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 2005-02-20

00:00:00HClthey hide its uglyness with defines
00:00:10HClwell, once the basic idea its there its fairly simple
00:00:14amiconnThe cpu emu as-is is really slow.
00:00:16HClhold on, let me cough you up a tarball
00:00:20Patr3ckNo one else got the floating point exception error for x11 iriver sim?
00:00:38rasherWorks For Me[tm]
00:00:46rasherNot that that's any help
00:00:54rasherBut at least it isn't *totally* broken
00:01:07pillopreglow: I thought that too... that's why I'm here :)
00:01:20HClits not finished
00:01:26rasherPatr3ck: did you update from cvs recently?
00:01:38HCland it'll scream "ah! unimplemented opcode!" if you try to run it, or at least, it should.
00:01:38Patr3ckTwo minutes ago
00:01:46amiconnHCl: When there is no graphical output, the cpu core does ~32000 z80 insns per 'frame'. From the framerate I get on archos, I calculated the average host cpu clocks per emulated cpu clock
00:02:02HClamiconn: i think dynarec will make a massive speed improvement
00:02:09amiconnGuess what: 100 (!) clocks!
00:02:25rasherThat's a whole bunch of clocks.
00:02:31HClfor at least all the loading instructions, 1 z80 instruction -> 1 m68k instruction
00:02:51HClthat is, if it'll work.
00:02:53amiconnDo you hold all z80 registers in m68k registers?
00:02:57 Join Diway [0] (~diway@
00:02:59HCli'm having debug nightmares over this already.
00:03:06HClwell, most.
00:03:13HCli don't have the flags
00:03:17amiconnThen this should be possible on sh1 as well
00:03:25HCli have 1 8bit reg in each 32bit reg
00:03:30HClits inefficient, but its faster.
00:03:46HClinefficient spacewise anyways
00:04:02amiconnI thought about making a highly optoimised interpretive core, and my first thought was to only hold certain registers in host registers
00:04:07HCltake a look at it...
00:04:18*HCl scratches his head and hmz.
00:04:31HCloh, delete cpudyna.c
00:04:39HCli split that into cpu.c and dynarec.c
00:04:47amiconnI could always move more z80 regs to host regs later
00:04:50HClcoding dynarec = fun
00:05:09HCltake a look at dynarec.c..
00:05:24preglowamiconn: how many players are with character cell lcd displays? just the archos player?
00:05:26amiconnBtw, number of m68k regs ~ number of sh1 regs, so porting shouldn't be that hard
00:05:43amiconnpreglow: As of now supported, yes.
00:05:56preglowpillo: i really think it'd be worth it, yes :)
00:06:20HClthe getupper/putupper is legacy from my old 2-gameboy-regs-in-one-m68k-reg implementation
00:06:26amiconnThere once was neo support in rockbox, but it was dropped. It never really worked, and the neo guys forked some time ago
00:06:37amiconnThe neo also had a charcell display
00:07:17amiconnI'd vote for having one single viewer plugin for both display types
00:08:47Patr3ckI just tried again from scratch, same error, strange
00:08:48pilloaha ok
00:08:58pillobut that's going to be a bit cumbersome... i'll try
00:09:08preglowamiconn: even thought the two wouldn't really share much code?
00:09:30amiconnIf done properly, they should
00:09:46Patr3ckWhat exception, I thought rockbox is pure c code
00:10:30pilloamiconn: i'll think about it a bit more then
00:10:57pillothanks, and keep up the good work!
00:10:58amiconnSimply assume width=1 for all glyphs on a charcell display, drop the calls to functions which calculate the glyph widths, and you're done
00:11:06rasherMy understanding is that the policy is to keep it C unless the benifits from using host-specific asm outweigh the disadvantages
00:11:43amiconnpillo: I planned to look into the viewer myself, but did not yet find the time to do it.
00:11:46preglowrasher: that should be common sense as well as policy ;)
00:12:20Patr3ckrasher: The error looks like a c++ exception ?!?
00:12:20pilloamiconn: good, i'll see what I can do!
00:12:45pilloHCl: can't wait for the rockboy to rock ;)
00:12:57pillonight all
00:12:59rasherPatr3ck: oh my.. sounds ... interesting.. maybe it's not even rockboxui exploding?
00:13:29amiconnpillo: I used to have a list of issues I wanted to fix, but can't find it atm
00:14:00Patr3ckrasher: maybe the codecs that were added, it stopped working for me at that time I think...
00:14:26rasherhrm.. maybe try doing a checkout before the codecs were added?
00:14:45Patr3ckI know I should start gdb now and take a real look where this exception comes from...
00:14:56preglowPatr3ck: codecs _shouldn't_ use floating point, at least
00:15:15preglowbut then again, no parts of rockbox should, heh
00:15:18Patr3ckRight, integer arithmetric and stuff...
00:15:25preglowbut yes, gdb
00:16:12Patr3ckOK, it will be my first time... always worked with vc++ until I met rockbox
00:16:36amiconnHCl: My thoughts of which registers should always be in host registers during emu: PC, A, and flags
00:16:51amiconnThe correct flag emulation is rather important for z80
00:17:21 Quit pillo ("CGI:IRC")
00:18:04 Join bobx2001 [0] (
00:18:05HClwell, PC isn't needed in dynarec
00:18:15HCli'm not sure what i'll do about flags just yet
00:18:25bobx2001rockbox for ioid now
00:18:44bobx2001ipodlinux is to slow
00:19:06rasherwell get hacking.. or convince the ipl people to do the port..
00:19:14rasherI think those are your options
00:19:27bobx2001the ipod linux guys have done most of the hacking
00:19:31rashersorry.. I'm a bit cranky.
00:19:35HClgoddammit, its cold here >.<;; the heating is out
00:20:09rasherhow unfortunate
00:20:22 Quit CoCoLUS ()
00:20:40 Quit bobx2001 ()
00:22:08preglowKILL ME
00:22:16 Join bobx2001 [0] (
00:22:30bobx2001i am back
00:22:46preglowbobx2001: so we see
00:22:51amiconnHCl: I have an asm-optimised implementation of updatepatpix() now. Doesn't push the frame rate significantly... at least not with the 2 roms I tried
00:23:18bobx2001what applications are you guys making?
00:23:25bobx2001i know your making rockbo
00:23:32bobx2001but what is it going to do
00:23:38bobx2001game ?
00:23:41preglowbobx2001: everyone's working on different things
00:24:16amiconnHCl: However, I did a slight alteration that pushes the framerate by >5 %. And I made rockboy play more fair in the multithreaded rockbox....
00:25:38HClyea, it doesn't do threading yet
00:25:53HCli'm kind of wanting it in cvs so we can easily merge changes..
00:26:48amiconnUnfortunately the simulator builds on cygwin are totally broken with the new build system, so I cannot test anything simulator-related until it is fixed
00:26:49HCli'm lucky.
00:26:59HCli just did an rm * in my plugins dir by accident.
00:27:12HClbecause it wasnt -r, it didn't delete my rockboy sources
00:27:17HCl*checks out again*
00:27:21rasherCareful tiger :)
00:27:46amiconn...and without being able to test, I do not want to mess up cvs...
00:28:20amiconnI tried fixing that together with Bagder yesterday, but we didn't find a solution yet :(
00:28:38preglowthe clashing headers issue?
00:28:55preglowwouldn't the only real solution to that be renaming the rockbox ones?
00:29:28HCli pretty much agree.
