#gemrb@irc.freenode.net logs for 21 Jun 2011 (GMT)

Archive Today Yesterday Tomorrow
GemRB homepage


[00:59:40] --> myownlittlworld has joined #gemrb
[02:00:24] <-- myownlittlworld has left IRC (Ping timeout: 240 seconds)
[02:04:03] --> _dhewg has joined #gemrb
[02:08:55] <-- dhewg has left IRC (*.net *.split)
[02:08:56] --- _dhewg is now known as dhewg
[05:21:53] --> barra_home has joined #gemrb
[06:15:14] --> Bo_Thomsen has joined #gemrb
[07:16:20] --> adominguez has joined #gemrb
[07:22:53] --> raevol has joined #gemrb
[07:40:24] <-- adominguez has left IRC (Quit: Saliendo)
[07:42:15] --> adominguez has joined #gemrb
[07:46:39] --> lubos has joined #gemrb
[08:00:02] --> lynxlynxlynx has joined #gemrb
[08:00:03] <-- lynxlynxlynx has left IRC (Changing host)
[08:00:03] --> lynxlynxlynx has joined #gemrb
[08:00:03] --- ChanServ gives channel operator status to lynxlynxlynx
[08:34:30] <-- raevol has left IRC (Quit: Leaving.)
[08:38:09] <-- lynxlynxlynx has left IRC (Read error: Operation timed out)
[09:07:45] <-- Cable_ has left IRC (Remote host closed the connection)
[09:20:18] <-- Bo_Thomsen has left IRC (Quit: Leaving.)
[11:05:47] --> lynxlynxlynx has joined #gemrb
[11:05:48] --- ChanServ gives channel operator status to lynxlynxlynx
[13:59:43] <-- lynxlynxlynx has left IRC (Remote host closed the connection)
[14:53:46] --> brad_a has joined #gemrb
[14:54:47] <brad_a> so what are my existing options for lowering gemrbs memory usage?
[14:57:32] <edheldil> implement MOS tiling and a general cache, probably :)
[15:00:08] <brad_a> ha ha i said existing :P
[15:00:45] <brad_a> not to say that im not willing to attempt that at some point...
[15:06:23] <fuzzie> don't play bg2
[15:11:08] <brad_a> thats an idea
[15:11:30] <brad_a> probably not what jonnypi wants to hear. break his heart.
[15:31:09] <-- lubos has left IRC (Quit: Leaving.)
[15:56:39] --> Beh0lder has joined #gemrb
[15:56:44] <Beh0lder> hi all
[15:57:03] <Beh0lder> I've warned about possible memory problems in ipad, but nobody believed me.)))
[15:58:54] <brad_a> lol i did believe you on first gen ipad which is what I am talking about
[15:59:19] <brad_a> 2nd gen has like 512 ram
[15:59:26] <fuzzie> it seriously runs in 32mb and a bit of swap on the a320
[15:59:42] <Beh0lder> i know
[15:59:47] <fuzzie> well, not bg2, but i already said don't run bg2
[15:59:53] <brad_a> yeah ios 4 has VM but apple has a limit still on how much memory a single process can use
[15:59:59] <brad_a> probably a way to disable that
[16:00:10] <-- Beh0lder has left IRC (Client Quit)
[16:00:14] <fuzzie> i don't know why it's taking crazy amounts now
[16:00:23] <fuzzie> although i do see that we seem to be discarding areas in completely the wrong way
[16:00:25] <brad_a> but im looking for less drastic solutions like the ability to just disable audio or something
[16:00:31] <fuzzie> i need to ask Avenger about that
[16:04:53] <brad_a> i need to check how much less memory is used running at 800x600 instead of 1024x768
[16:05:28] <brad_a> then i could maybe detect what gen ipad they have and force a lower res if that keeps it under the threshold
[16:05:52] <fuzzie> as Beh0lder said, resolution/audio probably make no real difference
[16:06:34] <fuzzie> well, i guess it depends on how your SDL implementation works..
