00:05
justinmcp_ has quit [Quit: No Ping reply in 180 seconds.]
00:06
justinmcp has joined #jruby
00:11
justinmcp has quit [Client Quit]
00:12
justinmcp has joined #jruby
02:46
_whitelogger has joined #jruby
07:57
jmalves has quit [Remote host closed the connection]
07:58
jmalves has joined #jruby
08:02
bga57 has quit [Ping timeout: 272 seconds]
08:49
_whitelogger has joined #jruby
09:19
jmalves has quit [Remote host closed the connection]
09:19
jmalves has joined #jruby
09:22
jmalves has quit [Remote host closed the connection]
09:51
jmalves has joined #jruby
10:24
drbobbeaty has quit [Ping timeout: 252 seconds]
11:08
KeyJoo has joined #jruby
11:43
KeyJoo has quit [Read error: Connection reset by peer]
11:50
KeyJoo has joined #jruby
12:09
KeyJoo has quit [Remote host closed the connection]
12:57
travis-ci has joined #jruby
12:57
travis-ci has left #jruby [#jruby]
13:00
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
13:21
slyphon has joined #jruby
13:44
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
14:39
travis-ci has joined #jruby
14:39
travis-ci has left #jruby [#jruby]
14:43
jmalves has quit [Ping timeout: 252 seconds]
14:46
slyphon has joined #jruby
14:48
slyphon_ has joined #jruby
14:51
slyphon has quit [Ping timeout: 244 seconds]
15:16
bga57 has joined #jruby
15:57
slyphon has joined #jruby
15:59
slyphon_ has quit [Ping timeout: 250 seconds]
16:03
xardion has quit [Remote host closed the connection]
16:08
xardion has joined #jruby
16:15
<
lopex >
not entirely correct though
16:59
travis-ci has joined #jruby
16:59
travis-ci has left #jruby [#jruby]
17:11
travis-ci has joined #jruby
17:11
travis-ci has left #jruby [#jruby]
17:37
subbu is now known as subbu|lunch
17:48
travis-ci has joined #jruby
17:48
travis-ci has left #jruby [#jruby]
17:57
travis-ci has joined #jruby
17:57
travis-ci has left #jruby [#jruby]
18:03
<
headius >
lopex: we support the "absent operator" from Onigmo now right?
18:06
subbu|lunch is now known as subbu
18:07
<
lopex >
headius: some
18:07
<
lopex >
headius: well, most
18:08
<
lopex >
very few cases blowing
18:08
<
lopex >
which will require us to sync with onigmos gpos thing
18:10
<
lopex >
I tried to tweak it without it but failed
18:11
<
lopex >
headius: it has separate opcode with quite peculiar implementation
18:12
travis-ci has joined #jruby
18:12
travis-ci has left #jruby [#jruby]
18:20
<
lopex >
untested on oxs
18:21
<
lopex >
if after gneration there's any diff then somethings wrong
18:22
<
lopex >
I know it depends on a toolchain but that's what we have for now
18:32
<
lopex >
lol @ OS.equals(LINUX) && OS_VERSION.contains("Microsoft");
18:42
<
enebo >
lopex: have you tried just adding Matcher as a static field to test overhead for match?
18:43
<
enebo >
lopex: I am curious if it makes a big difference or not. I was thinking about idea that callsite itself could cache it and we would just need to make sure it was only used by one thread
18:44
<
headius >
lopex: ok
18:55
<
enebo >
lopex: I just tried...0.29s and once I added static field just to keep matcher around it went down to 0.24/5...2.6preview is 0.14s
19:00
<
lopex >
so there we go
19:02
<
lopex >
enebo: so last chance would be smaller interpreter loop just for a test
19:02
<
lopex >
enebo: the other to run it on excelsior with index bounds checks off :P
19:02
<
enebo >
lopex: I changed bench to be a method with a 1m while loop to remove ips+block overhead
19:03
<
lopex >
otherwise all those methods are just too big
19:03
<
enebo >
lopex: jruby 9.2.1.0-SNAPSHOT (2.5.0) 2018-09-20 e9abd54 Excelsior JET 1.8.0_144 on 1.8.0_144-jdk_2017_08_24_19_55-b00 +jit [linux-i386]
19:03
<
enebo >
can I actually pass that in as an option?
19:03
<
lopex >
enebo: what ?
19:03
<
lopex >
enebo: I believe so, but on AOT
19:03
<
enebo >
I built jruby executable a couple of weeks ago
19:04
<
lopex >
enebo: is it aoted ?
19:05
<
enebo >
jruby: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux.so.2, not stripped
19:07
<
lopex >
enebo: -fno-bounds-check ?
19:09
<
lopex >
so I always assumed there's an option
19:11
<
lopex >
I'm unable to find anything now though
19:16
<
lopex >
enebo: it must be my bad memory then
19:17
<
lopex >
enebo: but I'd swear they supported some unsafe things
19:27
<
enebo >
lopex: I thought for array access checks we can do something odd like check somethingon the array before entering the loop which guarantees the size
19:28
<
lopex >
and the like
19:29
<
enebo >
lopex: ah unless it all inlines together or something
19:29
<
lopex >
and the opcode array
19:29
<
enebo >
but I can see what you mean
19:29
<
lopex >
those are the biggest offenders imho wrt bounds checks
19:30
<
lopex >
those are mostly predicted but the pipeline is still larger
19:30
<
lopex >
and we dont know how deep btb buffers are
19:30
<
lopex >
well, how far do they remember
19:31
<
lopex >
going to Cliff Click territory :P
19:56
travis-ci has joined #jruby
19:56
travis-ci has left #jruby [#jruby]
21:00
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
21:38
travis-ci has joined #jruby
21:38
travis-ci has left #jruby [#jruby]
22:04
slyphon has joined #jruby
22:16
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
22:33
slyphon has joined #jruby
23:29
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
23:29
travis-ci has joined #jruby
23:29
travis-ci has left #jruby [#jruby]
23:35
Eiam has joined #jruby
23:48
slyphon has joined #jruby