00:29:35HClhaving rockbox_stdio.h
00:29:36amiconnI think that should fix the problems once and for all. However, I don't know whether this is desired
00:29:39HClis much clearer anyways
00:30:04HClproblem is with porting you'd have to adjust everything
00:30:10amiconnI just had another idea, please correct me if I'm wrong:
00:30:14preglowamiconn: i can't see how anyone can not desire it, it is cleaner both semantically and practically, but yes, lots of work
00:30:51amiconn- The header files are already grouped by function.
00:31:43amiconn- I suggest splitting that a bit more, into those that contain rockbox specifics and those that define posix functions
00:32:02preglowthe posix ones could do a ifdef check
00:32:22amiconnpreglow: Unfortunately not.
00:32:28preglowwhy not?
00:33:03amiconnBecause some typedefs in the system includes do not provide magic defines to protect from redefinition
00:34:24amiconnBy having the posix includes in separate dir(s), we could easily select between rockbox posix and system defines by adding/not adding the rockbox posix defines to the include path
00:36:31preglowsounds pretty good
00:36:37preglowand clean
00:37:10amiconnOne problem remains - functions that have the same name, but different syntax/usage.
00:37:35preglowrename them :p that should be avoided anyway, heh
00:37:51amiconnThe only function that i remember causing this is sleep()
00:38:37 Quit quelsaruk ("KVIrc 'Realia'")
00:38:56amiconnPosix sleep(n) is sleep n seconds (and wake up by an event). Rockbox sleep(n) is sleep n*10 ms
00:40:22preglowyes, that's not good
00:40:23rasheroh my
00:40:31preglowit should not have been called that
00:40:38rasherthat's an odd definition of rockbox sleep(n)
00:41:17amiconnpreglow: I agree. Will discuss that with Bagder asap.
00:41:45amiconnThen I face the tedious task to walk through all rockbox headers and separate them...
00:41:58preglowit very probably has to be done
00:42:29amiconnHmm. I'd need a build envy on Linux, or someone who has one to test my changes....
00:42:51amiconncygwin is obviously not very posix compliant....
00:42:59rasheramiconn: I do
00:43:02rasherand I could
00:43:45amiconnOkay. I'll talk to Bagder when he's around.
00:44:09amiconnHCl: I'll work on the archos overlay loader in the meantime. That shouldn't be hard
00:44:35HCloverlay loader? you mean loading rockboy into mp3 buffer
00:44:58 Join gromit` [0] (
00:45:21Patr3ckhm, I tried to debug my floating point exception with gdb, but I don't think I reach the main function
00:45:40amiconnI simply need to define a little header format, that includes a magic, the load address, the length, and the plugin api version it is compiled for
00:46:26amiconnMy loader plugin would then load the header, claim the mp3 buffer, check whether the overlay fits, then load it and run it.
00:46:49HClohyea.. hm.
00:46:54HCldynarec needs a fair amount of memory
00:46:56thegeekhmm, would it be a big job to migrate away from using the mp3buffer later on?
00:47:03HCldunno how well the archos will do on that aspect
00:47:09Patr3ckgdb stops with Program received signal SIGFPE, Arithmetic exception.
00:47:18thegeekwould be a shame not beeing able to listen to mp3's at the same time as playing gb(if the dynarec works as well as it should;)
00:47:24amiconnThe malloc() implementation within rockboy will then need small adjustments to not use the whole buffer.
00:47:31Patr3ck0xb7ff390e in ?? ()
00:47:35amiconnOtherwise rockboy may overwrite itself...
00:48:11preglowthegeek: haha, please don't expect that without frameskip
00:48:17preglowat least the way things are looking right now
00:48:18thegeekI'm not
00:48:22Patr3ckbacktrace shows alot of hex codes and ??
00:48:30preglowPatr3ck: have you built with -g
00:48:32thegeekwith that super-optimized mp3decoder posted yesterday
00:48:36thegeekit might be possible
00:48:46thegeekI would sacrifice a bit of soundquality just while I'm playing
00:48:48preglowthegeek: it isn't super optimized, but it's better
00:48:49amiconnthegeek: I will certainly be impossible on the archos, and imho not feasible on iriver as well
00:48:51Patr3ckI tried with configure D
00:48:53HClwhat super optimized mp3decoder?
00:49:00amiconns/I will/it will/
00:49:02thegeekI thought it was pure asm
00:49:16preglowthere was lots of asm, yes
00:49:47thegeekwould it be a lot of work porting that?
00:50:02Patr3ckI thought I spotted an option check for that in the configure script
00:50:05preglowit's 68k asm :P
00:50:05thegeekif it produces fair quality, it could be usefull sometimes
00:50:14preglowunless we're talking about different decoders here
00:50:36thegeekI got it from reading the log
00:50:48thegeekwas in connection with some hw mod
00:52:08thegeekthat one
00:52:17preglowyes, that one
00:52:24pregloweven that's not realtime at 90mhz
00:52:52thegeekI thought it was superfast
00:52:56thegeekguess I got it wrong
00:52:57preglowi'm starting to wonder what the hell kind of magic motorola has done on their decoder, yes
00:53:17thegeekit must be pure asm
00:53:22Patr3ckpreglow: How can a debug version of the sim be built?
00:53:39preglowPatr3ck: wouldn't know, but for symbols, -g in gcc command line is enough
00:53:45 Quit bobx2001 ()
00:54:22Patr3ckpreglow: I know, but the command line isn't shown when building
00:54:41Patr3ckpreglow: There must be some configure option I think
00:54:51preglowPatr3ck: probably, i've never built sim
00:55:07Patr3ckOK, thanks anyway
00:55:29*HCl goes to test his dynarec on his iriver
00:56:33thegeekyou have the framwork done already?
00:57:39HClhrm. i'm getting the feeling rockbox isn't shutting down properly on my iriver
00:57:47HClthis is the second time i try to turn it on and its batteries are empty.
00:57:50***Saving seen data "./dancer.seen"
00:57:55HCldoes rockbox turn off after so many seconds unused?
00:58:35preglowmine doesn't
00:58:38preglowit used to, but not anymore
00:58:54preglowthe drive always turns off, though
01:00:30Patr3ckI have just seen that -g is always used in the x11 sim
01:00:38HClsomething went most definately wrong..
01:01:05HClrockbox froze on loading
01:01:57HCland the iriver firmware refuses to boot with the low battery..
01:02:47HCland yea, i have the framework pretty much done
01:02:54HClthere's just a little issue when calling dynarec blocks
01:02:59HCland the recompiler isn't done yet
01:04:05preglowHCl: did you do the "hash" and linked list thing?
01:04:27HClwell, linked list isn't created yet cause thats the recompiler's job
01:04:34HClbut its implemented
01:04:52 Quit gromit` ("Leaving")
01:06:37 Quit methangas (" HydraIRC -> <-")
01:09:12HClcan anyone confirm that the rockbox in current cvs won't boot correctly?
01:09:31HClor at least the one in
01:09:37HClon iriver, that is
01:12:47amiconnHCl: Do you know which gameboy buttons are the least important, i.e. seldomly used?
01:12:59HClthey're all important.
01:13:01HClstart isn't used much.
01:13:11HClits generally for pause / menu
01:13:30HClmost games work with A and B mostly
01:13:31amiconnHmm, that's not good :( Otherwise rockboy could work on Ondio too. It would be even ~10 % faster
01:15:15HClwho has an iriver and can test on my ftp for me ?
01:15:26rasherhang on
01:15:34HClit might freeze your iriver, so make sure the original firmware works
01:15:50rasherHeh, it does
01:15:52HCljust wanting to know whether its my battery level
01:15:57HClor the compile
01:16:56Patr3ckHCI: it is not booting...
