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 2005-02-17

00:00:41amiconnHCl: I even have one button left on recorder... :)
00:00:49HClnice :x
00:01:23amiconnUp/Don/Left/Right are obvious
00:01:35HClyou'll want a and b next to eachother
00:01:44HClstart somewhere where you can't press it by accident
00:01:54HCland select possibly near a and b as well
00:02:07amiconnI'll assign PAD A = F1, PAD B = F2, START = F3, SELECT = PLAY, QUIT = OFF
00:02:14HCldoes the archos allow multiple keypresses?
00:02:19 Join Digital007 [0] (
00:02:43amiconnHCl: yes and no.
00:02:57Digital007how is sound for iriverbox coming on?
00:03:15amiconnIt allows multiple keypresses for buttons hooked up on different adc ports, but not within one group
00:03:43HClwhat about iriver remote/base simultaneous presses?
00:04:17amiconnThis should work (different adc channels) if I read the button driver correctly
00:04:57HCli haven't really looked into how input works in gnuboy, i just made a driver for its interface
00:05:22amiconnOopsss! Binary size == 145 MB (!)
00:05:30HClwhat'd you do
00:05:32amiconnMost likely a section problem
00:06:31amiconnArgh! rockboy needs a .stack section!
00:06:54HCldon't all programs.. need a stack...?
00:07:26Bagderthere is a .stack already isn't it?
00:07:29amiconnYes. But the rockbox plugins use the stack of the rockbox main thread...
00:07:36Bagderah right
00:09:11amiconnHCl: May I take a look at your
00:09:42*HCl grins. like if he could say no without seeming really, really odd xD
00:12:01amiconnHCl: How many .o files do you have in build-dir/rockboy?
00:12:28HCllet me check.
00:12:43midkhey, guess what.
00:12:50midki recovered my old breakout code today.
00:13:06HCldoes it work?
00:13:08HCl :d
00:13:27 Nick Aison is now known as Aison|blabla (
00:13:44HClhrm, i think that flac checkin broke my build..
00:14:21linuxstbHCl: I forgot one file and then committed it about 30 minutes after the others. It should be OK now though...
00:14:32HCli just checked out again and it seems to work now
00:14:46HCltitania:/home/hcl/rockbox/rockbox-devel/iriver# ls rockboy
00:14:46HClcpu.o exports.o lcd.o main.o rccmds.o rtc.o split.o
00:14:46HClemu.o hw.o lcdc.o mem.o rcvars.o save.o
00:14:46HClevents.o iriver.o loader.o nosound.o rockboy.o sound.o
00:15:34preglowwoot, interpolating sine oscillator, i've _finally_ understood it
00:16:18amiconnHCl: Thanks, the same I get
00:16:30amiconnIf I find the cause why I get a bogus stack section and you dont, binary size will be roughly equal to yours (414328 bytes)
00:17:12amiconnThis is still with unmodified lcd driver though
00:17:31amiconnFirst I'd like to get a reasonable compilation result
00:18:49preglowbut yes, now i might actually know how to optimize the codecs
00:19:24HCl :)
00:22:28Bagdercurrent work
00:23:36Bagderworks for x11 sim on Linux at least
00:30:14preglowi take it there's not a gcc hacker present?
00:30:49 Join ashridah [0] (
00:31:52BagderNix on the mailing list seems like a gcc insider
00:34:00preglowi was going to ask how hard it'd be to add emac support to gcc, but i don't think that's something someone's going to be willing to do for me, heh
00:34:24Bagderprobably not
00:36:30 Quit methangas (" HydraIRC -> <- State of the art IRC")
00:36:57amiconnHCl: I have no idea what that stack section comes from. I'll discard it for now. I'll hack a bit on the lcd code, then I'll see how it goes...
00:37:09hubblepreglow: but how would you access the instruction from C?
00:37:34HClamiconn: okilydokiley
00:37:52preglowhubble: very good question
00:37:55hubblepreglow: or do you want an optimization trick that does it automaticly? =)
00:38:12preglowi was thinking something along the lines of intrinsics
00:38:48hubblepreglow: aha.. but it is implemented in gas? asm() blocks are nice?
00:38:53preglowgcc does a really good job with 3dnow and sse intrinsics, so i think it'd do good with the emac as well
00:39:01preglowhubble: that's what i'm doing now
00:39:15 Quit Strath ("Client closed")
00:39:24hubblepreglow: ofcourse intrinsics are better
00:39:32preglow#define MACL(x, y, acc) asm volatile ("mac.l %0, %1, %%" #acc : : "ad" ((x)), "ad" ((y)));
00:39:37preglowthere's an example
00:40:26hubblepreglow: ok..
00:40:27 Part hubble
00:40:30preglowbut it's a kludge, gcc doesn't even know the acc registers exist, so it tends to ignore everything i do that clobber sthem
00:40:46preglowunless i make everything volatile, which i've heard doesn't do wonders for code generation
00:48:41ashridahweee, lots of warnings when building the libFLAC code
00:48:49ashridahall the same warning, mind you
00:48:49amiconnHCl: Where is scan. defined/ what is it?
00:49:45preglowashridah: gcc 3.4 problem
00:50:52HClamiconn: structure with scanline info
00:51:35amiconnHmm. I don't seem to understand... where does it come from? How is the row determined?
00:52:29XShocKgrrr.... How do I make sure that interrupt is installed?
00:52:43XShocKi mean i wrote void DMA0(void) __attribute__ ((interrupt_handler));
00:52:49XShocKvoid DMA0(void) {...........}
00:53:06XShocKbut I have a feeling that it is not installed correctly.
00:54:00HClamiconn: lcd_refreshline collects data and builds the scanline into scan.buf
00:54:09HClthe HACKING doc contains some info about the format used
00:54:25HClpretty much, the lower 2 bits of each byte in buf is the 4 grayscales
00:54:47Quelsarukbye Bagder
00:54:55XShocKgood bight
00:56:17***Saving seen data "./dancer.seen"
00:58:03ashridahpreglow: right.
00:58:50amiconnHCl: I get some warning while compiling. Is that expected?
00:59:38 Quit ashridah ("out")
01:04:24HClamiconn: it has a few warnings
01:04:29HCli just removed most, not all
01:04:40amiconnlcd.c: In function `pal_dirty':
01:04:40amiconnlcd.c:764: warning: unused variable `i'
01:04:48amiconnfastmem.h:25: warning: `readw' defined but not used
01:04:48amiconnfastmem.h:45: warning: `writew' defined but not used
01:04:48amiconnfastmem.h:74: warning: `readhi' defined but not used
01:04:48DBUGEnqueued KICK amiconn
01:04:48amiconnfastmem.h:79: warning: `writehi' defined but not used
01:04:54HClthats cause of my most recent optimization to get palettes out
01:05:00amiconnmain.c: In function `doevents':
01:05:00amiconnmain.c:42: warning: implicit declaration of function `pad_set'
01:05:05HClsince we only use 2bit grayscale anyways
01:05:09amiconnloader.c: In function `rom_load':
01:05:09amiconnloader.c:170: warning: assignment from incompatible pointer type
01:05:19HCli wasn't sure how to get rid of the fastmem ones
01:05:27HCland the main one is cause of my input optimization
01:05:31 Join AC [0] (
01:05:34HCland the loader one i haven't been bothered to fix
01:06:19amiconnHCl: Archos rockboy launches and doesn't crash. It's counting up in the top left corner...
01:06:49HCltry mario
01:06:55amiconnWhen do I have to expect the first output (with mario)?