[16:06:40] <brad_a> yeah i figured since the resources take the same memory regardless. i was thinking mayble sdl would require less. it certainly wont take more
[16:06:40] <fuzzie> but i'm pretty sure not
[16:07:26] <brad_a> yeah it uses a whopping 2mb less... 9.9
[16:07:57] --> lynxlynxlynx has joined #gemrb
[16:07:57] <-- lynxlynxlynx has left IRC (Changing host)
[16:07:57] --> lynxlynxlynx has joined #gemrb
[16:07:57] --- ChanServ gives channel operator status to lynxlynxlynx
[16:08:50] <fuzzie> last time i did a massif it was all area tiles and MOS, but i didn't look at it recently, maybe we're doing something dumb again
[16:09:55] <brad_a> disabling audio drops usage by ~5mb
[16:10:10] <brad_a> probably not worth the loss of audio
[16:11:01] <brad_a> tho im only measuring gemrbs memory and not anything else like core audios memory or anything. maybe the savings applies there as well
[16:40:44] <brad_a> if somebody has a min i still have these patches
[16:40:45] <brad_a> http://dl.dropbox.com/u/13866402/sdl1.3patch.txt
[16:40:54] <brad_a> http://dl.dropbox.com/u/13866402/touchscreenconfigpatch.txt
[16:45:45] <lynxlynxlynx> i'll get back to that soon
[16:46:01] <lynxlynxlynx> met avenger irl this weekend :)
[16:46:27] <lynxlynxlynx> and it all started with a stupid fedex quest
[16:53:34] <lynxlynxlynx> http://lynxlynx.info/ie/HPIM0790.jpg <-- we took a picture for you too
[16:59:08] <brad_a> ill bet you goat loads of exp for completing that quest!
[17:11:32] <-- adominguez has left IRC (Quit: Saliendo)
[17:22:20] --> |Cable| has joined #gemrb
[17:34:12] --> Bo_Thomsen has joined #gemrb
[17:44:44] <lynxlynxlynx> actually it ended with me loosing a pretty important item
[17:44:50] <lynxlynxlynx> losing
[17:46:01] <lynxlynxlynx> but i laughed it off, my will saves are good
[18:00:29] --> raevol has joined #gemrb
[18:10:50] --> Beh0lder has joined #gemrb
[18:15:14] <wjp> you and Avenger met up? cool
[18:28:19] <brad_a> beholder: you say BG1 takes the same memory as BG2?
[18:32:40] <-- Beh0lder has left IRC (Quit: Beh0lder)
[18:37:57] --> Beh0lder has joined #gemrb
[18:37:58] <Beh0lder> I don't know how memory needed for BG2. But BG1 eats up to 130Mb. I checked.
[18:40:14] <Beh0lder> You may test it. Maximum memory usage present if you go out from Candlekeep. (Gorion death scene)
[18:44:02] <fuzzie> hmm
[18:44:10] <fuzzie> and it stays that high?
[18:44:24] <Beh0lder> more then 100Mb
[18:44:35] <fuzzie> hmph
[18:44:43] <fuzzie> i can only assume DelMap is failing for some reason
[18:45:06] <Beh0lder> but if I save and load game mem usage is more small
[18:45:23] <fuzzie> let me see if it changed in the last year or so
[18:46:44] <fuzzie> maybe CanFree is returning false due to IF_ACTIVE or something
[18:47:25] <Beh0lder> May be it changes in last versions, I checked it in 0.6.3
[18:47:52] <fuzzie> in Game.cpp, you could add a printf in the (!map->CanFree()) check and see if it fails there
[18:48:45] <fuzzie> i mean, in DelMap
[18:50:00] <brad_a> well in BG 2 i load the autosave at the very beginning and its using 115
[18:50:09] <brad_a> right from the get go
[18:50:15] <fuzzie> ok, it really shouldn't do that
[18:50:31] <brad_a> i should hope not
[18:50:39] <brad_a> but im largely ignorant
[18:50:43] <fuzzie> it really uses 40mb or so max for me there
[18:51:06] <brad_a> you are talking to me correct?