01:17:03rasheractually, I'll pass
01:17:12HClokay.. then my question is, whats wrong with cvs
01:17:20rasheroriginal firmware not booting :D
01:17:32rasherI think there's slight corruption on my firmware
01:17:41HClyea, run scandisk first
01:17:49HClthats why i said make sure the original firmware is working :P
01:17:51rasherI had that earlier this evening
01:17:52preglowrasher: seems like a lot og people are getting this now, if it's stuck at 'read file system'
01:17:57rasherit is
01:18:01rasherresetting worked last time
01:18:05preglowyes, you're the third i've heard of
01:18:14Patr3ckHCl: Your is not booting
01:18:20HClPatr3ck: okay.
01:18:27HClthen something is wrong with current cvs...
01:18:33HCl :(
01:19:45HCli'm glad my original firmware booted, heh, otherwise i'd have a bricked player now
01:20:20Patr3ckHCl: I don't think so, I just recompiled cvs from scratch and it worked
01:20:48HClthats rather odd since i don't have any changes aside from a slight makefile adaption, and an plugin api extension
01:24:41amiconnHCl: Tried 'make clean' before recompiling?
01:25:53HClamiconn: ofcourse, i cleared out my entire object dir before compiling
01:26:12HCli'm re-doing my api extensions, making sure there aren't any other difference
01:26:31amiconncvs diff may tell you something
01:28:30preglowi've got a fairly recent compile here
01:28:34preglowlet's say about six hours
01:28:37preglowworks like a charm
01:28:43*rasher prods his iriver
01:28:47rasherI guess it's time for a scandisk
01:29:54preglowHCl: i built from cvs around 18:30, works as a charm
01:30:09rasherrockbox working fine
01:30:59rasher"Windows was unable to complete the disk check"
01:31:11rasherWhat now, do I need to be Administrator to do Scandisk on an external drive!
01:31:18rasherIf I do, I'll kill.
01:31:28*rasher punches randomly
01:32:16preglowif you intend to kill, please let me know of it in advance, i've got a list, you see
01:32:34 Quit DrRick ()
01:47:27HClpreglow: yea, but there have been changes since then
01:47:35rasherho hum
01:47:41rasheririver still not booting
01:47:43preglowHCl: yes, just wanted to help you narrow it down
01:47:49HClthanks :)
01:55:08 Quit linuxstb (Read error: 60 (Operation timed out))
01:56:40*HCl tries 3.4.3 instead
01:58:33HClcan someone compile current cvs with 3.4.x and give me an md5sum of rockbox.iriver?
01:58:55preglowgimme a sec
01:59:48 Join linuxstb [0] (
01:59:57HCl1d76b73d8e9e9b8eea79c69a80735c27 rockbox.iriver
02:00:10preglowbfc6d21b942794b737ac82377bbaa55c rockbox.iriver
02:00:18HClwhat compiler do you have?
02:00:24preglowthomj@glow build $ m68k-elf-gcc −−version
02:00:25preglowm68k-elf-gcc (GCC) 3.4.3
02:00:41HClas far as i know i have a current cvs checkout with no changes whatsoever..
02:00:50preglowi checked out one minute ago
02:01:10preglowdo a cvs diff and check
02:01:16preglowi've only got diffs in some plugins
02:01:18preglowwhich is my code
02:01:33rasherI'd love iriver firmware to boot
02:01:47preglowrasher: it will, eventually :)
02:01:57HClhm, wait, i made a slight typo
02:02:01HCllet me fix that and try again
02:04:25 Quit Sucka ("( :: NoNameScript 3.81 :: )")
02:04:51HCli still don't get yours.
02:04:56HCllet me completely take out all my changes
02:05:24HClpreglow: does calculator.c compile for you?
02:08:02*HCl sighs.
02:08:12HClok, that suddenly compiles here now too. but i still don't get your md5...
02:08:20linuxstbCurrent CVS with gcc-3.4.2:
02:08:22linuxstba8cd8bfad8440a785699b54cebf49981 rockbox.iriver
02:08:34HCla9855d6f21193805806e9b9e4a59e8b1 rockbox.iriver
02:08:41HClwith 3.4.3....
02:08:53HClmeh. i'll just go to sleep.
02:12:21rasherI guess my only option is a format
02:12:27rasherthis is sortof annoying
02:12:34rasherstill not booting
02:14:45preglowrasher: how did it happen?
02:14:58rasherI'm not sure unfortunately
02:15:14preglowok, so no one has identical md5
02:15:22rasherI recorded some clips with the stock iriver firmware
02:15:30rasherand tried opening playlists with iriver firmware
02:15:34rasherthat's about it
02:18:55 Quit edx (Read error: 60 (Operation timed out))
02:28:14 Quit Patr3ck ()
02:28:55preglowbut time for sleep
02:29:41 Quit preglow ("for old times.")
02:33:34amiconnHmm. How do I cast a long into a function pointer?
02:35:21HCli do that.
02:35:42amiconnI need that too for my overlay loader.
02:35:51HCl( ( int (*)() ) b->block)();
02:35:51 Quit cYmen (Read error: 113 (No route to host))
02:35:57HCl :P
02:36:27HClthe type is int (*)() :P
02:37:44amiconnThanks. :)
02:37:55amiconnThat results in a funny line, btw
02:37:58HClyup :P
02:38:02amiconnreturn ((enum plugin_status(*)(struct plugin_api*, void*))header[3])(api, parameter);
02:38:13HCl :p
02:38:16 Join ashridah [0] (
02:39:12 Quit Aison ("( :: NoNameScript 3.72 :: )")
02:39:55HCli'm gonna go sleep now.
02:40:06amiconnnight HCl
02:57:51***Saving seen data "./dancer.seen"
03:08:15 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
03:29:56 Join Aison [0] (
03:43:24 Quit Aison ("( :: NoNameScript 3.72 :: )")
04:05:26 Join QT_ [0] (as@area51.users.madwifi)
04:09:00 Join edx [0] (
04:22:31 Quit QT (Read error: 110 (Connection timed out))
04:28:49 Quit Hohoman (Read error: 104 (Connection reset by peer))
04:57:52***Saving seen data "./dancer.seen"
05:13:16 Quit XShocK (" HydraIRC -> <- The future of IRC")
05:13:44ashridahhm. not a good sign. pass the flac decoder in iriver a 60M flac file and it cries because it's too big
05:13:59ashridahdo it on a smaller file, and it hangs while grinding away at the disk
05:54:29 Join DMJC [0] (
06:29:24ashridahwow. linuxstb is right. the vorbis decoder is horribly slow.
06:53:40 Quit DMJC ("Leaving")
06:57:55***Saving seen data "./dancer.seen"
07:31:15HClbah, i need to unwind
07:31:21*HCl caught himself dreaming in sourcecode
07:42:04ashridahheh. was it the software equivalent of a flux capacitor? :)
07:42:31HCli was dreaming about a mixture of c headers and my gf
07:43:56*HCl goes back to sleep
08:18:07 Join methangas [0] (
08:30:17 Join DMJC [0] (
08:57:59***No seen item changed, no save performed.
09:33:56 Join Bagder_ [0] (
09:50:38HClhello... something...
09:50:53 Quit Bagder (Read error: 110 (Connection timed out))
09:52:06 Quit pill (Read error: 110 (Connection timed out))
09:54:04*HCl has a deadly hello.
09:56:34 Join markun [0] (
09:56:49markunGood morning HCl
10:01:03 Join webguest59 [0] (
10:01:06 Quit webguest59 (Client Quit)
10:01:18 Join webguest59 [0] (
10:01:37 Quit webguest59 (Client Quit)
10:05:18markunI read you had bad dreams?
10:06:04markunsource code mixed with gf..