01:06:55HClit starts displaying at 14
01:06:58preglowamiconn: counting at what rate?
01:07:26amiconnroughly 4 fps
01:07:40amiconnHowever, I get no display :(
01:07:55HClpaste your vid_update routine?
01:11:36amiconnArgh, found the prob.
01:11:45amiconnI expected it to be called every scanline...
01:11:55HClit gets called every 8 at the moment
01:11:59HClfor the caching stuff.
01:12:33amiconnYes... and the number is always odd... with caused my version to return immediately ... since I want to discard odd lines...
01:13:08 Nick Aison|blabla is now known as Aison (
01:13:54HClumm... also look at the scan structure thing, i modified buf and the definition of BUF in order to make caching possible
01:21:05Quelsarukgood night to all, don't work too hard
01:21:13 Quit Quelsaruk ("KVIrc 'Realia'")
01:27:38 Quit Patr3ck ()
01:29:51 Join ashridah [0] (
01:30:17 Quit mecraw ()
01:33:17 Quit xen` (Read error: 113 (No route to host))
01:34:17HClamiconn: any progress?
01:34:20AClets check if libwavpack compiles
01:34:48amiconnHCl: It runs slower (about 2.5 fps) now. Still no output...
01:36:10HClwhat did you modify?
01:37:09amiconnlcd.[ch] and iriver.c
01:37:31amiconnThe buf needs 16 lines (for now), and scanline_index is calculated % 16
01:37:54HClwouldn't that still only feed you even numbers, and even less?
01:38:25amiconnThe lcd driver the only uses the even lines from scan.buf, and skips every second column (cnt += 2)
01:38:51HClwell, your routine should be much like mine, really, aside from those changes and the buf / modulo increase
01:39:05HClyou can always uncomment the draw and clearpixel routines
01:39:15HClthey're meant as a legacy/backup thing thats guaranteed to work
01:39:25amiconnYeah. I simply copied your routine and did those changes
01:39:46amiconnI added all changes with #if LCD_HEIGHT == 64 ... #else ... #endif
01:40:31amiconnBtw: I think there is still plenty of room for optimisation in the lcd driver.
01:41:09amiconnThe best thing would be to use asm for the bit shuffle, that would make it fly...
01:41:38HClmaybe, i'm hardly familiar with m68k asm
01:41:52amiconn...this kind of asm routine is rather straightforward... on SH1... I already did similar things in the grayscale lib.
01:42:20HCli suggest using the drawpixel routines while getting it to work though
01:42:30linuxstbAC: Any progress with wavpack?
01:42:30HClfirst get it to work right, slowly, then look at optimizations
01:43:09amiconnm68k shouldn't be hard either... and there are highly optimised routines originally developed for faaast c2p on Amiga, which do pretty much what is needed here as well
01:44:11preglowgood news
01:44:21amiconnYou Makefile integration is a bit dirty... I always have to remove rockboy.[o|rock] to make it recompile
01:44:26preglowis the lcd blitter a resource hog as it is?
01:44:41midkfo' sho!
01:45:20AClinuxstb: i will use the wavpack decoder desidned for embbled systems
01:45:22amiconnpreglow: The lcd driver, although already optimised, is still a bitshifting mess of hell
01:45:45preglowmidk: it's spelled "fo' shizzle"
01:46:05linuxstbAC: Sounds perfect.
01:46:25ACThis "tiny" decoderThis "tiny" decoder library works with less than 32k of code and less than 4k of data
01:46:26midkpreglow: only if you're 'g
01:49:09HClamiconn: yea, i know...
01:49:32 Quit cYmen_ ("leaving")
01:49:37CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
01:49:37*HCl grins at amiconn's accurate description of vid_update
01:52:35amiconnI have some nice bit-"weaving" asm routines that will hopefully speed things up...
01:58:38amiconnYeehah! It works!
01:59:16HCl :P
01:59:19HClmake pictures
02:02:14amiconnHmm. I guess it's not really playable. The graphics look rather odd at half the size with no interpolation
02:02:43preglowamiconn: framerate?
02:03:11amiconnAdditionally, there is some graphical garbage at the left & right, because the gameboy screen at 1/2 doesn't cover the whole archos screen width
02:04:15amiconnFramerate varies a bit depending on what happens, but i get ~3.3 fps
02:04:22XShocKaaa... can the interrupt handler function be in plugin?
02:04:23 Quit Digital007 ("CGI:IRC (EOF)")
02:04:27preglowdefinitely faster than on the iriver
02:04:44preglowXShocK: i rather doubt that
02:04:50amiconnI only needs to remap 1/4 of the pixels...
02:05:01XShocKhrr.. i didn't think about that...
02:05:07ACwhy can i not use #include <inttypes.h> in a codec?
02:05:26amiconnHCl: Mario demo mode
02:05:35HClmake pics?
02:06:07preglowi wish i understodd fixed point filtering
02:06:24 Quit midk (Read error: 104 (Connection reset by peer))
02:06:36amiconnBaah, digicam batteries are empty :(
02:07:57HCl :X
02:07:59HCli had that..
02:08:45amiconnHmm. I can't obviously rip the batteries off my recorder...
02:08:58HClrip half of them?
02:09:01HClso both have half power
02:09:18linuxstbAC: You can - it's deifned in firmware/include/
02:09:23preglowso both will fail while taking pictures
02:09:57preglowi have two sets of rechargable batteries, prevents stuff like thise very well
02:11:09AClinuxstb: #include "../../../firmware/include/inttypes.h"
02:11:48amiconnpreglow: I tried more than two sets... These NiMH cells tend to be empty when I want to use them... :(
02:12:12preglowamiconn: yes, my camera batteries are always empty if i haven't used it for a while
02:12:39linuxstbAC: No, the Makefile should include -I$(FIRMDIR)/include - look at apps/codecs/libmad/Makefile for an example - you should be able to just copy that (with some minor changes).
02:13:02linuxstbAC: The Makefile reads a text file called SOURCES (in the same directory) listing the .c files that need to be compiled.
02:15:12ACINCLUDES=-I$(APPSDIR) -I.. -I. -I$(FIRMDIR)/include -I$(FIRMDIR)/export\ -I$(FIRMDIR)/common -I$(FIRMDIR)/drivers
02:15:33AC#include <inttypes.h> wont work
02:16:54ACrockbox/apps/codecs/libwavpack/wavpack.h:12:35: inttypes.h: No such file or directory
02:17:39linuxstbSilly question, but does "inttypes.h" exist in your firmware/include directory?
02:18:30 Quit edx (Read error: 60 (Operation timed out))
02:19:18linuxstbAnother silly question, which directory are you in when you type "make" ?
02:19:43ACrockbox/apps/codecs/libwavpack/wavpack.h:12:35: inttypes.h: No such file or directory
02:20:00AC~/rockbox/build $ make
02:21:03linuxstbI'm not sure what the problem is then - at least some of the other codecs use inttypes.h
02:21:59ACcan you show me which codec?
02:22:21*HCl prods amiconn
02:22:26HClamiconn: try it in the archos sim?
02:23:04 Quit Aison (Read error: 54 (Connection reset by peer))
02:23:30linuxstbAC: FLAC uses it (in libFLAC/include/FLAC/ordinals.h), and liba52 uses it in every .c file
02:23:44amiconnHCl: I'll need to look into the Makefiles, but I guess that has to wait until tomorrow
02:24:07*amiconn is watching the pokemon yellow intro on his recorder
02:24:15HCl :P
02:24:19AClinuxstb: got it now
02:24:27preglowahhh, finally
02:24:38linuxstbAC: What was the problem?
02:24:42preglowi think i'll go to bed, then
02:25:11AClinuxstb: Makefile problem
02:25:57linuxstbAC: It always is.
02:27:00AClinuxstb: compiles with about 8 warnings :)
02:27:38linuxstbCool. Don't bother to fix those yet - we'll add it to CVS with the warnings (i.e. the original unmodified code), and then fix the warnings afterwards.
02:28:09ACok.. i will write now a wv2wav plugin to test everything
02:28:49linuxstbOK. I've cleaned up my decoders a lot, so it should be easier for you to write one in the same style now.
02:29:25preglowwho'd have thought things would have progressed at this rate
02:29:35amiconnHCl: rockboy never yields, so the hd keeps spinning....
02:30:02amiconnAh, demo mode!
02:31:36linuxstbAC: The current decoding speed record is FLAC at about 8% of real-time on the iRiver. I'm hoping that wavpack will take the record.
02:31:46HClwhat do you mean never yields?
02:32:22preglowcall yield()
02:32:38amiconnRockbox is cooperatively multithreaded... every thread needs to yield() from time to time
02:32:47AClinuxstb: we will see :)
02:33:05linuxstbSo my ???2wav plugins should be doing the same thing?
02:33:21preglowamiconn: does rockbox have a cpu% per thread measurement thing?
02:33:31XShocKhardware guys, help me. :) how should I set up an interrupt?
02:33:38XShocKin iriver
02:34:13HCloh o.o
02:34:14preglowlinuxstb: i guess most things should be doing it from time to time
02:34:32XShocKI have a feeling that void DMA0(void) __attribute__ ((interrupt_handler)); void DMA0(void) {....} does not work well
02:34:47XShocKdo i need to add anything else?
02:34:54linuxstbHow often should it be done? All my plugins have decoding loops which more or less decode one frame of audio on each iteration. Should I put a yield() inside that loop?
02:35:11amiconnXShocK: This cannot work from within a plugin
02:35:34XShocKI already put it into the firmware itself
02:36:01amiconnlinuxstb: There is no general rule for a yield rate, but I'd aim for at least once per tick
02:36:28linuxstbYou mean 200 times per second?
02:36:32amiconnOf course, for your test plugins it isn't really important, but certainly more often than once per second
02:36:44amiconnA tick is 10 ms
02:36:45preglowdoes rockbox always pass control to other threads when you call yield() ?
02:37:25amiconnIirc it's a simple round robin scheduler, so yes, unless some threads are sleeping
02:37:52linuxstbAre there any other threads running on the iRiver at the moment though?
02:38:04amiconnYes of course
02:38:08preglowbacklight, ata, scroll
02:38:14XShocKamiconn: I want to set up a interrupt handler. and I need interrupt vector. i guess the index of the interrupt. How should I do that?
02:38:35linuxstbI assume it's rb->yield() ?
02:39:04XShocKI think I am pretty done with DMA sound, I only cannot catch the interrupt itself.
02:39:09amiconnIf you give your routine the correct name, this should be done automatically. The default handler for interrupts have a weak symbol, so it is replaced by yourt routine
02:39:20amiconnlinuxstb: yes
02:39:33XShocKok. then how do i get the irq vector? :)
02:39:36 Quit Hohoman ("( :: NoNameScript 3.81 :: )")
02:40:21amiconnXShocK: You don't need to know yourself. Just name your interrupt routine like the vector name.
02:40:42XShocKBut I need the number itself. :)
02:40:48XShocKthe vector
02:41:41XShocKi certainly need it... :) because interrupt's vector will be put into DIVR register of the DMA
02:42:08XShocKMaybe I am completely wrong, but I suspect it is right
02:42:46amiconnAh ok. Hmm, then I don't know. Obviously coldfire works a bit different...
02:43:08XShocKmmmm, maybe you can guess? :)
02:43:29XShocKi really don't want to wait one more day, I will go crazy. :)
02:45:56preglowisn't it listed in a .h file somewhere?
02:46:04amiconnIn case an archos recorder (v1) owner wants to try rockboy... Get it here: Put rockboy.ajz in the root (or wherever you prefer to keep .ajz's), put rockboy.rock into /.rockbox/viewers and replace your viewers.config with the supplied one. Then put a gameboy rom file somewhere, RoLo romboy.ajz, then "play" the rom file
02:46:30HCland don't forget to take pictures >.o
02:46:34*HCl goes to sleep
02:46:45*amiconn goes to sleep as well
02:47:03amiconnPokemon still cycling... (> 5000 now)
02:47:26HClyou need to press start/A, or it'll keep cycling
02:47:34preglownight indeed
02:48:13amiconnI know that I need to start somehow... I simply left it running to see whether it's stable
02:48:47XShocKpreglow: As I understood it is listed, but not for 5249, i mean it is not given the vector explicitly
02:54:20DeadManHey guyz. How far along is iRiver?
02:56:20***Saving seen data "./dancer.seen"
02:56:27 Quit preglow ("w00p")
03:00:55ACgood nigth... later the day, the wavpack codec should be finished
03:01:19 Quit AC ("CGI:IRC")
03:03:07DeadMannew worm? mydoom.m
03:05:57DeadMannew variant rather
03:07:18 Join mij [0] (
03:07:34 Part mij
03:08:27DeadManhmmm. pc-cillin seems to be playing me for a fool
03:50:49 Join amx [0] (
03:57:56 Join midk [0] (
04:05:47 Join QT_ [0] (as@area51.users.madwifi)
04:09:30 Join edx [0] (
04:10:24 Quit QT (Read error: 60 (Operation timed out))
04:27:29 Quit XShocK (" HydraIRC -> <- IRC for those that like to be different")
04:56:24***Saving seen data "./dancer.seen"
05:17:30 Join wacky_ [0] (
05:17:43wacky_how would I make ogg files recognised as music files (just to create playlists, that I'll load in the original firmware)
05:30:00 Quit amx ("Client Exiting")
05:38:02 Quit skav ()
05:45:41DeadManyo heads up
06:11:24wacky_no vorbis supporters here ?! no one wants to work on tremor ?
06:11:41wacky_Hey I have 14 gigs of vorbis stuff.. I hope free codecs are going to get supported before proprietary ones huh?!
06:12:01wacky_we *are* open-source, aren't we ? thus we encourage open source, don't we ?
06:20:26*ashridah hands wacky_ a compiler :)
06:20:53ashridahit'll get done. there's plenty of other things that need to get taken care of as well, vorbis will probably work its way in.
06:31:38DeadManI want to see audio working first :)
06:32:49 Join The_Jester [0] (
06:33:08 Part The_Jester
06:34:02ashridahdon't hold your breath
06:34:15ashridahthere's a lot of infrastructure to build before rockbox will be able to play music
06:37:18wacky_yeah I know
06:37:45ashridahyou could, of course, help
06:37:56DeadManif I knew how I would :)
06:38:10ashridahthat's what books are for
06:56:26***Saving seen data "./dancer.seen"
08:05:20 Quit Taxi|3 (Read error: 60 (Operation timed out))
08:08:36 Join webguest12 [0] (
08:10:53 Quit DMJC-L (Read error: 60 (Operation timed out))
08:11:41 Join Spookie [0] (~ejp@
08:12:49 Quit webguest12 (Client Quit)
08:19:03 Join LinusN [0] (
08:20:22amiconnHCl: u there?
08:24:18midkgood night all.
08:25:14dwihnomorning all :D
08:26:19 Part Spookie
08:27:13 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
08:34:27amiconnHCl: Anyway, I have 2 changes for rockboy on archos. (1) It uses the full screen width now, i.e. scales 160->112 by a bresenham-like algorithm. (2) A first asm optimised lcd driver. Frame rates vary of course, but Mario @start screen does 6 fps now...
08:36:04dwihnoamiconn: You do 68k assembly TOO?! :)
08:38:22amiconn??? I said *on archos*
08:38:53amiconnHowever, 68k asm shouldn't be hard...
08:38:56dwihnoMust be the weekend ghost that got me all nibards :)
08:39:32amiconnThe asm part is only 10 instructions on SH1
08:39:44dwihnoamiconn: have you implemented the grayscale in the lcd driver for rockboy?
08:39:48dwihno10 instructions?
08:40:03amiconnNo grayscale, would be way too slow
08:40:23dwihnohow many more fps do you think you can get, optimizing stuff?
08:40:57amiconnWell, I need to look at the gnuboy sources more thoroughly. All uncommented stuff :(
08:41:36amiconnI have an idea how to run this with plain rockbox (unmodified plugin ram size)
08:44:54 Join Shulberry [0] (
08:47:21linuxstbHmmm. My H140 is suddenly not booting Rockbox. I've started it, and I just get a completely blank screen. But there is hard disk activity - so I'm hoping it's calculating the fsinfo.
08:47:50linuxstb... Ah, it's now started. So I guess it was calculating fsinfo. Strange.
08:48:07linuxstbBTW, Morning all.
08:48:50linuxstbIs it possible to display a "calculating fsinfo..." message, or is the LCD not initialised at that point?
08:50:46Bagderis it really rockbox that does that?
08:50:48LinusNit's possible, but only the ata driver knows, and it doesn't have access to the lcd
08:51:04LinusNBagder: yes, it's our ata driver
08:51:14Bagderwhat is it doing more specificly?
08:51:30LinusNit is scanning the FAT, counting the free clusters
08:51:41linuxstbIt's during fat_init() - the symptoms from a user's point of view is a completely blank LCD screen.
08:51:52Bagderdoes it always do that on boot?
08:52:04LinusNno, only if the fsinfo is blank
08:52:23LinusNblank, as in uninitialized
08:52:29linuxstbNo, some people had it when first booting after flashing. My H140 never did it, but suddenly wanted to today.
08:53:12LinusNi guess it was invalidated by the fat driver on your pc
08:53:25linuxstbBut maybe it won't be too bad when the CPU is at full speed.
08:53:33LinusNguess not
08:53:35Bagderlinuxstb: windows or linux?
08:53:42linuxstbOnly linux.
08:56:27***Saving seen data "./dancer.seen"
09:00:24 Join einhirn [0] (
09:05:09 Join MO-Pantsu [0] (
09:06:33linuxstbMaybe a solution would be to display a message on the LCD before the disk_mount_all() function is called. Just something like "Rockbox initialising, please wait...". Under normal circumstances it would disappear almost instantly, but in case of fsinfo recalc, at least a message wouild be there, and not a blank screen.
09:08:10 Join markun [0] (
09:08:18markungood morning
09:08:49markunTremor now compiles, but I did not make a ogg2wav yet.
09:09:08Bagderlinuxstb: yes, it sounds like a working approach to me
09:09:15Bagdermarkun: nice
09:09:18linuxstbmarkun: Good morning. That's good news. Were there many changes?
09:09:38linuxstbShould it be called vorbis2wav?
09:10:22markunNo, just some small ones. It gives quite some warnings and I didn't get to compile vorbisfile.c (high level api)
09:10:40linuxstbThat's OK, we won't be using vorbisfile.
09:10:57linuxstbI assume it uses stream IO (FILE* etc), which isn't in Rockbox.
09:10:57markunyes, vorbis2wav would be better
09:11:42markunyes, it does
09:11:50linuxstb"AC" is working on wavpack, so we should have 5 codecs soon. More than enough to design a codec API around.
09:12:37BagderI hope to get the SOURCES-sim build up soon too, making it easier to make a sim version that can do this too
09:12:40markunI could fix some of the warnings, but maybe the first commit should have a few alterations as possible?
09:13:10linuxstbBut wavpack is interesting - it has a hybrid mode where the file is compressed losslessly to two files - a normal lossy ".wv" file, plus a ".wvc" correction file. It can either decode the .wv by itself (lossy playback) or use both files to decode losslessly.
09:13:41 Quit DeadMan (Read error: 60 (Operation timed out))
09:13:44Bagdermarkun: it doesn't really matter, as long as we carefully note which version you started with
09:14:06markunpulled Tremor out of SVN yesterday.
09:14:20Bagderthen you have a revision number
09:14:30linuxstbMy approach is to put the unmodified files into CVS first (without linking them into the Rockbox build system), and then make my changes from there. So every change is tracked in CVS.
09:14:46Bagderyes, I guess that is a good way
09:15:09linuxstbOnce it compiles cleanly, then I link it to the build system in CVS.
09:15:11 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
09:15:19markunlinuxstb: I cannot commit code, would you like to take a look at my code and commit it?
09:15:37Bagdermarkun: I can set you up for commit
09:15:43Bagderto ease this operation
09:16:00Bagdermarkun: /msg me your real name, a username and a preferred password
09:16:39BagderI didn't find your nick in the IrcNicks
09:17:17linuxstbmarkun: As long as you don't commit your modified version of apps/codecs/Makefile, then Tremor won't be built as part of Rockbox. So you can commit in safety.
09:17:28markunI'll put it in IrcNicks then.
09:17:41linuxstbOnce it's compiling cleaning, commit apps.codecs/Makefile.
09:18:00Bagdermarkun: ok, you are in. Just proceed with care and ask if in doubt
09:18:12Lynx_are the codec's speed problems resolved yet? Or does the processor run at full speed?
09:18:15Bagderwelcome aboard
09:18:33linuxstbmarkun: Have you seen the "README.rockbox" files I've put in the other codec directories? We would need something similar for Tremor.
09:18:49markunI'll first do a 'man cvs' first :)
09:18:54linuxstband welcome to the codec club.
09:18:58LinusNLynx_: i am working on the cpu clocking
09:19:08LinusNit isn't trivial
09:19:17linuxstbThe Rockbox website has a useful "brief intro to CVS" page.
09:19:27LinusNlots of datasheet digging to find the timings for the different peripherals
09:19:44Lynx_LinusN: ok :)
09:21:18linuxstbmarkun: One last thing, I've heard that theres a "lowmem" branch of Tremor. Is this the one that you are using? I have no idea if that would be better than the normal Tremor for our purposes.
09:21:25Bagderthe daily tarball grew 500K over night ;-)
09:21:48Bagderall the codecs I believe
09:21:56Bagderthe FILES weren't fixed until yday
09:22:08markunNo, I don't think we should use the lowmem brach. It trades in speed for memory.
09:22:26linuxstbOK. speed is more important.
09:24:27 Quit ElGringo ("CGI:IRC (EOF)")
09:25:46 Join bobTHC [0] (
09:26:04bobTHChi all!
09:27:26Bagderusing 2.95.3 for the win32 cross-compile sure is painful
09:27:38Bagderin case you wanted to know ;-)
09:27:59markunTremor uses alloca in some places, should we change that in malloc + free?
09:28:21 Nick QT_ is now known as QT (as@area51.users.madwifi)
09:29:15linuxstbmarkun: Rockbox doesn't have malloc(). At the moment, I'm using a very simple local malloc() implementation in plugins/lib/xxx2wav.c - there is no free().
09:30:19linuxstbIt's possible that when we implement the real codec system, we would have a proper malloc()/free() implementation using a codec malloc buffer. But ideally we would want to try and avoid that.
09:30:52linuxstbCould you use a static buffer instead?
09:31:58linuxstbIn FLAC, I'm ignoring the VORBIS_COMMENTS block at the moment - I'm thinking that we should write our own parser that can be used by both the FLAC and Vorbis codecs.
09:32:35linuxstbThe FLAC one does lots of inefficient mallocs and frees.
09:33:30markunYes, I think so too. Should we also combine some other code from the all the codecs like the mdct code?
09:34:27LinusNi think we should change the codecs as little as possible
09:35:30linuxstbBut in the long term, a highly optimised "libcodecs" could be useful - with functions such as mdct.
09:37:06linuxstbBut I agree 100% that we should leave the codecs as untouched as possible
09:37:11linuxstb.. at the moment.
09:40:19markunIf I use 'cvs add Tremor', will all the files in it also be added to cvs?
09:42:18linuxstbNo, just the directory.
09:43:03markuncvs [add aborted]: there is a version in Tremor already
09:43:04linuxstbThere possibly is a command to do it all at once, but I normally just traverse the directories myself, adding and committing things.
09:43:24linuxstbAh, are there already CVS directories in there from SVN?
09:45:02linuxstbI mean ".svn" directories - maybe they need to be deleted.
09:45:23markunI think the adding worked now. Can I commit the files recursively?
09:46:07linuxstbYou can check the current contents of CVS here:
09:46:22linuxstbI don't know how to commit recursively - I never do it that way.
09:50:27LinusNyou just "cvs ci" without arguments in the root dir of the dir you want to commit recursively
09:50:36LinusNcd tremor
09:50:48LinusNcvs ci -m "blabla"
09:51:43 Join DrRick [0] (
09:52:56markunHm, don't know if it worked. I got:
09:52:58markuncvs commit: Examining .
09:53:04markuncvs commit: Examining debian
09:53:09markuncvs commit: Examining doc
09:53:27linuxstbI think you need to do a "cvs add" before the commit
09:53:58markunI thought I did that. Must have done something wrong then.
09:54:20linuxstbmarkun: Also, I think you can delete directories like "debian" (and any automake/conf tools) - they just get in the way. But that's just my opinion.
09:55:08 Quit Shulberry ()
09:55:27linuxstbmarkun: Try typing "cvs status" in your Tremor directory.
09:55:49linuxstbThat shows you the current status of the files - "locally added" would mean it's added but not committed.
09:57:16markuncvs status: failed to create lock directory for `/cvsroot/rockbox/apps/codecs/Tremor' (/cvsroot/rockbox/apps/codecs/Tremor/#cvs.lock): No such file or directory
09:57:53markunCan I just login with cvs when I did the cvs co as anonymous?
09:58:20linuxstbAh, no - I think you need to do a brand new checkout after logging in as yourself.
09:59:19markunShould I also delete the .svn dirs?
09:59:39linuxstbmarkun: Yes, I think so.
10:03:12markunlinuxstb: what about the doc dir?
10:06:23linuxstbmarkun: I think delete it.
10:06:55linuxstbBut keep README, COPYING etc in the main directory.
10:07:22markunYes, I will.
10:09:07markunI think it worked! :)
10:09:54linuxstbYes - I can now see it at
10:10:31linuxstbI've just looked at your README.rockbox, and I think you've forgotten to change some of the information when you copied my file from libFLAC
10:11:32 Join DMJC [0] (
10:12:14markunok, I will add the things I removed from the source tree.
10:13:46linuxstbAlso, I think you need to set the OUTPUT variable in the Makefile to $(OBJDIR)/libTremor.a (so -lTremor will work when linking), but I'm not 100% about that.
10:14:22DMJCUM... is cvs broken?
10:15:13linuxstbDMJC: The daily builds table looks fine.
10:15:44DMJCI just updated
10:15:48DMJCit won't build now
10:15:58linuxstbWhich target?
10:16:07DMJCCC mpa2wav.c
10:16:07DMJCmpa2wav.c:25:31: codecs/libmad/mad.h: No such file or directory
10:16:29linuxstbAh, you may have to do "cvs update -d" in the apps/ directory to get the codecs
10:16:31markunlinuxstb: I was not sure if lib should be in front of it. I can add it.
10:17:20linuxstbmarkun: I think you will need to - in any case, it does no harm.
10:17:55DMJC/home/james/development/iriver/rockbox/firm/mpa2wav.o: In function `plugin_start':
10:17:55DMJC/home/james/development/iriver/rockbox/apps/plugins/mpa2wav.c:156: undefined reference to `mad_stream_init'
10:18:33linuxstbDMJC: Do you have the codecs directory? Is there a "libmad.a" in your build directory?
10:19:09 Join Patr3ck [0] (
10:19:33DMJCno libmad.a
10:19:37DMJCbut there is a codecs dir
10:25:13linuxstbHave you re-run tools/configure? There was a new variable added to enable compilation of the codec libraries.
10:25:26DMJCah k
10:28:48 Nick ashridah is now known as Lost-tv (
10:30:13markunlinuxstb: I have to go now. When should I commit the codecs/Makefile?
10:31:57linuxstbmarkun: You can leave that until you are happy with how Tremor compiles - it is easy for someone else to add it in their local copy.
10:32:10markunAnd what is apps/FILES used for?
10:32:24linuxstbTo generate the .tar.gz file - so you will need to add Tremor into it.
10:36:49 Quit DrRick ()
10:44:41 Quit markun ()
10:49:36DMJCso now that these soundcodecs compile
10:49:52DMJCany idea how well they'll run when the iriver runs at 140mhz?
10:51:03linuxstbNo. The Iriver is currently at 11MHz, so we should see about a 12 -13 times speed-up at 140MHz. I don't know if there are any other factors involved.
10:51:50linuxstbBut that's one reason I wanted to get them in CVS as soon as possible - so others can look at optimising them.
10:54:23 Join Aison [0] (
10:56:29***Saving seen data "./dancer.seen"
10:57:10dwihnois there a way you could tell gcc to link only used functions? like borland used to do in borland c/c++ ver. 3 :)
10:57:50linuxstbI thought that is what happened. Are you saying it doesn't?
10:58:36dwihnoI don't think so
10:58:54dwihnonobody but I would be happier if my guesses are wrong
10:59:05LinusNno you can't
10:59:27LinusNnot on a function level
10:59:47LinusNyou can do it on the object file level, by using a library
11:01:02dwihnoThat would be pretty swell if the compiler could skip unused stuff
11:04:51linuxstbLinusN: If you create a library, and then statically link against that library, are only the used functions from that library copied into the output file?
11:05:17LinusNlook at the bootloader, for example
11:05:29linuxstbThat's OK then. I was worried I would have to remove unused functions from the codec libraries.
11:05:53LinusNbut remember that if there are several functions in an object file, all of the functions will be included
11:06:05LinusNeven if only one of them is ised
11:06:28LinusNnote: a library is a collection of object files, not a collection of functions
11:07:33linuxstbAh, so unused functions in the codecs probably are still being copied? Unless the whole .o file is unused.
11:08:19LinusNthis is why the plugin grayscale library contains so many .o files
11:08:33linuxstbSo there is no advantage to libraries then (in that respect). I could just link against the individual .o files that I know are needed.
11:09:57linuxstbThanks for the explanation - that's very useful to know.
11:13:01 Join B4gder [0] (
11:25:31dwihnoLinusN: does the compiler do it automatically, or do you have to utilize additional flags?
11:25:59LinusNthat's the way the linker does it
11:26:12LinusNno extra flags needed
11:26:21dwihnoneato! :)
11:26:28dwihnotoo bad it's only on object level.
11:26:33LinusNprovided that the .o files are in a library
11:26:43dwihnoit would be swell if it would be on a function level
11:26:58dwihnohow do you create a library using gcc? using ar?
11:27:13*LinusN goes to lunch
11:28:25*dwihno went packing printouts
11:31:52 Quit Cassandra_ (Read error: 110 (Connection timed out))
11:32:06 Join Cassandra_ [0] (~christi@
11:39:04 Nick Lost-tv is now known as ashridah (
11:40:48 Quit DMJC ("Leaving")
11:42:38*B4gder pats his compiler... now be nice
11:49:21 Join preglow [0] (
11:55:57preglowis dither really necessary for 16 bit samples?
12:05:18 Join cYmen [0] (
12:07:17linuxstbpreglow: Are you talking about my libmad implementation?
12:07:22preglowlinuxstb: yes
12:07:39preglowthat's what i was looking at
12:08:14linuxstbI've taken that code from a program I wrote a couple of years ago. I remember at the time that that dither routine noticably improved the sound quality compared to just truncating the samples to 16-bits.
12:08:21preglowi'll have to do a comparison sometime
12:08:31linuxstbPlease do.
12:08:41preglow'cause i somehow doubt i can hear it
12:08:49preglowmy hearing isn't the best in the world, though
12:09:37linuxstbMy hearing is not the best either when it comes to comparing different versions of the same muisc, but I do remember noticing a big improvement with that dither routine. Buit my memory isn't the best either...
12:09:58preglowi'll check it out sooner or later
12:10:12preglowh1x0 permits higher resolution output than 16, yes?
12:10:17preglowi think someone mentioned 20 bits
12:10:20 Join ripnetuk [0] (
12:11:10preglowbut yes, i know completely understand all emac dealings
12:11:23linuxstbYes, I think the audio hardware supports 24-bits, but there is another bottleneck (i2s?) that limits it to 20 bits.
12:11:24preglowso look forward to the time i've got time to start optimizing
12:12:00linuxstbBut I'm not sure if the complication of 20-bit samples would be worthwhile. I guess we need listening tests to find out.
12:12:04preglowwhich most certainly won't be for a little while :/
12:12:13preglowwhy is that a complication?
12:13:23linuxstbBecause one sample is then 2.5 bytes. It would also increase the memory usage of the PCM buffer.
12:13:57 Join Quelsaruk [0] (~kvirc@
12:14:21preglowwell, we really should be flexible with the pcm buffer format anyway, the h1x0 won't be the last player rockbox is ported to
12:14:37preglowand it would have to packed in 3 or preferably 4 bytes, yes
12:15:05preglow4 probably will be faster than 3, the coldfire being a 32 bit processor
12:15:27preglowbut yes, i understand the concern
12:15:31preglowlistening tests will have to be done
12:15:40linuxstbWe need to think about the format the low-level audio driver needs, we don't want to convert between two many different PCM formats.
12:15:48linuxstb^too many
12:16:11preglowpersonally i don't think i stand a chance to hear the difference
12:16:16preglowbut then again, my hearing is a bit damaged
12:16:44preglowhear the difference between 16 and 20 bit
12:17:35linuxstbI would be surprised if I could tell the difference as well. But who knows? Audio is not a science.
12:19:41preglowbut i need to go
12:19:55 Quit preglow ("woop")
12:25:31 Join jyp [0] (
12:33:14 Join DMJC [0] (
12:53:36 Quit ripnetuk ("Leaving")
12:53:48 Join ripnetuk [0] (
12:56:33***Saving seen data "./dancer.seen"
13:08:18 Join Zagor [242] (
13:10:12 Quit B4gder ("ChatZilla 0.9.35 [Mozilla rv:1.5/20031007]")
13:10:27 Quit Nibbler (Remote closed the connection)
13:10:41 Join Nibbler [0] (
13:10:41 Join NibbIer [0] (
13:10:45 Join xNibbler [0] (
13:10:58 Quit Nibbler (Client Quit)
13:11:45 Quit NibbIer (Client Quit)
13:15:07 Join Patr3ck_ [0] (
13:17:51 Quit xNibbler (Remote closed the connection)
13:18:24 Join Nibbler [0] (
13:18:24 Join NibbIer [0] (
13:18:30 Join xNibbler [0] (
13:18:43 Quit NibbIer (Client Quit)
13:18:45 Quit Nibbler (Client Quit)
13:19:58 Nick xNibbler is now known as Nibbler (
13:30:19 Quit Patr3ck (Read error: 110 (Connection timed out))
13:35:35 Join lolboy [0] (
13:38:26 Quit bobTHC (Read error: 60 (Operation timed out))
13:40:53 Quit lolboy ("CGI:IRC")
13:42:35 Join [IDC]Dragon [0] (
13:43:16*[IDC]Dragon is swamped in cvs emails
13:43:37[IDC]DragonI should unsubscibe
13:44:43 Quit Nibbler (Remote closed the connection)
13:45:10[IDC]Dragonto even track Rockbox is too much for me these days, let alone contributing
13:46:58Quelsaruki had to unsuscribe once from normal mail list
13:47:27Quelsaruknot being able to read 30 mails is horrible, after a week you have too many
13:55:34 Join NibbIer [0] (
13:55:41 Quit NibbIer (Read error: 104 (Connection reset by peer))
14:14:17 Join elinenbe [0] (~elinenbe_@
14:25:40 Part LinusN
14:28:57 Join lostlogic [0] (
14:30:19 Join LinusN [0] (
14:31:10 Join midk_ [0] (
14:38:20 Quit lostlogic ("Going to the moon")
14:38:21 Join R3nTiL_ [0] (~zorroz@
14:44:07 Quit midk (Read error: 110 (Connection timed out))
14:48:43 Join bobTHC [0] (
14:50:01 Quit Patr3ck_ ("User pushed the X - because it's Xtra, baby")
14:53:47 Join Nibbler [0] (
14:56:36***Saving seen data "./dancer.seen"
15:04:37MO-PantsuNo movement on those 6 red items on the iRiver webpage for a few days now. Hmmm. Bummer.
15:06:18 Join midk [0] (
15:06:51ashridahlots of infrastructure to build first
15:07:09ashridahgotta agree on design and apis too, or there'll be holes and duplicated incompatible work
15:07:33ashridahkeep in mind that rockbox hasn't had to do its own decoding and buffering to anything like this extent before
15:14:37rasherMO-Pantsu: things are not standing still at all
15:14:38MO-PantsuPushing the chips ;)
15:15:00MO-PantsuI know. I just like to see the page change every now and then or I get paranoid ;)
15:15:14MO-Pantsuwrong word.
15:15:24MO-Pantsutypo hell :)
15:20:31 Join lolo-laptop [0] (~lostlogic@
15:22:40 Quit midk_ (Read error: 110 (Connection timed out))
15:30:31ashridahMO-Pantsu: if you want to see action, sign up to the cvs log mailing list
15:30:45ashridahthen you'll have mountains of incomprehensible junk fly past that'll be nicely reassuring
15:31:06 Join Baggder [0] (
15:31:18ashridahit gets noisy tho
15:31:35 Quit ripnetuk (Read error: 54 (Connection reset by peer))
15:32:27 Quit ashridah ("sleep")
15:32:51Baggderany cygwin user around who feels like testing some rather extensive build changes for me?
15:33:15rashersorry, not a cygwin user
15:33:25rasherand religiously not a cygwin user as well
15:34:05BaggderI've tried x11 on Linux and win32 cross-compile on Linux and they work now
15:37:22Baggderanyway, for the log, the patch to try is here
15:38:57[IDC]Dragonwhat's new?
15:39:22Baggderit uses SOURCES
15:39:27Baggderfor the sim too
15:39:41BaggderI've tried to generally unify sim and target builds
15:39:55[IDC]Dragongood, yes
15:40:10*[IDC]Dragon wonders how we dis without
15:40:26Baggderlots of kludges
15:40:48[IDC]Dragonbut I can't test atm
15:41:31Baggderno worries, there's no hurry
15:41:51BaggderI've also tried to move more logic into configure
15:41:57Baggderto clean up the Makefiles
15:42:04Baggdermakes things easier to follow
15:42:51Baggderand of course I must've broken things I've just not discovered yet
15:44:30 Quit R3nTiL_ ()
15:44:38 Join methangas [0] (
15:45:54 Part LinusN
15:48:40 Quit linuxstb ("Leaving")
16:04:36 Quit elinenbe (" HydraIRC -> <- Get hot chicks here!")
16:06:26 Join Patr3ck [0] (
16:16:04amiconnHCl: u there?
16:17:19 Quit Baggder ("CGI:IRC (EOF)")
16:22:35 Join Hohoman [0] (
16:32:55Quelsarukamiconn: how's your rockboy port for archos ? :D
16:33:09amiconnDidn't you read the logs?
16:33:30amiconnI have it running, using the full lcd width
16:33:59amiconnFrame rate varies of course, but Mario title screen does a little above 6 fps
16:33:59Quelsarukwell, seems i didn't read the logs
16:34:20amiconnI already have ideas how to push it further
16:34:45amiconn...and how to make it work without modified rockbox build
16:34:46Quelsarukcould you send me the firmware to test it? thx!
16:35:36amiconnQuelsaruk: [02:46:22] <amiconn> Oops,
16:36:03amiconnPut rockboy.ajz in the root (or wherever you prefer to keep .ajz's), put rockboy.rock into /.rockbox/viewers and replace your viewers.config with the supplied one. Then put a gameboy rom file somewhere, RoLo romboy.ajz, then "play" the rom file
16:36:33amiconnI hope to break 10 fps this evening
16:37:26HCl :P
16:37:59Quelsaruki'll test right now
16:39:56HClamiconn: i'm interested in your scaling algorithm for 144 -> 128 on iriver
16:40:24HCland ofcourse i want to see pictures
16:40:27HCland the code in general :p
16:41:29amiconnThis algorithm does only work well for horizontal scaling
16:41:43HClwell, i was thinking of blending two scanlines into one
16:41:54amiconnBlending is rather costly.
16:41:56HClevery 9 scanlines
16:42:02HClwhy? its just an or
16:42:25amiconnIt's not exactly an or, because this would prefer black pixels
16:42:33HClgotta go
16:42:44amiconnOn the contrary, one coud blend with and
16:43:01amiconnDid oyu change some code meanwhile?
16:44:16Quelsarukanyone can give me a gamboy rom?
16:45:26 Join mecraw [0] (
16:46:39amiconnQuelsaruk: Look here: There's Mario and Pokemon yellow
16:52:26Quelsaruklooks nice
16:54:00Quelsaruka bit slo, but nice
16:54:20Quelsarukwho said a mp3 player is only for listening music or watching rvf? :D
16:55:17 Quit jyp (Remote closed the connection)
16:56:38***Saving seen data "./dancer.seen"
16:57:00Quelsarukwell... i have to say that on the iriver looks nicer
16:59:14 Join preglow [0] (
17:00:25*[IDC]Dragon spots Jens
17:01:16[IDC]Dragonand is exited about gameboy emulation on JBR
17:01:57Quelsarukexited or excited? :D
17:02:06amiconnI hope you mean *excited* ;-)
17:02:58amiconnI have an idea how I can make it run without such a huge plugin ram area (currently the modified version uses 512 KB)
17:03:27[IDC]DragonPIC in the mp3 buffer?
17:03:46amiconnI'll write a small 'loader plugin' which will then load the main binary into the mp3 buffer (towards the end, where the address will be constant)
17:04:07[IDC]Dragonah, no PIC, better
17:04:19amiconnSomething like an overlay in the old days of DOS...
17:04:46[IDC]Dragonthe code (without tables) has no chance for 32 kB?
17:05:27amiconnYes, it will be linked to a fixed address. I'll use ram_end - 512 KB, this leaves 480 KB (without the original plugin ram)
17:05:42*[IDC]Dragon has 8 MB
17:06:08amiconnThen you'll need to compile yourself. This should enable loading of all gb roms.
17:06:24amiconn2 MB boxes cannot load all roms.
17:06:27[IDC]Dragonmario is only 64 KB?
17:06:39amiconnPokemon yellow is 1 MB
17:06:47[IDC]Dragonquite a range
17:06:52amiconnThe maximum is 4 MB iirc
17:07:24amiconnThe CPU is a Z80 lookalike. I wonder how the >64 KB roms are addressed
17:07:44[IDC]Dragonmust be bankswitched or so
17:07:58amiconnMany banks there are...
17:09:02[IDC]Dragondid you look at the CPU interpreter, about how muck "potential improvement" there is?
17:09:18amiconnNot at all yet.
17:09:59[IDC]Dragononce there's been a famous console emulator, I forgot for which
17:10:11amiconnI only looked at the lcd driver, and there was indeed some potential. The C version (without proper scaling 160->112 but only halving to 80) did ~3.3 fps.
17:10:11[IDC]Dragonbut Zelda was the killer app
17:10:21amiconnThe asm version with scaling does 6 fps..
17:10:35[IDC]Dragonyes, saw that, nice
17:11:08[IDC]Dragonthe above re-used code which was already interpreted, iirc
17:11:09amiconnI hope to speed it up further by skipping the calculation of the dropped scanlines completely
17:11:58amiconnNo, but the C version uses individual bit shifts by 1..7, then ORs together
17:12:53amiconnThe asm version tests the source bits, the assembles the output byte by rotating in the carry bit
17:14:00 Join Stryke` [0] (
17:14:02amiconnI hope to get this to work as well for 68K, because it allows to drop another loop across the scanline. (That's part of the fps gain with asm)
17:14:16amiconnThe asm is only the core loop, 10 insns on SH1
17:16:50amiconnOf course I have to keep a (changed) C version to make it work on the sim (or use x86 asm - urgs!)
17:19:08[IDC]Dragonhow many fps do we need?
17:19:33amiconnHCl: Did you change some code since yesterday?
17:19:34HClamiconn: does it work in the sim?
17:19:37HClamiconn: nope
17:19:43HCli don't think i have
17:19:49HCldoes your version do savegames?
17:19:53HClwhen you load pokemon?
17:19:57amiconnNo, not yet, but I'm going to look into it
17:20:02HClit should..
