Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2014-10-13

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:01 colomon joined #moarvm
00:13 cognome joined #moarvm
00:16 cognome joined #moarvm
00:17 cognome_ joined #moarvm
00:44 cognome joined #moarvm
00:45 cognome_ joined #moarvm
01:01 lizmat_ joined #moarvm
01:18 camelia joined #moarvm
01:46 cognome joined #moarvm
02:01 colomon joined #moarvm
02:13 colomon joined #moarvm
02:17 JimmyZ joined #moarvm
02:19 JimmyZ Hello, what is callsite intern?
02:33 colomon joined #moarvm
02:46 cognome joined #moarvm
02:50 mlilenium_ joined #moarvm
02:50 mlilenium_ left #moarvm
03:03 JimmyZ_ joined #moarvm
03:46 cognome joined #moarvm
03:54 cognome joined #moarvm
04:45 ggoebel1111118 joined #moarvm
04:46 cognome joined #moarvm
05:46 cognome joined #moarvm
06:28 JimmyZ_ joined #moarvm
06:34 nine TimToady: sudo cpupower frequency-set -g performance # gives stabler benchmark numbers
06:35 JimmyZ cpupower doesn't exist on ubuntu ...
06:37 JimmyZ ok, installing it.
06:38 nine An alternative would be: sudo cpufreq-set -g performance
06:44 JimmyZ hmm , I didnt' see -g performance in the man page :(
06:46 cognome joined #moarvm
06:54 timotimo JimmyZ: we only generate (or look for?) spesh candidates when we invoke something if the callsite has been interned
06:54 timotimo interning means we stash the callsite in our interning array and have made sure that the pointer points to the interned callsite rather than the one we may have just created
06:57 timotimo okay, so stuff for today:
06:58 timotimo - look more closely at how we spesh decont and maybe turn deconts on known container types into a direct pointer lookup
06:58 timotimo - look more closely at assign and maybe turn assign into assignunchecked and perhaps even into a direct pointer assignment
06:59 timotimo oh, also: a decont on a code-ish container type (the proxies) could become an invoke directly and then we could inline the little piece of code it has in BOOTSTRAP.nqp
07:01 Ven joined #moarvm
07:02 JimmyZ other quesiont:  I see calliste is intened after deserialization, can we intern the callsite before serialization? :0
07:03 JimmyZ that is, seriallize the interned calliste :P
07:05 timotimo - run some proper tests to see if (or why not) the invocations from all over our c code get treated by spesh at all
07:06 zakharyas joined #moarvm
07:13 kjs_ joined #moarvm
07:46 cognome joined #moarvm
08:04 _sri joined #moarvm
08:07 _sri joined #moarvm
08:23 lizmat joined #moarvm
08:36 woolfy joined #moarvm
08:46 cognome joined #moarvm
09:08 FROGGS joined #moarvm
09:09 Ven joined #moarvm
09:27 kjs_ joined #moarvm
09:46 cognome joined #moarvm
09:53 cognome joined #moarvm
10:13 flaviusb joined #moarvm
10:32 timotimo guess what was wrong with my trigonometry?
10:32 timotimo i was using the atan function to implement the atan2 op
10:33 jnthn It's broken 'cos you comitted a sin
10:33 tadzik *g*
10:33 dalek MoarVM: eb73c5e | (Timo Paulssen)++ | src/jit/graph.c:
10:33 dalek MoarVM: atan is a very poor implementation for atan2.
10:33 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/eb73c5e72a
10:34 tadzik even the argument list is different :D
10:34 tadzik gotta love gotos
10:47 timotimo wtf.
10:47 timotimo what, argument list?
10:47 timotimo OH FFS
10:47 timotimo thank you
10:48 tadzik :)
10:48 tadzik well, you already fixed it anyway, I think?
10:49 timotimo something else is wrong, too
10:50 timotimo why is this still wrong >:(
10:50 timotimo ah doh
10:50 timotimo no, damn, i ah'd too quickly
10:53 dalek MoarVM: a71fb99 | (Timo Paulssen)++ | src/jit/graph.c:
10:53 dalek MoarVM: atan2 actually takes two arguments.