10:06:33*HCl listens to his flatmate acting really stupid with his cat.. mrf
10:06:55HClsome people should seriously learn that some people are still trying to sleep at 10 am sunday
10:07:08HClwell, that wasn't the bad dream
10:07:21HCli kind of meant i was both dreaming about source code and my gf at the same time
10:07:58markunIn my dream my eyelids were glued shut, very strange..
10:08:20HClfortunately, when i have a bad dream i almost always instantly wake up.
10:18:22 Join pillo [0] (
10:21:11 Join XShocK [0] (
10:23:28 Quit XShocK (Client Quit)
10:27:50markunDuring linking of the uisimulator I now get:
10:27:53markunbuilds/iriver-sim/librockbox.a(debug.o)(.text+0x1f): In function `debugf':
10:27:59markunfirmware/debug.c:231: undefined reference to `stderr'
10:28:14markunWhat did I do wrong?
10:28:22HClsims are broken afaik
10:29:23pillohmm... sounds like we have a C guy lurching somewhere... :)
10:29:47markunI thought only the win32 simulator..
10:30:26 Join pill [0] (
10:38:47 Join Patr3ck [0] (
10:41:18 Join XShocK [0] (
10:41:19linuxstbmarkun: the iRiver X11 simulator (using latest CVS) compiles fine for me under Linux.
10:41:31XShocKhmmm... rockbox does not boot.
10:41:50XShocKwhen reset, cookie works and boot the original.
10:42:11XShocKwhen trying to boot rockbox i just hangs with red led on.
10:42:18XShocKand hard disk spinning
10:42:27markunlinuxstb: I use FreeBSD so I changed a few things to get it to compile, but it used to work so I don't know what I did wrong. Will look at cvs diff.
10:43:11amiconnmarkun: Currently, all simulator builds are not working *on cygwin*. However, these builds fail much earlier due to conflicting/ missing prototypes
10:45:05XShocKamiconn: what do you mean by "all simulator builds are not working *on cygwin*" ? it does not work not or never worked, because i had no problems running simulator, didn't test new stuff with audio codecs yet though..
10:45:24XShocK*it does not work NOW or...
10:46:25amiconnmarkun: "Currently,...", that means since Bagder introduced the new build system (on friday)
10:48:08XShocKso, does anybody had that problem with booting rockbox?
10:48:50markunXShocK: I have no problems. Did an cvs update this morning and rebuild the firmware.
10:49:09Patr3cklinuxstb: mine abort with floating point exception when started :-(
10:49:27amiconnHCl: My overlay loader for rockboy on archos is working :)
10:50:32amiconnmarkun: ???
10:50:40XShocKI actually have this problem with an old not up-to-date version of rockbox.
10:50:48markunamiconn: this is what I changed to get it to compile on freebsd so far, can you see if I made any studpid mistakes?
10:50:49linuxstbmarkun: on cygwin, I think the problem is conflicts between the system includes and the rockbox firmware includes (those in firmware/include). It's very likely to be a similar problem with FreeBSD.
10:50:56XShocKand i really did not update anything from cvs
10:52:19linuxstbXShock: could it be that Rockbox is recalculating the fsinfo? That happened to me the other day - you just need to wait a few minutes, and it will then start.
10:53:38XShocKaah, maybe. I just do not understand why would it do that since the only thing that i did yesterday was listening to a couple of songs.
10:54:25linuxstbI had the same problem, it worked fine since I flashed, and then suddenly one day last week, it decided it needed to calculate the fsinfo..
10:54:45amiconnmarkun: I don't spot any obvious mistakes. I only wonder about the extra x11 include...
10:54:48XShocKscandisk did the job, didn't need to recalculate. :)
10:55:22linuxstbI spoke to Linus and Bagder about it on IRC, and they decided my O/S (Linux) must have invalidated the fsinfo during the previous time I used it.
10:55:24XShocKthat actually is more confusing why did it happen
10:55:43amiconnmarkun: Btw, the ssize_t declaration is also a problem on cygwin, but there are more and different ones
10:55:44markunamiconn: it could not find the header files so I added it there. It works for you without it?
10:55:46linuxstbBut they didn't know why that would happen.
10:56:23amiconnmarkun: It used to work. Currently, I can't compile any simulator.
10:58:03***Saving seen data "./dancer.seen"
11:01:14markunamiconn: This is why I added the line: button-x11.c:25:24: X11/keysym.h: No such file or directory
11:02:16DMJCirc on projector is cookl
11:29:57 Join Aison [0] (
11:31:21 Join R3nTiL [0] (~zorroz@
11:32:03 Quit Patr3ck ()
11:40:01 Part amiconn
11:50:48 Join jyp [0] (
11:54:01 Quit markun (Read error: 54 (Connection reset by peer))
12:14:16 Join cYmen [0] (
12:23:52linuxstbIf anyone is interested, I've just finished editing the Wiki with my proposal for the codec part of the new audio playback system:
12:24:03 Join Quelsaruk [0] (~kvirc@
12:29:25 Join markun [0] (
12:30:24*HCl yawns.
12:30:30HCli'm finally awake, i think
12:30:35markunI can now see unicode filenames on my iriver
12:32:42HClnice :)
12:32:47HClwhat about memory usage though?
12:33:48 Quit R3nTiL (Excess Flood)
12:34:16markunI made a very small unicode font. Only iso-8859-* in it.
12:34:40 Join R3nTiL [0] (~zorroz@
12:34:41markunBut I can now display russian, greek, esperanto.. files.
12:35:43markunI was unable to create a japanese file. cp and touch told me: Invalid argument
12:37:33 Quit R3nTiL (Client Quit)
12:40:30 Join muesli_ [0] (
12:41:36 Quit Diway (Read error: 54 (Connection reset by peer))
12:44:02 Join Diway [0] (~diway@
12:49:06DeadManXShocK you got audio working properly on iRiver? Or did I read that wrong that you got mp3 playback?
12:58:05***Saving seen data "./dancer.seen"
13:06:52 Quit XShocK (Read error: 104 (Connection reset by peer))
13:18:05linuxstbJust been browsing the iPodLinux forums and found some very useful profiling information on libmad and libFLAC done by "normalperson": I'll add the link to the wiki somewhere.
13:19:16markunIriver with utf-8 support:
13:21:14 Quit courtc (Read error: 110 (Connection timed out))
13:21:22 Join courtc [0] (
13:22:39 Join amiconn [0] (
13:29:11 Join CoCoLUS [0] (
13:36:28 Join preglow [0] (
13:36:57preglowlinuxstb: that's pretty good information to have
13:37:25linuxstbYep. It saves us some work. I suppose there are advantages to having Linux running on the target...
13:37:48preglowlooks like they did the profiling on an athlon, though
13:39:08linuxstbSorry, you're right. But it's still useful information.
13:42:17preglowi wonder if it's possible to profile it in rockbox, somehow
13:42:34preglowi wonder if there are any good simulators around
13:44:49linuxstbI don't know about that, but if you look in libmad/synth.c, there appears to already be some optimisation flags we can set to enable speed/accuracy trade-offs.
13:45:54preglowyeah, i know of them
13:45:57preglowneed to do some testing on that
13:47:40preglowi think i'll go have somthing to eat
13:47:51preglowand then have a look to see what i can do with libmad
13:48:43linuxstbI'm just trying the OPT_SSO thing now.
13:48:55 Join R3nTiL [0] (~zorroz@
13:49:58 Join muesli- [0] (
13:50:57ashridahpreglow: there's plenty of m68k simulators, the problem will be finding one that supports the emac
13:51:29ashridahand other coldfire features
13:58:24preglowashridah: not really essential for profiling, we really want to find how things perform without emac right nowe
13:59:41 Join hubble [0] (
13:59:58 Join [IDC]Dragon [0] (
14:00:35[IDC]Dragonamiconn, do you read?
