Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2019-08-10

00:02:02Bilgusspeachy as long as they'll boot thats all I really need though I suppose you need the dpad to get into OF so nm
00:36:18 Quit ulmutul (Quit: Leaving)
00:37:16 Quit ZincAlloy (Quit: Leaving.)
00:47:15***Saving seen data "./dancer.seen"
01:13:53 Quit lebellium (Quit: Leaving)
02:09:32*__builtin stumbles upon pocket quake
02:09:42__builtin"fixed-point optimizations"
02:40:00foolsh__builtin: still got the pocketpc device I gave u?
02:40:11*__builtin has it somewhere
02:40:53__builtinright now I'm just merging some of PQ's changes into our quake
02:41:07*foolsh still has a horde of them
02:42:02foolshCan't find the source for quakeDS though, they must have optimized the crap out of that
02:42:58__builtinit's on archive, I think
02:43:06__builtinI read it a bit, but it's actually not that much use
02:43:10__builtinthey're using GL
02:43:52foolshhmm did not know the toolchain for DS had hardware api calls
02:44:36foolshno wonder it can run on the DS
02:47:18***Saving seen data "./dancer.seen"
03:00:08*__builtin has to say... this is really nice :)
03:00:23__builtinall I have to do is copy someone else's code into our quake
03:16:43__builtindamn, that's a 34% performance increase right there
03:18:39foolshI was poking at the atari2600 emulator today, for some reason button presses stopped working, it had "if (buttons & JOY_LEFT)" etc which didn't make sense to my meager C understanding, so I changed it to "if (buttons == JOY_LEFT)" and now the buttons are hit and miss sorta works on my c250 but not my clip zip
03:19:09foolshscratching my head on that one, it should have worked
03:19:58foolsh34% is very specific
03:20:08*foolsh wants to see __builtin's math
03:20:34__builtinquake has a `timedemo' command which runs a specific demo at max speed and times it
03:20:46foolshAH! very useful
03:20:50__builtinbaseline is 9.7FPS (before these current patchsets)
03:21:02__builtinfor demo1.dem, that is
03:21:08__builtinit's now getting 13.0FPS
03:21:54*foolsh was only kidding :) but is impressed
03:24:02__builtinthis feels far more playable :)
03:24:12__builtintoo bad there's some mipmapping that degrades rendering quality a little
03:36:48 Join massiveH [0] (
03:38:48fs-bluebotBuild Server message: New build round started. Revision 4397194, 280 builds, 11 clients.
03:58:46fs-bluebotBuild Server message: Build round completed after 1198 seconds.
03:58:47fs-bluebotBuild Server message: Revision 4397194 result: 12 errors 0 warnings
04:05:09__builtinnot my fault this time :)
04:47:21***No seen item changed, no save performed.
05:05:56fs-bluebotBuild Server message: New build round started. Revision a9a891b, 280 builds, 11 clients.
05:28:05 Quit TheSeven (Disconnected by services)
05:28:14 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
06:47:24***Saving seen data "./dancer.seen"
08:17:26 Quit [7] (Ping timeout: 264 seconds)
08:22:01 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
08:47:28***Saving seen data "./dancer.seen"
09:01:26 Join ZincAlloy [0] (~Adium@2a02:8108:9440:dfc:f94d:c363:b7b0:1699)
09:37:48 Join lebellium [0] (
09:58:38 Join dys [0] (
10:47:30***No seen item changed, no save performed.
11:28:08 Join ulmutul [0] (~ulmutul@rockbox/developer/ulmutul)
11:51:08ulmutulfoolsh: don't put too much work in the arari emulator, I already worked upon it a lot. My version now works stable as far as I can tell. However I completely turned the code upside down and rewrote a lot of stuff, so it wouldn't make sense to merge it into the base code, rather create a new patch. I have to clean up my code a lot though.
11:53:27ulmutulI first removed the warnings from the original code, then removed unused parts of the code, moving things around, renaming things etc. and hoped that the bugs popped out. They didn't though.
11:58:30ulmutulThen I rewrote parts of the code, and an idea occured to how to better handle the rendering of sprites. A clean approach, that especially would speed up the moving a lot! I learned however that the moving part is insignificant the way atari 2600 programs work, so I ended up at about the same emulation speed.
12:03:48ulmutulAnd the initial "cleanness" is also gone (I had to add something that made it slower, and now have done things slightly different to catch up the emulation time).
12:06:32ulmutulSo what's left is: an emulation that runs stable (some graphical glitches still left though), but too slow for my needs (I want to have it working in rael time on my PP5020 targets. Emulation time per frame is about 40ms though, without rb->lcd_update)
12:10:51ulmutulI already added some neede UI for the user, like VCS switch setting, setting bank switching, setup keys (like rockboy) and some profiling and test code.
12:14:17ulmutulLast thing I did was changing the cpu core and run it against a test suite to ensure correctness.
12:16:21 Quit ulmutul (Quit: Leaving)
12:47:34***Saving seen data "./dancer.seen"
12:55:21Bilgusulmutul are you clearing the screen?
12:56:04Bilguslike are you doing a whole frame update each time or incremental scrween update?
14:30:05 Quit Rower (Read error: Connection reset by peer)
14:30:07 Join Rower- [0] (
14:31:39 Join Rower [0] (
14:34:35 Quit Rower- (Ping timeout: 245 seconds)
14:35:02 Join Rower- [0] (
14:38:09 Quit Rower (Ping timeout: 244 seconds)
14:47:38***Saving seen data "./dancer.seen"
15:13:02 Quit copper (Remote host closed the connection)
15:30:34 Join copper [0] (~copper@unaffiliated/copper)
16:47:41***Saving seen data "./dancer.seen"
18:04:36 Join Rower [0] (
18:07:30 Quit Rower- (Ping timeout: 258 seconds)
18:47:44***Saving seen data "./dancer.seen"
20:47:48***No seen item changed, no save performed.
22:35:05 Join IncognitoMan_ [0] (~Incognito@unaffiliated/incognitoman)
22:39:05 Quit IncognitoMan (Ping timeout: 246 seconds)
22:47:52***Saving seen data "./dancer.seen"
23:11:59 Join ulmutul [0] (~ulmutul@rockbox/developer/ulmutul)
23:14:37ulmutulBilgus: I'm doing a whole screen update. First rendering everything to an internal "atari-framebuffer", and on every vsync writing to rockbox framebuffer with scaling and everything.
23:39:23__builtinulmutul: I looked at your g#1665
23:39:26fs-bluebotGerrit review #1665 at : Add support for Windows shortcuts (*.lnk files) by Sebastian Leonhardt
23:43:28ulmutulThanks! Have you looked at the manual entry? Is the wording OK?
23:47:12Bilgusulmutul I was going to say you can speed up rendering by only updating portions of the screen each time but being a emulator it might be hard to decide what needs updated
23:48:17Bilgus__builtin I think you are right on the bufferlib for overlays I don't think lua has a fully implemented overlay looking at pictureflow as an example

Previous day | Next day