the bots myself, amn't i
AlexDaniel unicodabl6: uptime 22:19
unicodable6_ AlexDaniel, 3 weeks, 4 days, 10 hours, 19 minutes, and 43 seconds, 408.871094MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
AlexDaniel so I have to spell it incorrectly for the bot to figure it out 22:20
timotimo oi AlexDaniel are the bots built with debug set to anything other than 0?
AlexDaniel timotimo: no
timotimo: nowadays it's easy to run the bots
timotimo damn
AlexDaniel they'll fetch rakudo releases and stuff automatically
timotimo i would really want to know how often it has run the GC in those 3 weeks
Kaiepi there was a period not too long ago where i was getting deadlocks in it that i wasn't getting with the same code in a slightly earlier compiler release, but that doesn't happen anymore afaik
AlexDaniel as long as you're on 62-bit linux
ooooh 22:21
MasterDuke AlexDaniel: that's your problem, you've 2 bits too few in your linux
AlexDaniel 62-bit linux, yes… 62… haha
timotimo: anyway I can put one of the bots down so that you can run a replacement 22:22
installation info here: github.com/perl6/whateverable/wiki/Installation
timotimo i suspend2ram my computer over night 22:23
Kaiepi 62-bit linux, finally a use for those types like uint_least_16_t :^) 22:24
timotimo AlexDaniel: which bot do you think is the least likely to be missed?
coverable i think was recently mentioned as "never used"
AlexDaniel bloatable6: uptime 22:25
bloatable6 AlexDaniel, 3 weeks, 4 days, 10 hours, 25 minutes, and 1 second, 376.6875MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
AlexDaniel bloatable6: help
bloatable6 AlexDaniel, Like this: bloatable6: d=compileunits 292dc6a,HEAD # See wiki for more examples: github.com/perl6/whateverable/wiki/Bloatable
AlexDaniel bloatable6: releases
let's see if it even works
bloatable6 AlexDaniel, gist.github.com/c87bcb8f269fc88862...f48eaaf7cc
timotimo all these broken archives back then, i wonder how that happened 22:26
AlexDaniel it sorta does
bloatable6: uptime
bloatable6 AlexDaniel, 3 weeks, 4 days, 10 hours, 26 minutes, and 53 seconds, 430.289063MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
MasterDuke timotimo: btw, any idea why i'm getting this warning? travis-ci.org/libtom/libtommath/jobs/576029036. i do have `#include <stdio.h>` in the file 22:27
AlexDaniel timotimo: I'd say bloatable is a good choice because at least it doesn't allow people to eval code :)
timotimo ah
good point 22:28
AlexDaniel quotable6: uptime
quotable6 AlexDaniel, 1 day, 20 hours, 40 minutes, and 57 seconds, 228.8125MiB maxrss. This is Rakudo version 2019.07.1-94-gd1f9d2848 built on MoarVM version 2019.07.1-50-gb614a7b4d implementing Perl 6.d.
timotimo AlexDaniel: whateverable-data seems to be empty? 22:32
AlexDaniel timotimo: it's not used yet
I was trying to set it up and then got distracted with something else… 22:33
timotimo OK
AlexDaniel bloatable needs moarvm builds which it'll automatically attempt to download from shareable
timotimo right 22:34
22:34 bloatable6 left 22:36 Ven`` joined 22:40 Ven`` left
timotimo i didn't make the feature to spit out the FSA data optional 22:42
so it's either turned on or off at compile time
don't want to run sake or anything else with that on, i don't think
ah, just using the stuff in xbin i guess 22:44
AlexDaniel I was thinking: PERL6LIB=lib xbin/Bloatable.p6 < config.json
default config will also work of course 22:45
timotimo oh, config
AlexDaniel but hmm that doesn't seem to do much
timotimo i'll have to give it the right hostname for shareable 22:46
AlexDaniel the default one is probably right
timotimo "localhost"
AlexDaniel no that's if you want to run shareable
"mothership" is the one you have in mind
timotimo oh, ok 22:47
AlexDaniel PERL6LIB=lib DEBUGGABLE=1 xbin/Bloatable.p6 < config-default.json
that works
timotimo OK
AlexDaniel DEBUGGABLE will make IRC::Client more verbose 22:48
timotimo Failed to open file /home/timo/perl6/ecosystem/whateverable/lib/Whateverable/Replaceable.pm6: No such file or directory
AlexDaniel ehh… trying to install it? :)
timotimo no 22:49
oh, you wanted me to perl6lib lib
13 processes already 22:50
and no output?
did i do it wrong? 22:51
AlexDaniel timotimo: fixed in github.com/perl6/whateverable/comm...367c2f3dc4
timotimo: well, there are some possibilities…
either you're waiting for it to precomp… 22:52
timotimo oh
i forgot to pipe the config into it
well that isn't going to work well
AlexDaniel timotimo: or that
or without DEBUGGABLE it may look like nothing is happening
timotimo: also, there's this line: github.com/perl6/whateverable/blob...e.pm6#L200 22:53
timotimo no, debuggable is turned on
AlexDaniel comment it out
it's for running 18 bots, otherwise they get k-lined
timotimo ah, right 22:54
i'll just let it sit and see if it GCs a bunch without noticeable activity 23:21
like, does the thread pool scheduler supervisor thread happen to allocate stuff again?
it rose above a meg 23:24
but every entry in that file is kind of big 23:25
maybe i should also zstd it on the way to the file
hurm, just reading the big messagepack file grows more and more in memory usage 23:58

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!