14:00:59amiconnI've tried your activity icon, works nicely
14:01:15[IDC]DragonI was also wondering if the activity is inverted for the old bridge
14:01:15amiconnHowever, I ask myself what the icon should resemble?
14:01:30[IDC]Dragonhow do you know, and the level?
14:01:39[IDC]Dragonthe icon is a disk
14:01:39amiconn[IDC]Dragon: It is, I've documented it in the wiki. If you would have checked...
14:01:48[IDC]Dragonbut I don't like it neither
14:01:57preglowso that's what that is
14:02:10amiconn[IDC]Dragon: I got the info from the disassembled archos fw
14:02:11 Join Sucka [0] (
14:02:22*[IDC]Dragon is not subscribed do wiki changes
14:02:40amiconnIt's been there for quite some time..
14:02:55[IDC]Dragonok, beat me
14:03:22amiconnMy rockboy loader is working nicely :)
14:03:32amiconnNow we need a fast gbz80 emu...
14:03:54[IDC]Dragonyes, like 4 cycles/instr.
14:04:11CoCoLUSwhats the dynarec status? :)
14:04:37*[IDC]Dragon found the activity values
14:04:48ashridahCoCoLUS: uh. dynarec is the study of months/years. not days :)
14:04:57ashridah(depending how far you go, that is)
14:05:43CoCoLUSjust asking
14:05:51[IDC]Dragonamiconn: feel free to come up with a better icon
14:06:02HClamiconn: nice work on the loader :)
14:06:03[IDC]Dragonif we know what we should show there...
14:06:35 Quit Nibbler (Read error: 104 (Connection reset by peer))
14:06:59 Join Nibbler [0] (
14:07:02amiconnI would prefer something that looks like an MMC, but without the false association with a paper doc...
14:07:33[IDC]Dragonto the users, internal is no mmc
14:08:07amiconnHCl: The loader actually "wastes" some ram. Plugin ram area is 32KB on archos, of which the loader only takes ~600 bytes
14:08:15[IDC]Dragonthe icon can be as wide as the clock was
14:08:41[IDC]Dragoncan you do something useful with 31 kB?
14:08:47 Quit muesli_ (Read error: 113 (No route to host))
14:08:50amiconn[IDC]Dragon: I even thought of having 2 icons, separate for internal & external
14:09:00[IDC]Dragonit could be provided to the started module
14:09:45[IDC]Dragonthen it's no plain LED emulation any more
14:09:49amiconnI already thought of adding it to the (private) malloc pool, but that may cause problems by the memory being not contiguous
14:09:59[IDC]Dragonthe status bar would have to know
14:11:01linuxstbamiconn: Regarding your logo, how about havin the "corner" on the bottom edge. "paper" icons normally have the corner at the top.
14:11:44linuxstb(sorry, icon, not logo)
14:12:22 Quit R3nTiL ()
14:13:11[IDC]DragonI'd draw it the way the MMC is inserted, corner at the lower left
14:13:25amiconnAlternatively I could put some routines that now are in appls/plugins/rockboy/rockboy.c into appls/plugins/rockboy.c, but that would need a separate callback api, and the main rockboy.c would look much more messy with all those #ifdefs
14:13:27[IDC]Dragonand wider than tall
14:14:07[IDC]Dragonnot enough pixels for the contacts on the left, though
14:18:45[IDC]Dragonbye for now
14:18:49 Quit [IDC]Dragon ()
14:26:44 Join R3nTiL [0] (~zorroz@
14:39:32 Join midk [0] (
14:40:39 Quit muesli- ("ich will Kühe!!!")
14:44:55 Join muesli- [0] (
14:46:55DeadMan4chan is known by a lot of folks I see
14:47:11DMJC4chan is an internet legend
14:47:19DeadManthat it is :)
14:47:31DeadManCock Mongler 4ever!
14:48:30DeadManwhat's this about basic sound playbackon the iRiver? I saw a screencap of Ogg being played?
14:48:57DeadManor is it just showing the tag and not outputting?
14:49:21DMJCyeah, how is sound coming?
14:49:30DeadManprobably just a font test
14:49:35DeadManI dunno
14:49:47markunDeadMan: That's just the listing of a dir to test unicode.
14:49:49DMJCthat's not playing
14:49:59DeadManXshock mentioned something earler about mp3 playback but I did not get what he was saying
14:50:14markunogg's can now be decoded to a file, but only at 0.9% real-time :(
14:50:40 Quit jyp ("poof!")
14:50:42DeadManI want MP3 first anyhow
14:50:45DeadManOgg can wait
14:50:53linuxstbAnd I'm sure XShocK is only working on playback of raw PCM samples - to get the low-level audio working.
14:50:55preglowdamn, the arm has heaps of nice instructions i wish the coldfire had :/
14:51:15DeadManlinuxstb sounds reasonable to assume that
14:51:27linuxstbpreglow: Time for a hardware mod? :-)
14:58:07***Saving seen data "./dancer.seen"
15:00:16 Join DMJC-L [0] (
15:01:04ashridahheh. if only the coldfire had onboard programmable gate arrays :/
15:01:29preglowyou can get customized coldfire cores, but no fpga style shit
15:02:05DMJCboot amigaos on an iriver
15:02:09DMJCthen I'll be impressed heh
15:02:43preglownot all 68k code is valid coldfire code
15:02:51preglowbut the opposite is true, however
15:04:09DMJC-Lheh boot iriver firmware on an amiga..
15:04:12DMJC-Lnow that'd be scary
15:04:49pilloI just managed to get proportional fonts to work nicely in the viewer plugin!
15:05:14pillonow to fix all the bugs i added in the parts that once worked... ;)
15:05:47amiconnpreglow: Iiuc the opposite (all valid coldfire code is valid m68k code) is not true either, thinking of mac/emac... stugg
15:06:00preglowamiconn: ahh, yes, but that's the only exception
15:06:07*pillo digs back into viewer.c
15:06:55amiconnm68k code set differs across family members, e.g. some addressing modes are 68020+
15:10:05 Quit R3nTiL ()
15:11:08amiconnHCl: Did you have a look at the rcvar* stuff? Imho this can be removed, saving space...
15:12:17preglowcoldfire has a subset of addressing modes that are common to all 680x0, afaik
15:13:48amiconnHCl: Btw, I squashed all warnings except the fastmem ones
15:17:03 Quit muesli- ("ich will Kühe!!!")
15:17:17 Join muesli- [0] (
15:18:44 Quit muesli- (Client Quit)
15:18:57 Join muesli- [0] (
15:21:13 Quit Ka (Nick collision from services.)
15:22:06 Join Ka [0] (
15:23:00 Quit Aison (SendQ exceeded)
15:23:27preglowlinuxstb: you should try compiling libmad with OPT_SSO and OPT_SPEED
15:26:36linuxstbpreglow: I did, but only noticed a very small speed increase.
15:27:34preglowlinuxstb: that's really strange...
15:28:12preglowOPT_SSO bloody eliminates a ton of muls
15:28:48linuxstbFeel free to try it yourself in case I did something wrong though.
15:29:14preglowi might, i'm looking into optimizing synth_full right now
15:30:41linuxstbLooking at synth.c - if FPM_DEFAULT is defined (which it is - see the Makefile), then OPT_SSO is enabled anyway. So I think the current speed is already with OPT_SSO enabled :-(
15:31:13preglowhow do i use the mad decoder viewer?
15:31:20preglowlooks like rockbox hooks the mp3 first
15:31:59linuxstbYou need to hold down the joystick, and then use "open with". Or you may want to rename your MP3 files to .mp3.mad, and then modify viewers.config
15:32:15linuxstb^hold down the joystick button
15:32:29preglowyeah, figured as much
15:32:57preglowyou don't count disk acitivity in the timing?