[18:51:17] <fuzzie> hehe, yes :-)
[18:51:21] <Beh0lder> I may check in Windows
[18:51:24] <fuzzie> it's strange
[18:51:35] <fuzzie> the trouble is, i have no real time right now, so i can't experiment
[18:51:38] <brad_a> let me check the os x build that is using sdl 1.2
[18:52:05] <fuzzie> well, actually, maybe scripts run *after* the autosave?
[18:52:20] <brad_a> btw music stopped working on the os x build but works fine on ios. no idea why
[18:52:22] <fuzzie> there is a problem with the bg2 intro because it loads several areas at once
[18:52:37] <brad_a> maybe im not compiling a needd file or missing a plugin
[18:53:04] <fuzzie> openal on OS X used to be pretty difficult
[18:53:15] <brad_a> well be that as it may on os x it only uses 90 mb
[18:53:24] <fuzzie> that is still crazy. hmph.
[18:53:32] <fuzzie> what if you make a new save and load that?
[18:53:37] <brad_a> well the thing is audio was working fine before i moved to the latest git
[18:54:06] <fuzzie> oh, is it just music missing?
[18:54:10] <brad_a> yes
[18:54:24] <brad_a> and sometimes the voices
[18:54:34] <fuzzie> maybe we added something new, i'm honestly not sure
[18:55:26] <brad_a> [OpenAL]: Unable to create new source: 0xa003 [ERROR]
[18:55:27] <brad_a> Error: MVE player couldn't open audio. Will play silently.
[18:55:37] <fuzzie> hmm
[18:55:41] <fuzzie> which openal library are you using there?
[18:55:47] <brad_a> that is the same error when the voices fail
[18:55:50] <fuzzie> the OpenAL.framework?
[18:56:10] <brad_a> the openal framework that ships with os x yes
[18:56:18] <Beh0lder> BG 1 after loading save - 63Mb usage, after move in another area - 82Mb (windows)
[18:56:23] <brad_a> like i said tho ios is fine
[18:56:35] <fuzzie> *not* the macports one?
[18:56:41] <brad_a> nope
[18:56:50] <brad_a> apple supplied for both
[18:57:41] <fuzzie> if i load bg2, and load a save, gemrb uses 22mb here
[18:58:19] <brad_a> about the momory ussage on os x i should mention that my build include all the appkit overhead as im not building a commandline executable anymore
[18:58:22] <Beh0lder> odd
[18:58:41] <brad_a> so 90 mb is probably expected
[18:58:45] <fuzzie> i am not using git head though, let me try it
[19:02:28] <-- Beh0lder has left IRC (Quit: Beh0lder)
[19:03:34] <-- Gekz has left IRC (Ping timeout: 276 seconds)
[19:09:22] <fuzzie> in fact gemrb is not even working for me
[19:10:30] --> Maighstir has joined #gemrb
[19:10:57] <fuzzie> oh, right, because everyone decided it would be awesome fun to break the GameOnCD code
[19:12:48] <fuzzie> ok, gemrb with git head is using 35mb if i load a save and 80mb if i move areas, ouch
[19:14:17] <brad_a> thats less than im getting on save loading alone
[19:14:34] <wjp> 35MB RSS?
[19:15:06] <fuzzie> yes, 35mb RSS, ~33mb allocations
[19:16:12] <wjp> seems to depend a lot on the area for me
[19:16:16] <fuzzie> i really don't want to start messing with heap debugging now
[19:16:26] <wjp> just tried two saves; one got to 45MB, one to 69MB in BG2
[19:16:48] <brad_a> I have left gemrb running in the background and its memory has dropped to ~70. so thats better still wired to me that it starts high then drops
[19:17:55] <fuzzie> last time i tried, the highest user was TIS (so, depends on area size), then MOS (fairly constant), then BAM (depends on actors present), then SDL overhead
[19:18:58] <fuzzie> however it was quite likely without sound turned on.
