#exult@irc.freenode.net logs for 25 Sep 2010 (GMT)

Archive Today Yesterday Tomorrow
Exult homepage


[01:55:15] <-- Sevalecpp has left IRC (Quit: Leaving)
[01:59:17] --> Sevalecpp has joined #exult
[03:11:43] <-- Morde has left IRC (Read error: Connection reset by peer)
[03:27:38] <-- Marzo has left IRC (Ping timeout: 245 seconds)
[06:20:36] <-- RadoS has left IRC (Remote host closed the connection)
[06:22:09] <-- jvlee has left IRC (Quit: jvlee)
[10:23:06] --> Marzo has joined #exult
[12:36:42] <Dominus> colourless: http://vogons.zetafleet.com/viewtopic.php?p=198514#198514 question about our fork of munt
[13:56:45] <-- Kirben has left IRC ()
[16:09:53] --> Morde has joined #exult
[16:54:07] <-- Morde has left IRC (Read error: Connection reset by peer)
[17:43:08] --> Morde has joined #exult
[18:03:00] <-- Morde has left IRC (Read error: Connection reset by peer)
[18:38:08] --> Morde has joined #exult
[18:47:46] <-- Morde has left IRC (Read error: Connection reset by peer)
[21:51:34] <Colourless> our fork of munt has really trivial changes done to it. it really should be updated at some point
[22:00:03] --> Testerrr has joined #exult
[22:00:41] <-- Testerrr has left #exult
[22:08:27] <Dominus> heya :)
[22:19:20] <Colourless> as far as i know, the changes are only so the code sits well with the way exult and pentagram do things like data loading, and doesn't step on anything else in the code. May have made some minor changes to other things but can't remember
[22:20:26] <Dominus> I remember you talking about stuff they were doing strangely or wrong... but then I don't remember good as wel :)
[22:22:18] <Dominus> while you are here, what about deactivating the Alt+ +/i key combination in Exult. With your rendering changes this doesn't work well at all.
[22:32:40] <Dominus> as for munt, the update would probably need to be a mixture of actual munt code, ScummVM and Exult changes to the original code :)
[23:30:33] <Marzo> Colourless: regarding the new munt code: I had already tried merging it earlier this year into Exult when I saw that they had updated their code; but the new version was far more CPU intensive and stuttered heavily in my computer
[23:30:53] <Marzo> It is probably more accurate, but I decided to leave the one that was working correctly
[23:32:44] <Dominus> there was a message about the higher CPU load on vogons and why... looking for it right now
[23:35:48] <Dominus> ah found it http://vogons.zetafleet.com/viewtopic.php?t=25023&sid=774fbb971170226f3a3f8ab43c0d4f71
[23:36:02] <Dominus> We had a time last year where development on MUNT was improving rather rapidly. To improve our analysis of the few remaining parameters it was decided to switch all math from fixed point (intergers only) to floating point. This had a pretty signficant performance penalty--even on very modern CPU's.
[23:37:43] <-- Dominus has left IRC (Remote host closed the connection)
[23:37:54] --> Dominus has joined #exult
[23:37:54] --- ChanServ gives channel operator status to Dominus
[23:38:22] <Marzo> I had figured out it was the use of floating point already
[23:39:10] <Marzo> Back then, I had even tried optimizing their use of it by using several floating point tricks that made a minor improvement in performance -- but it just wasn't enough, sadly
[23:39:17] <Dominus> :) you just triggered my memory that I had read this before about the high cpu usage. had to look it up :)
[23:40:21] <Marzo> About the only thing I didn't try that might improve performance would be to use a fast high-quality sine wave generator instead of standard sines and cosines
[23:42:24] <Marzo> (example of floating point tricks: change all exponentiation to base 2 because there are very fast high-precision functions to do 2^x with floating point numbers)
[23:43:22] <Marzo> It is sad, because it sounds sweet -- I did manage to dump some audio to a file and listen to it without the stutter, and it sounded way better
[23:58:00] --> Kirben has joined #exult
[23:58:00] --- ChanServ gives channel operator status to Kirben