15:34:01linuxstbYes - the "speed" just uses the number of ticks (100 ticks=1 second) since decoding started. You could disable the disk write if you wanted.
15:34:34linuxstb(search for rb->write)
15:35:07lostlogicwhat came of Linus' bad news about the ATA controller and 140mhz?
15:35:19preglowi think he's been busy with life lately
15:35:32lostlogicahh, tragic when that happens.
15:39:29 Part amiconn
15:40:45 Join amx [0] (
15:43:05markunlinuxstb: have you seen this site?
15:43:07 Quit pillo ("CGI:IRC")
15:44:10markunHe got a big speed improvement by replacing the mdct by a fft with a pre and post sorting function.
15:44:43markunThere is a fft in coldfire assambly by freescale.
15:46:18linuxstbI'm just going to concentrate on the higher-level audio/codec code. Are you planning to have a look at optimising Ogg?
15:46:41preglowmarkun: interesting
15:47:10markunDon't know if I will be capable of doing this, maybe.
15:47:21linuxstbIt's a shame the iRiver doens't have an ARM - most codecs seem to already have optimised assembler for them.
15:47:38preglowi don't think there's anything to gain by doing that in libmad, since the mdct blocks are very short, but with vorbis
15:48:35linuxstbDid the oggonachip project use Tremor or libvorbis?
15:48:45 Join Hohoman [0] (
15:49:13markunlow-mem branch of Tremor.
15:49:44preglowpretty logial for a hardware project
15:50:24markunCan we use the fft source of freescale?
15:50:48 Join Aison [0] (
15:51:04linuxstbmarkun: I have no idea. Can you find a license anywhere?
15:51:45markunNo, I haven't found it.
15:51:53preglowit's fixed length
15:51:57preglow1024 samples only
15:52:59preglowcan't find a license
15:58:38 Quit ashridah ("sleep")
16:01:10markunI think I mixed up 2 things in my head. oggonachip was done using libvorbis, but this guy did the same with tremor low mem:
16:01:18*HCl grumbles at rockboy already crashing when it shouldn't.
16:02:24markunI cannot find his thesis and his code patch anywhere though..
16:04:30preglowlink to the thesis is dead
16:06:22markunI downloaded it to my computer a while ago:
16:10:27thegeekthe coldfire is 16-bit?
16:12:14markunThis guy also tried to get vorbis to run on a DSP:
16:17:57preglowthegeek: 32 bit
16:25:29preglowit's so bloody hard to keep track of which format numbers in in mad
16:28:32 Quit hubble (Read error: 110 (Connection timed out))
16:33:01 Join Tang [0] (
16:46:32HCligh, i think i broke rockboy...
16:46:53HClwhere's amiconn when you need him
16:47:30TangHello :)
16:47:40TangSorry can't help
16:47:53HCli don't really expect you to..
16:48:00TangI've read the new "howtogapless" edit
16:48:00muesli-hi tang ;)
16:48:10TangHi Museli
16:48:20Tang@HCl: yes i imagine
16:48:37TangA question: HCl it's a i or a l?
16:49:43TangOkay thanks preglow
16:49:47linuxstbTang: I think you need a font that shows "eyes" and "ells" differently.
16:50:01TangAh yes maybe
16:50:17Tangi didn't chheck if chatzilla offer the possibility
16:50:55TangI've received my new remote yesterday
16:52:00Tangif you see Linus you can transmit him
16:52:21Tanghe was "worrying" about my iHP without remote lol
16:58:11***Saving seen data "./dancer.seen"
17:06:27thegeekkk preglow
17:26:06 Quit einhirn (Read error: 104 (Connection reset by peer))
17:36:15 Join pillo [0] (
17:42:58 Join einhirn [0] (
17:50:11 Quit einhirn (Read error: 104 (Connection reset by peer))
17:50:29 Join einhirn [0] (
17:57:05 Quit einhirn (Read error: 104 (Connection reset by peer))
17:58:59rasherSoooooooooo........... iRiver firmware booted now.
18:00:00rasherBut it didn't yesterday.
18:00:12 Join einhirn [0] (
18:17:37 Quit Tang ("Chatzilla 0.9.66 [Mozilla rv:1.7.5/20041108]")
18:18:23 Join FiZZ [0] (
18:19:10 Quit Hohoman ("( :: NoNameScript 3.81 :: )")
18:23:55 Join Hohoman [0] (
18:24:55 Quit Sucka ("( :: NoNameScript 3.81 :: )")
18:25:01 Part Hohoman
18:47:52 Nick FiZZ is now known as fizze`AwAy (
18:57:57 Join Tang [0] (
18:58:14***Saving seen data "./dancer.seen"
19:03:16 Quit thegeek (Read error: 104 (Connection reset by peer))
19:03:25 Quit fizze`AwAy (Read error: 54 (Connection reset by peer))
19:04:29 Join G [0] (
19:04:46 Join FiZZ [0] (
19:05:01 Quit muesli- (Read error: 104 (Connection reset by peer))
19:06:50 Join hubble [0] (
19:07:36 Join muesli- [0] (
19:12:58FiZZanyone alive here ? ;)
19:14:26FiZZsweet. anyone from the devs around ?
19:14:46preglowwell, kind of
19:14:55preglowif you've got a question, just ask
19:15:06 Nick pillo is now known as pillo_away (
19:15:26FiZZhehe well im curious about the H1x BDM interface
19:16:38preglowdon't know much about it, but some
19:17:16FiZZwell have there been any attempts about a H3xx ?
19:17:41preglowno, too much to be done on the h1x0 yet
19:17:51preglowbut the bdm interface should fit like a glove in the h2x0
19:17:54preglowh3x0, yes
19:18:23FiZZyeah I would assume that too
19:18:41FiZZwell i have access to some BDIs
19:18:56FiZZthet bdm they "built" on the site seems very very geekish lol
19:19:10FiZZi guess a bdi2000 or 4000 would work like a dream though
19:19:59preglowi don't know, but there's not much we _need_ a bdm for now anyway
19:20:20preglowprimary work on h3x0 should be to first map out the differences in the hardware
19:20:27FiZZim guessing the hardware difference in a h1xx and h3xx isnt really big
19:20:35preglowthat would be my guess as well
19:20:40preglowthe obvious difference would be the lcd
19:20:49FiZZyeah, lcd controller is different ;)
19:20:57FiZZbut i guess the uP is different too
19:21:06FiZZmight most likely have other units
19:21:09FiZZexact same one ?
19:21:16preglowthink both are 140mhz 5249 coldfires
19:21:17preglowafaik, yes
19:21:42preglowboth are 5249's, at least
19:21:47FiZZthey are compatible, yeah yeah, but Im guessing that the h3xxs have some more extensions
19:21:56preglowwhat extensions would that be?
19:21:56FiZZi dont think the xvid decoding is a piece of cake
19:22:24preglowwell, no, but i don't think the cpu has any more facilities to ease it
19:22:26FiZZbut maybe thats the 10fps and 400kbit bandwith limit they have ;)
19:22:38preglowit's probably just very tight code
19:22:46FiZZits definetly tight code ;)
19:23:07FiZZim gonna take a look at the reverse engineered firmware some day
19:23:14preglowgood luck
19:23:29FiZZjust curious about their assembly
19:23:36preglowi wonder if there's a 5249 optimizing guide around
19:23:44preglowi sure as hell could need one
19:24:48preglowour codecs are shit slow at the moment
19:24:52preglowand need very extensive optimizing
19:24:54FiZZwhat are you coding atm ?
19:25:41preglowi'm working on optimizing the codecs
19:25:51FiZZoh i c
19:26:13FiZZhappy inverse discrete cosine transformations, eh ? ;)
19:26:34preglowyes, it'll be fun doing those in asm
19:27:08FiZZis there any way to test without an actual h1xx ?
19:27:38preglowi am wondering about the same thing
19:27:54preglowthere are coldfire simulators about, but none that support the 5249 extensions
19:28:02preglowwhich we'll rely pretty heavily on while optimizing
19:28:41FiZZwell i defenitely wanna aid u guys
19:28:54FiZZbut seeing that theres emulator issues and i just own a h3xx im kinda lost
19:30:43CoCoLUSdo we have a 4 in a row plugin? :)
19:30:55rasherCoCoLUS: nope :)
19:31:15preglowFiZZ: you could work on finding an emulator :P
19:31:40rasherI'll be happy to test out code on my iRiver
19:32:15FiZZhehe well
19:32:36FiZZi can try. I can ask some of my profs
19:32:44FiZZi think we had a project that involved a coldfire
19:32:56FiZZbut i dont think theres a GPL'd solution
19:37:14FiZZhm that linked site doesnt show lots of progress :-/
19:37:58preglowno, indeed
19:38:12preglowwith just a bit of luck, it's abandoned
19:38:25FiZZand dates would be much appreciated for wiki "news" entries
19:38:29FiZZ"a bit" hehe
19:38:38FiZZi can almost hear the despair
19:38:59 Join LinusN [0] (
19:39:17FiZZah nice
19:41:43LinusNFiZZ: we didn't build our bdm emulator
19:42:01FiZZi c
19:43:13LinusNit's a simple parallel port wiggler from P&E
19:43:37FiZZhm nice
19:43:51FiZZi thought I read somewhere that some device was involved
19:44:13LinusNyes, the parallel port wiggler from P&E
19:44:42FiZZyou got a point there
19:44:47FiZZhow much are these things, roughly ?
19:45:26LinusN$200 something
19:45:54FiZZmaybe i can trick the uni into buying one of these ;)
19:46:21LinusN$149+shipping to be exact
19:48:09preglowLinusN: didn't you say that thing managed to function also at 140 mhz?
19:49:03LinusNworks for me
19:49:07preglownot bad for something parallel port based
19:49:12FiZZthx for the info LinusN
19:49:24LinusNnah, it doesn't need to communicate in 140MHz
19:49:45LinusNthe comm pace it set by the pc
19:50:10LinusNthe only trick is that it has to handle the 140MHz clock when breaking the target
19:50:29LinusNand that is handled "locally" inside the wiggler
19:50:50LinusNi have to go now
19:51:00LinusNcu around
19:51:03 Part LinusN
19:54:19 Quit markun ()
19:55:54TangHi Linus
19:56:07preglowlinus just left :]
19:56:38TangAh sorry
19:57:02Tangjsut wanted to say hello
20:02:01 Join montana16 [0] (
20:06:23 Nick G is now known as thegeek (
20:06:54 Quit montana16 ("—I-n-v-i-s-i-o-n— 2.0 Build 3515")
20:13:01 Join Stryke` [0] (
20:13:12 Nick pillo_away is now known as pillo (
20:24:06 Join XShocK [0] (
20:24:07 Quit Quelsaruk (Read error: 104 (Connection reset by peer))
20:31:07CoCoLUSa question
20:31:13CoCoLUSis it normal that the simulators can't run plugins?
20:32:41pilloI think I've heard that, but I never tried myself.
20:33:05 Join mrmags [0] (
20:33:26 Quit pillo ("CGI:IRC")
20:33:33 Join pillo [0] (
20:36:47 Quit amx ("Client Exiting")
20:38:11linuxstbCoCoLUS: No, the simulators should run the plugins - at least, the X11 simulator on Linux does.
20:39:53CoCoLUSi'm using the win32 version
20:40:20CoCoLUSand every plugin i start gives an error message "blah.rock is not a valid windows image"
20:40:24CoCoLUSand nothing happens
20:41:14linuxstbSorry, never tried the win32 version.
20:41:56CoCoLUSseems like i have to copy that plugin to the iriver via usb each time i modify it :/
20:43:00CoCoLUSthat error message is somewhat funky... "please check against your installation disk"
20:43:04CoCoLUSi wonder what thats supposed to mean
20:54:26FiZZmüsli lol. geiler nick ;)
20:54:26muesli-i suppose wouldnt help you, isnt it?
20:54:33muesli-sicher dat ;)
20:54:41CoCoLUSuh deutsche :P
20:54:47FiZZ<- ösi :p
20:54:54muesli-hihi :-)
20:55:02CoCoLUS<- ösi :)
20:55:05FiZZstemplt mi do als scheipi ob heast
20:55:07muesli-vadder würde sagen, die schluchtenjodler ;)
20:55:08preglowenglish, please
20:55:25FiZZmeh. party pooper. :( ;)
20:55:45preglowi prefer "destroyer of fun"
20:58:15***Saving seen data "./dancer.seen"
20:58:28 Join muesli|tarn [0] (muesli_tv@
20:58:55muesli|tarninternet doesnt like me too much today
21:00:41HClanything new?
21:00:45preglowHCl: that's a stretch
21:01:02preglowunless you're gmt+12 or something
21:01:17muesli|tarnso...if you hold in a coldfire cpu in your hands..would that help you in any waya?
21:02:20HClif someone sees amiconn, can you ask him to upload his rockboy sources to my ftp at
21:02:50CoCoLUSbroke yours? :)
21:02:55HClyes :(
21:03:11HClso if i have to revert slightly anyways, i might as well incorperate amiconn's changes
21:04:24CoCoLUSis that rockboy.rock on your ftp working?
21:04:28CoCoLUSi want to try it out :)
21:05:05 Quit muesli- (Read error: 113 (No route to host))
21:07:22 Join jyp [0] (
21:08:27HCli don't think it is :X
21:08:33HClno, it most definately isn't...
21:08:37HCland i don't have a working copy at the moment
21:09:10thegeekyou really should setup a local cvs/svn you know;)
21:09:58HCli know, i know.
21:10:03CoCoLUSdidn't you send me a copy of your source tree?
21:10:05HCli'll do that tonight.
21:10:10HClactually., yes.
21:10:20HClhm, wait.
21:10:26HClthat already contains my dynarec changes
21:10:31HClwhich is causing my current crashes.
21:10:43HCli'm lagging.
21:10:48HClstupid wireless.
21:11:05CoCoLUSthat version's also not working?
21:11:18HCli have an old source that should work.
21:11:59HClthe should contain tain rockboy-0.3.tgz, which contains a working copy of rockboy
21:12:04HCl*sslaps lag*
21:27:59 Quit jyp ("poof!")
21:29:03 Quit muesli|tarn ("ich will Kühe!!!")
21:41:47preglowlinuxstb: i just emac-ified the entire synth_full loop and if anything, it actually got slower ://///
21:47:55preglowand is it just me, or is make for codecs a bit broken? it doesn't seem to detect changes to the files
21:51:16elinenbeHCl: what is the current rockboy speed?
21:51:21HClelinenbe: too slow.
21:51:34elinenbeI know that, but is it still 1.5fps?
21:51:39HClpretty much
21:51:54HCldynarec is making the interpreter part slower
21:52:14elinenbeoh... that sucks...
21:52:18HClbut once it works, it should provide a deadly speed increase
21:52:26elinenbewhat speed is the CPU running at on it?
21:52:35preglowsame old, 11mhz
21:52:37elinenbeand what exactly is "deadly"? full speed?
