01:17 kvw_5_ joined 01:20 kvw_5 left 01:27 MasterDuke left 02:33 maggotbrain joined 03:53 frost-lab joined 04:20 frost-lab left 04:28 frost-lab joined 05:38 frost-lab left, softmoth_ left 05:41 frost-lab joined 06:36 MasterDuke joined
[Tux] Yesterday: 06:44
Rakudo v2021.03-163-ga68d2579f (v6.d) on MoarVM 2021.03-47-gdba4b528e
csv-ip5xs0.867 - 0.876
csv-ip5xs-208.788 - 8.944
csv-parser25.853 - 26.001
csv-test-xs-200.375 - 0.377
test8.095 - 8.247
test-t2.409 - 2.459
test-t --race0.956 - 1.008
test-t-2042.571 - 42.754
test-t-20 --race11.944 - 12.033
06:50 domidumont joined 08:02 patrickb joined 08:06 patrickb left 08:51 Altai-man_ left 08:53 sena_kun joined 09:49 epony left 10:00 frost-lab left 10:03 frost-lab joined 10:17 epony joined 11:39 Kaeipi joined 11:40 Kaiepi left 12:02 frost-lab left 14:39 dogbert11 joined 14:42 dogbert17 left 16:05 softmoth_ joined
MasterDuke m: my $a; $a := $_ ** 2 for 1..10_000_000; say now - INIT now; say $a 16:07
camelia 7.175648797
100000000000000
MasterDuke m: my $a; $a := $_ ** -2 for 1..1_000_000; say now - INIT now; say $a
camelia 2.300756496
0.000000000001
MasterDuke i can decrease the first one to 5.6s, but at the cost of increasing the second to 3.7s 16:08
i'll PR the change, maybe someone can improve it 16:11
github.com/rakudo/rakudo/pull/4320 16:19
lizmat m: my $a; $a := $_ * $_ for 1..10_000_000; say now - INIT now; say $a
camelia 2.905321862
100000000000000
lizmat even faster ?
m: my $a; $a := $_ * $_ * $_ for 1..10_000_000; say now - INIT now; say $a 16:20
camelia 5.742425597
1000000000000000000000
lizmat I guess just special casing 0..3 would already be a win ?
MasterDuke i believe they are already in the libtommath implementation of pow_I 16:21
but fwiw, that change would also help the not-so-likely-to-be-worthy-of-special-casing-cases 16:22
lizmat m: my $a; use nqp; $a := nqp::pow_I($_,2) for 1..10_000_000; say now - INIT now; say $a
camelia ===SORRY!===
Arg count 2 doesn't equal required operand count 5 for op 'pow_I'
lizmat m: my $a; use nqp; $a := nqp::pow_I($_,2,Int) for 1..10_000_000; say now - INIT now; say $a
camelia ===SORRY!===
Arg count 3 doesn't equal required operand count 5 for op 'pow_I'
MasterDuke Num,Int
lizmat m: my $a; use nqp; $a := nqp::pow_I($_,2,Num,Int) for 1..10_000_000; say now - INIT now; say $a 16:23
camelia 2.397020746
100000000000000
lizmat m: my $a; use nqp; $a := nqp::pow_I($_,3,Num,Int) for 1..10_000_000; say now - INIT now; say $a
camelia 2.573100814
1000000000000000000000
lizmat so, why is $_ ** 2 so much slower then ?
MasterDuke github.com/rakudo/rakudo/issues/3910
[Tux] Rakudo v2021.03-169-g71d31a613 (v6.d) on MoarVM 2021.03-49-gd11bd7f11
csv-ip5xs0.864 - 0.906
csv-ip5xs-208.751 - 8.854
csv-parser25.732 - 26.809
csv-test-xs-200.376 - 0.387
test7.823 - 7.895
test-t2.300 - 2.347
test-t --race0.967 - 0.978
test-t-2042.712 - 43.167
test-t-20 --race11.462 - 11.620
 16:24
MasterDuke the speedup for **5 isn't quite as great. 7.5s before, 6.0s after 16:25
it'd be nice if pow_I had a way of signalling overflow that wasn't returning a different type 16:29
lizmat what strikes me as odd, is that even for something like: 16:34
sub a(\a) { a ** 2 }; a($_) for ^1000000
there is nothing the the speshlog that indicates an attempt of inlining infix:<**> into sub a
whereas sub a *is* inlined 16:35
raku -MSIL -e 'sub a(\a) { a ** 2 }; a($_) for ^1000000'
17:27 domidumont left
lizmat afk for the rest of the day (probably)& 18:02
18:09 patrickb joined 18:32 patrickb left
[Coke] TIL about -MSIL 18:32
18:35 patrickb joined
releasable6 Next release in ≈4 days and ≈23 hours. 2 blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 19:00
19:22 patrickb left, softmoth_ left 19:24 patrickb joined
japhb Since there are more folks here at this hour -- we *may* have found a problem with the libuv bump that happened a couple months ago. I and ugexe have both been seeing "Failed to open dir: Too many open files" errors when doing a `zef uninstall`. ugexe is on MacOS and I'm on Linux, so it doesn't appear to be platform-specific. When I rolled back to just before the libuv bump, I couldn't recreate the 19:30
problem.
Gah, I intended that to go to #moarvm, though perhaps it's useful here too. 19:31
nwc10 "this hour" - the optimal time of morning? :-)
japhb Well, I suppose "optimal" would be a couple hours earlier here, but this will have to do. :-) 19:33
s/here//
Clearly my brain is not optimally engaged
japhb attempts to apply music and tea to the problem 19:34
nwc10 :-)
I hope your tea making has been more sucessful than some of my recent efforts
japhb *chuckle* So far so good. 19:35
nwc10 colabti.org/irclogger/irclogger_lo...-04-12#l54
and the previous report about not using tea (and attempting to make coffee with cold water)
maybe I should stick to whisky. I think that I'm reasonably competant with the input side of whisky 19:36
19:58 softmoth_ joined 20:19 patrickb left 20:36 b2gills left, b2gills joined 21:07 dogbert17 joined 21:09 dogbert11 left
lizmat bisectable6: enum Str ( :Free<f>, :Inuse<n> ); note "ok %s".sprintf: Free; 21:22
bisectable6 lizmat, Will bisect the whole range automagically because no endpoints were provided, hang tight
21:22 dogbert11 joined
bisectable6 lizmat, Output on all releases: gist.github.com/f9ec2d9611ef24a058...3090cc017d 21:22
lizmat, Bisecting by exit code (old=2021.03 new=b1556cb). Old exit code: 0
21:22 dogbert11 left
bisectable6 lizmat, bisect log: gist.github.com/3b4673a7fe4c317290...28f555aa55 21:22
lizmat, (2021-04-11) github.com/rakudo/rakudo/commit/a6...59a3df9ff1
lizmat, Output on all releases and bisected commits: gist.github.com/fc0c8da7e599106a9d...9e78e80096
21:23 dogbert11 joined 21:24 dogbert17 left
lizmat bisectable6: enum Stringy ( :Free<f>, :Inuse<n> ); note sprintf("ok %s",Free) 21:26
bisectable6 lizmat, Will bisect the whole range automagically because no endpoints were provided, hang tight
lizmat, ¦6c (52 commits): «ok f␤»
lizmat, Nothing to bisect!
21:56 dogbert17 joined 21:59 dogbert11 left 22:01 dogbert17 left 22:02 dogbert17 joined, Geth joined 23:10 gfldex left 23:12 gfldex joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!