10:53 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/a71fb99f98
10:53 timotimo this is still wrong :(
10:53 timotimo tadzik: tell me what i'm doing wrong :(
10:54 timotimo i'm staring at the code and can't see what is wrong
10:54 tadzik I woke up at 5 today, I can't help anyone with anything :)
10:55 timotimo :|
11:34 kjs_ joined #moarvm
11:40 dalek MoarVM: ca1f3b2 | (Timo Paulssen)++ | src/jit/graph.c:
11:40 dalek MoarVM: fix atan2 op; ven++ for tracking this down
11:40 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/ca1f3b26ea
11:41 timotimo how the fucking fuck can this fix it?
11:42 Ven \o/ I fixed it.
11:51 dalek MoarVM: 641391a | (Timo Paulssen)++ | src/jit/graph.c:
11:51 dalek MoarVM: Revert "fix atan2 op; ven++ for tracking this down"
11:51 dalek MoarVM:
11:51 dalek MoarVM: This reverts commit ca1f3b26eaf465880a122d66a43efc5c921d000a.
11:51 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/641391adc6
12:04 timotimo something must have been in a weird state before i committed ca1f3b2
12:04 timotimo because that commit was apparently not necessary
12:04 brrt joined #moarvm
12:12 JimmyZ joined #moarvm
12:23 jnthn joined #moarvm
12:24 nine joined #moarvm
12:37 brrt joined #moarvm
12:43 daxim joined #moarvm
12:55 dalek MoarVM: 256554b | (Timo Paulssen)++ | src/jit/graph.c:
12:55 dalek MoarVM: fix atan2 and friends for real this time; jnthn++ and nwc10++
12:55 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/256554b8a2
12:56 * jnthn tries building that
13:04 kjs_ joined #moarvm
13:17 brrt yeah, that's a good idea timotimo :-)
13:17 brrt that should work a bit better
13:21 dalek MoarVM: 03ac9a7 | (Tobias Leich)++ | src/io/syncfile.c:
13:21 dalek MoarVM: implement set_separator for syncfiles
13:21 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/03ac9a70f6
13:30 nwc10 timotimo++ # fixed
13:35 timotimo "a bit"? %)
13:35 timotimo i think it only passed the spec tests on my machine because the right input arguments happened to have been put onto the "floating point stack"-like thing before by other ops (like unbox_n or something)
13:37 brrt that happens often
13:37 brrt that can be a really annoying way to mask bugs
13:37 brrt in fact
13:37 brrt i have a blog post about just that
13:38 timotimo oh damn
13:38 timotimo i seem to recall reading that
13:41 brrt :-)
13:41 timotimo it's especially annoying because the day before i tried to build neg_n
13:41 timotimo and had to use the special movsd or whatever instruction
13:41 timotimo (also, it didn't work out)
13:41 brrt no, neg_n is hard
13:42 JimmyZ looks like some ops which calls MVMROOT is a bit hard to jit, except wrapping them to function?
13:43 * JimmyZ hopes we can jit the MVMROOT too :P
13:44 brrt hmmm
13:44 brrt let me think about that for a bit
13:44 JimmyZ i.e: add_I etc..
13:45 kjs_ joined #moarvm
13:45 brrt but yeah... in the general case wrapping-to-function is not going to cost you /that/ much
13:45 JimmyZ these ops avoids a MVROOT, by using REGS, since gc updates REGS address.
13:47 brrt yeah, i don't think they need a MVMROOT
13:49 brrt hmm
13:50 brrt basically, this is for stack-variables mostly
13:50 brrt i don't really do that often
14:05 kjs_ joined #moarvm
14:10 lizmat joined #moarvm
15:29 brrt joined #moarvm
15:40 woolfy left #moarvm
16:00 woolfy joined #moarvm
16:00 woolfy left #moarvm
16:00 lizmat joined #moarvm
16:01 kjs_ joined #moarvm
16:28 cognome joined #moarvm
20:46 FROGGS joined #moarvm
20:48 Ven joined #moarvm
21:02 timotimo we could block sigterm with sigprocmask and check for sigterm in the "main thread" and get an actually useful stack trace like people who come from python are used to.
21:03 timotimo except signal(SIGTERM) doesn't wrok
21:03 timotimo work
21:28 brrt joined #moarvm
21:44 brrt ok, numeric negation is just xor with the highest possible byte
22:06 camelia joined #moarvm
23:13 dalek MoarVM: e75de84 | (Bart Wiegmans)++ | src/jit/ (2 files):
23:13 dalek MoarVM: Add numeric negation
23:13 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/e75de84f84
23:13 brrt left #moarvm
23:38 cognome joined #moarvm

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary