00:14:16  * TheJHquit (Quit: goodbye)
00:17:42  * C-Manquit (Quit: Connection reset by beer)
00:21:13  * C-Manjoined
00:46:45  * bnoordhuisquit (Ping timeout: 256 seconds)
01:50:40  * C-Manquit (Quit: Connection reset by beer)
02:44:48  * joshthecoderjoined
04:36:25  * Net147joined
06:53:28  * RT|Chatzillaquit (*.net *.split)
06:53:28  * briancline|awayquit (*.net *.split)
07:11:11  * RT|Chatzillajoined
07:11:11  * briancline|awayjoined
07:14:43  * RT|Chatzilla_joined
07:16:01  * RT|Chatzillaquit (Read error: Connection reset by peer)
07:16:18  * RT|Chatzilla_changed nick to RT|Chatzilla
07:45:16  * joshthecoderquit (Quit: Leaving...)
08:14:09  * rendarjoined
09:19:22  * TheJHjoined
09:35:13  * C-Manjoined
10:25:08  * decoderquit (Ping timeout: 246 seconds)
10:26:38  * decoderjoined
10:39:11  * eoh|quit (Read error: Connection reset by peer)
10:39:23  * eohjoined
11:04:00  * [[zzz]]joined
11:07:15  * [[zz]]quit (Ping timeout: 256 seconds)
11:08:35  * bnoordhuisjoined
11:22:24  * eohquit (Read error: Connection reset by peer)
11:35:49  * eohjoined
12:32:38  <luite>how much overhead (bytes and instructions) can i expect for allocating lots of short lived 4-element arrays? is cleanup free if an object is never promoted?
12:57:12  <mraleph>luite: overhead in memory will be around 6 words. in instructions hard to say without counting. also depends on the way you allocate. (allocation in new space is just a pointer bumping) but you additionally need to initialize the object (all 10 words), plus there might be some small overhead to reach allocation stub in the first place if you do like new Array(4).
12:58:55  <mraleph>luite: reclamation always costs nothing for young objects, and pretty cheap for old objects… however if it survives one minor GC it will be copied, if it survives two it will be promoted — both times you pay something for copying and scanning.
13:01:46  <mraleph>(when I say word I mean "pointer size" not word in terms of say intel refman).
13:13:21  * [[zzz]]changed nick to [[zz]]
14:09:35  * eoh|joined
14:12:51  * eohquit (Ping timeout: 246 seconds)
14:30:51  * Net147quit (Quit: HydraIRC -> http://www.hydrairc.com <- Po-ta-to, boil em, mash em, stick em in a stew.)
14:50:04  <luite>mraleph: sorry was away for a while, thanks!
15:51:41  * RT|Chatzillaquit (Quit: ChatZilla 0.9.86.1 [Firefox 2.0.0.22pre/2009081014])
16:53:09  <luite>i just wanted some ballpark figures, have lots of very small arrays that i statically know the size of, so i can store them in a big array as offsets without overhead, but then i'd need to do the scanning/copying manually in javascript
16:54:42  <mraleph>I see
16:55:48  <luite>and 99% will probably not survive the first collection
17:18:28  * eoh|quit (Read error: Connection reset by peer)
17:18:45  * eohjoined
17:46:24  * bnoordhuisquit (Ping timeout: 248 seconds)
19:13:32  * CodeWarjoined
19:14:05  <CodeWar>constant pool and reloc information, is it feasible to add a 256B padding between code and data of this form?
19:14:09  <CodeWar>(for ARM)
19:16:17  <mraleph>reloc information does not reside in the code object at all.
19:16:32  <mraleph>it should be in the separate byte array floating arround.
19:16:57  * AndreasMadsenjoined
19:17:24  <mraleph>as for constant pool, it might be feasible, though the way constant pool is done is so fragile that I would not promise anything. But it should be doable.
19:18:26  <CodeWar>mraleph, thanks. Essentially the problem is that the ARM instructions JITted by your JIT is further JITted (what fun :-)) by a processor trace JIT and stored in a translation cache on some ARM CPUs (long story)
19:18:40  <CodeWar>writes to data close to code makes us drop these entire translations hurting IPC ..
19:21:02  <CodeWar>the constant pool is it always at the end of functions or do we have code- pool-code-pool for a method?
19:28:14  <mraleph>they are intermixed yeah.
19:29:10  <CodeWar>ouch padding might hurt ifetch
19:29:29  <CodeWar>guess we need to go and fix our coherence logic then, thanks
19:45:35  * CodeWarquit (Quit: Leaving)
20:08:41  * seventhjoined
20:08:58  * seventhquit (Client Quit)
20:09:31  * seventhjoined
20:22:50  * AndreasMadsenquit (Remote host closed the connection)
21:38:06  * bnoordhuisjoined
21:55:20  * rendarquit
22:19:07  * RT|Chatzillajoined
23:59:24  * TheJHquit (Ping timeout: 244 seconds)