[19:19:20] <brad_a> sound for me under open al seems to only take ~10 mb
[19:20:04] <fuzzie> it was 48mb total allocations (when loading Trademeet), so 10mb is quite a lot on top
[19:20:16] <fuzzie> but i can only assume that we're failing in DelMap or something
[19:20:28] <fuzzie> also that you all have crazy memory-eating viruses
[19:20:54] <wjp> it also rather depends on how you measure memory usage
[19:21:58] <fuzzie> well, presumably RSS really ought not to be that much higher than actual malloc/new calls
[19:22:48] <brad_a> what is RSS? excuse my ignorance :P
[19:23:22] <fuzzie> 'resident set size', the amount of actual RAM a process is using
[19:23:29] <fuzzie> RSIZE in OS X's top, I think
[19:24:09] <brad_a> ah let me see if thats comparable to what ive been quoting from AM
[19:25:22] <brad_a> yeah it looks like rsize from top is what activity monitor is telling me
[19:25:40] <brad_a> says 92+ in top
[19:26:20] <brad_a> waht is #TH in top # threads? how is it possible to have 4/1 in that col?
[19:26:41] <brad_a> everything else is just a single number
[19:27:13] <fuzzie> no idea
[19:29:51] <brad_a> uppon quitting im getting more openal errors on os x
[19:29:58] <brad_a> Unable to stop source: 0xa003 [WARNING]
[19:30:18] <brad_a> all open al erros have that save hex value
[19:30:33] <brad_a> i have no idea what that means
[19:31:15] <fuzzie> i think it's AL_INVALID_VALUE, which is an amazingly useless error
[19:31:37] <brad_a> ill have to brows the changes since 0.6.4 to see what happened to cause this to fail between now and then
[19:31:44] <wjp> fuzzie: yes
[19:32:34] <brad_a> still odd that it works on iOS but not OS x. maybe ill static link os x to see what effect if any that has.
[19:33:56] <fuzzie> iOS's openal is quite different
[19:34:32] <fuzzie> it would help if the openaldriver differentiated between AL and ALC errors
[19:35:35] <brad_a> you seem to know more about iOS then i do!
[19:36:08] <brad_a> not that i claim to know alot or even claim to be an iOS dev
[19:37:55] <fuzzie> rest assured that I know nothing useful :)
[19:39:11] <brad_a> whatever! ill bet you are the "guest" that ported gemrb to iOS first then decided to troll everyone by not releasing it ;)
[19:41:55] <fuzzie> i think if i had a shiny fun iPad to play with, you'd never see me again
[19:43:54] --> Beh0lder has joined #gemrb
[20:13:46] <-- Beh0lder has left IRC (Quit: Beh0lder)
[20:14:40] <lynxlynxlynx> brad_a: that sdl patch looks wrong
[20:15:14] <lynxlynxlynx> doesn't the brace removal cause a compiling error?
[20:17:37] <lynxlynxlynx> the other one makes no sense without the code that actually implements those options
[20:18:20] <lynxlynxlynx> and it should also modify GemRB.cfg.noinstall.sample
[20:19:04] <brad_a> one of those options is for beholders patch the other i still need to submit to you
[20:19:22] <brad_a> the sdl one compiles for me on both 1.2 and 1.3 are you sure?
[20:19:41] <brad_a> anyhow im late for school but ill read the chat log after
[20:19:54] <-- brad_a has left IRC (Quit: brad_a)
[21:27:23] <-- barra_home has left IRC (Quit: Verlassend)
[22:24:12] <-- lynxlynxlynx has left IRC (Remote host closed the connection)
[23:31:28] <-- Maighstir has left IRC (Read error: Connection reset by peer)