21:52:56HCli'm hoping for more than full speed
21:53:02HClbut we'll see.
21:53:22HClits really a nightmare to debug
21:53:29HClso that'll hold me back mostly
21:55:16elinenbeunderstandably... I can't imagine debugging a dynarec core...
21:55:31elinenbeIs it home-grown, or did you get it from somewhere?
21:56:22pilloquestion: in the viewers.config file, what are those numbers after the .rock file name?
21:57:30preglowpillo: an icon
21:57:35elinenbepillo: if I am not mistake, they are the icon
21:58:03pilloyeah, ok, i can "see" (like matrix ;)
21:58:35pillothanks a lot
21:58:49HClelinenbe: i enjoy making it, so, i'm making it :P
21:59:00HClno one ever made an open source gameboy dynarec anyways
21:59:11pilloso they're _definitely_ not the reason my viewer is not recognized ... :/
22:00:05pilloman... maybe I should reboot sometimes XD
22:00:18pilloforget about that ^^''
22:03:35preglowthat's a healthy habit
22:03:58HClpillo: make sure that in general options -> file view -> show files -> supported is set
22:04:01HClrather than music
22:04:44pilloHCl: I did that like 3 or 4" after installing rockbox for the first time ;)
22:05:01pilloI hated not seeing everything on the iriver firmware...
22:05:39pilloit lacked simply a reboot
22:05:55pillothe new .config was not parsed
22:11:39elinenbeHCl: what is the current status of the dynarec?
22:11:43elinenbeHCl: sounds good!
22:11:51preglowi don't frigging get this
22:11:59elinenbeyou're going to make it so there are gameboy emulators for cellphones, etc!
22:14:26linuxstbpreglow: Just read your recent messages. Yes, I know the codec Makefiles are broken - I was going to ask Bagder about it, but he seemed to have more urgent problems with cygwin.
22:15:24linuxstbAlso, about libmad - I think the OPT_SSO already enabled, so maybe your emac versions are more accurate, but at a similar or slower speed to the inaccurate C versions.
22:17:55 Join jyp [0] (
22:18:03preglowlinuxstb: my emac version assumes OPT_SSo at the moment
22:18:43preglowlinuxstb: it should be exactly equal to the code that's already there
22:18:55preglowlinuxstb: of course, right now it produces white noise, but i'm fixing that :P
22:19:55preglowi don't know how much the coldfire depends on instruction ordering and such
22:19:59preglowi'd really love to know that
22:46:07 Join pappou [0] (
22:48:47preglownot only does it seem to be slower, it produces garbage as well
22:48:53preglowgarbage that just sounds a bit like the mp3
22:48:54 Part pappou
22:49:34HCl :P
22:50:14 Quit DMJC-L ("Leaving")
22:50:15*HCl bites rockboy and lets it rest for today
22:50:32HClfor some reason my transition to seperating registers into their own 32bit vars is failing
22:56:32preglowfailing how?
22:56:37 Quit methangas (" Like's GUI? Then try HydraIRC -> <-")
22:58:18***Saving seen data "./dancer.seen"
23:00:37 Join amiconn [0] (
23:01:51 Quit Aison ("( :: NoNameScript 3.72 :: )")
23:04:59 Quit FiZZ ()
23:06:25HClpreglow: um.
23:06:29HClpreglow: it did odd stuff.
23:06:37HClit counted to about 4
23:06:42HClthen it suddenly skipped to 20
23:06:47HClthen it went from 20 to 121 really fast
23:06:49HCland then it froze
23:07:19 Join Muushuu [0] (
23:08:14MuushuuI'm having trouble upgrading my JBRHD15
23:09:40HCli wouldn't know anything about that
23:10:18Muushuuseems like alot of AFK people here
23:10:23*HCl shrugs
23:10:26HClpatience is a virtue
23:10:42Muushuuthanks anyways
23:11:01HCllol, i suggest you just wait a while
23:11:11HClsomeone's bound to wake up, know about it, and respondd
23:11:18Muushuui'll prolly be back on later
23:11:44 Quit Muushuu (Client Quit)
23:13:57amiconnArgh, he went. I just digged up an old ml thread about Toshiba drives, their problems and a solution....
23:19:00amiconnhi HCl
23:19:00HClamiconn: can you send me your rockboy sourcetree?
23:19:28amiconnI'll make up an archive with the changed files (against current cvs)
23:20:21 Join muesli- [0] (
23:22:49muesli-ur welcome ;)
23:24:31preglowlinuxstb: seems it got a teeny, itty, bitty bit faster
23:26:32linuxstb:-) Are we up to 5% yet?
23:26:54preglowhell no, with luck it now rundt 0.1% faster
23:27:17linuxstbHave you tried MP3s at different bitrates?
23:27:19preglowi really would have thought it would be about a percent
23:27:29preglowno, just a vbr snippet i've made myself
23:29:00amiconnI have a nice mp2/mp3 test set, the same song encoded with *all* possible combinations (mpeg1 layer 2,3; mpeg2 layer 2,3; mpeg2.5 layer 3 with all possible bitrates)
23:30:13amiconnOf course some combinations don't make sense for actual use, cause they sound really crappy
23:33:44preglowi really wonder where are all the cycles are going
23:33:48preglowlibmad is insanely slow
23:35:35thegeekthe list from paris's hacked cellphone
23:35:46thegeeklots of celeb phonenumbers;)
23:39:06amiconnHCl: The Makefiles are not finished 100%. They contain all necessary stuff, but conditional actions (iRiver/ Archos) are not yet there. I work around that by commenting/uncommenting
23:39:13HClokay :)
23:39:23HCli got a local cvs set up now, so its more safe for me to fiddle with it..
23:39:50amiconnAlso, the Makefile in apps/plugins doesn't even check the one under rockboy if it finds an up to date .elf
23:40:11amiconnI don't know why; I could swear it worked once...
23:40:41amiconnWhat I did to it so far:
23:42:04 Join ashridah [0] (
23:43:00amiconn- optimised lcd driver (asm for archos), also skipping unneeded lines completely
23:43:11amiconn- asm optimised updatepatpix() for archos
23:43:37amiconn- a bit of further speedup by not refreshing the whole lcd for the frame counter
23:44:02amiconn- rockboy now yield()s once per frame
23:44:28amiconn- overlay loader for archos
23:45:34amiconn- squashed all warnings except the fastmem ones
23:47:04 Join Patr3ck [0] (
23:49:15linuxstbpreglow: Interesting forum post from the FLAC developer about our performance problems:
23:51:33 Quit Tang ("Chatzilla 0.9.66 [Mozilla rv:1.7.5/20041108]")
23:51:36linuxstbMakes me wonder if there are better C compilers than gcc we could be using.
23:51:50preglowi really don't know, the code gcc makes doesn't seem to be half bad
23:53:36preglowbut it surely makes you wonder
23:54:02linuxstbBut he says that the JetAudio has the same CPU, but can only manage low order LPC. Which is similar to our performance - I think we may already be over 100% with FLAC for "easier" files.
23:54:03rasherso a lot of time is spent in FLAC__lpc_restore_signal ?
23:54:41 Nick Bagder_ is now known as Bagder (
23:54:43linuxstbWe haven't profiled it, but it seems the most "difficult" function.
23:55:07preglowi've already done some naive optimization for it
23:55:14preglowand it really didn't help much
23:55:16linuxstbIt's also a function where there are a lot of existing ASM versions for other processors - so I assumed it would be a good place to start.
23:58:00pilloI've made viewer supporting proportional fonts properly.
23:58:47pillobut I need to clean up the code before submitting the patch
23:58:50 Join amiconn_ [0] (

Previous day | Next day