17:20:08amiconnsavegames - seems so
17:20:18amiconnThe 'no' was for the sim question
17:20:23HClthen no, haven't changed anything
17:20:52amiconnAt first I'll try to improve the make mechanism
17:21:12amiconnThis is currently ugly
17:21:39amiconnFor the sim, I'll wait for Bagder's new build system
17:22:16[IDC]Dragondid you see him write above
17:22:25[IDC]Dragonhe issued a patch
17:22:35amiconnI'll test as soon as I'm at home
17:22:57*[IDC]Dragon peeks into gnuboy's sources
17:23:15[IDC]Dragonthere already is asm in there!
17:23:26amiconnYeah - x86 asm...
17:23:34[IDC]Dragonfrom when is this code, that they need this?
17:24:01[IDC]Dragonit has the benefit that you know what to look for
17:24:39 Quit Cassandra_ (Read error: 110 (Connection timed out))
17:25:42 Join Cassandra_ [0] (~christi@
17:31:52 Quit bobTHC ("( :: NoNameScript 3.81 :: )")
17:31:59 Join wacky__ [0] (
17:32:17wacky__anyone tested my plugin and willing to put it in the CVS ?!
17:32:45 Quit wacky_ ("BitchX-1.1-final -- just do it.")
17:32:51 Nick wacky__ is now known as wacky_ (
17:55:28rasherlooks like libmad's pretty fast
17:55:48rashermaybe I'm reading this wrong
17:55:56rasheryes I am
17:55:59rashernevermind me
17:56:52*[IDC]Dragon checks out Mario
17:59:17 Quit Patr3ck ("User pushed the X - because it's Xtra, baby")
18:02:42 Join markun [0] (
18:05:29Quelsaruk[IDC]Dragon: do you like it??
18:09:27[IDC]Dragonto be honest, no
18:09:35[IDC]Dragonbut it's cool
18:12:36Quelsarukhave you seen HCl screenshots in iriver?
18:12:42Quelsarukthat really looks cool
18:13:15 Quit wacky_ ("ahuh")
18:16:06 Join CoCoLUS [0] (
18:18:54 Part CoCoLUS
18:20:10 Join CoCoLUS [0] (
18:20:35Quelsarukthose are really interesting
18:33:09 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
18:49:37Quelsarukaren't yours those screenshots of the rockboy plugin in the iriver?
18:50:04HClonly the last two, since my cam didn't have batteries
18:50:14HCli own an h140
18:50:47preglowmost of the photos are mine, i think
18:50:49Quelsarukthe supermario one
18:50:53preglowunless there are some new ones around
18:50:57Quelsarukdoesn't matter
18:51:07preglowthere's more there
18:51:14HClyea, its linked on the wiki
18:51:26Quelsarukthose in titania.something ftp
18:51:33CoCoLUShow fast is the gb emu?
18:52:01Quelsarukamiconn said that in archos recorder is 6fps
18:52:03HCltoo slow, but we're planning on getting it to full screen
18:52:04Quelsarukright now
18:52:14HClat least, on iriver.
18:52:37preglowi think we'll manage that
18:52:37CoCoLUSfull screen or full speed :)
18:52:45preglowwith clever asm and higher clocked cpu
18:52:54HClim tired
18:52:55HClsue me
18:54:05CoCoLUSnice to hear
18:54:26CoCoLUSthough the thought of a firmware playing gameboy games but no mp3s seems somewhat funny :)
18:54:51 Join _aLF [0] (
18:56:42***Saving seen data "./dancer.seen"
18:57:22CoCoLUShow much work was it to port gnuboy?
18:57:31HCli got it running fairly quickly.
18:57:54 Join Chamois [0] (
18:58:32CoCoLUSso porting standard c++ apps to m68k/rockbox is fairly "easy" ?
18:58:37preglownot c++
18:58:45CoCoLUSc only? :)
18:58:55preglowof course not, but gnuboy is c
18:59:04preglowand we'll try to stick with c
18:59:13HClit completely depends on the application
18:59:15CoCoLUSc# would be nice :P
18:59:17HClgnuboy was written for portability
18:59:20preglowbut yes, the difficulty varies
18:59:26preglowCoCoLUS: haha, forget it
18:59:39preglowc# is rather nice
19:00:05preglowbut i mostly use c++ these days
19:00:09preglowi don't like it very much, though
19:00:15CoCoLUSwell bringing the whole .net fx to the iriver would be quite impossible, i think
19:00:31preglownah, not impossible
19:00:34preglowbut very impractical
19:00:40CoCoLUSnothings impossible :)
19:00:41HClthere isn't a c# for linux, is there?
19:00:42preglowand it would perform badly
19:00:46preglowHCl: there is
19:00:47CoCoLUShci, there is
19:00:47preglowHCl: mono
19:00:59HCli'll still stick with java / c / c++
19:01:07preglowand i'll stick with c/c++/perl
19:01:08CoCoLUSc# == java :)
19:01:18HClc# was made by microsoft.
19:01:22HCli trust sun more than microsoft.
19:01:25rasherthey're very alike
19:01:32CoCoLUSi trust neither
19:01:40rasherI don't put Sun much higher than MS
19:01:50preglowbut yes, c# is an ok language, beats c++ by far
19:02:00preglowand once the mono guys are done, you're not dependent on microsoft tools
19:02:11CoCoLUSif they're ever done...
19:02:17preglowthey've gotten pretty far
19:02:19preglowand it's a huge job
19:02:23CoCoLUSi know
19:02:35CoCoLUSbut the whole windows forms implementation...
19:03:06CoCoLUSi think they will never get it so far so you could take just any windows application written in c# and run it unter mono
19:03:12preglowahhh, that might be so
19:03:14CoCoLUSif it has any kind of graphical output :)
19:03:34preglowbut windows forms stuff is very tied to the windows api, so small wonde¨r
19:03:49CoCoLUSas is the whole drawing stuff
19:04:27CoCoLUSback to "the roots"
19:04:48CoCoLUSsound with our beloved ihp-1x0
19:05:17CoCoLUSi think i'll go listen to sine waves :)
19:06:00rasherare there sine waves in the current cvs?
19:06:07Chamoisspeakin sounds, any news about sounds on iriver ?
19:06:25preglowdon't think there's any sound yet
19:06:45Chamoissome codecs seem to be ready
19:07:11Chamoisor in the good way
19:07:17HClthere's a sinewave thing on the wiki...
19:07:19preglowwell, ready and ready
19:07:22preglowthey compile, at least
19:07:53CoCoLUSok so i don't go listening
19:08:21preglowHCl: where?
19:08:33HCliriver port
19:08:41HCllow level code
19:08:44HClaudio driver
19:08:48preglowwell, yes, but he broke the code
19:08:54preglowdon't know if he's gotten it working now
19:08:55HClxshock fixed it
19:08:58HCla few days ago
19:09:13CoCoLUSyesterday, i think
19:11:52preglownot much of a sine
19:12:23HClwhat does it sound like?
19:12:31preglowbut there's sound, alright
19:12:42preglowbut calling this a sine is the overstatement of the century
19:12:57preglowi don't think sound quality was hubble's focus :P
19:14:22preglowHave fun.. Almost works, it produces noise of what should be a nice sinus wave..
19:16:30preglowso i guess he's aware of the issue ;)
19:16:52rasherguess I messed that up
19:17:59rasher- Remove dummy i2c_init function from firmware/drivers/i2c.c
19:18:02rasherI don't quite get that
19:18:25preglowremove the dummy i2c_init :)
19:18:26rasherI mean I see the dummy funktion
19:18:33preglowline 285, it's right there
19:18:56rasheroh.. I think I get it now
19:20:32CoCoLUShas anyone experience with gentoo?
19:20:44Quelsaruknot me
19:20:46CoCoLUSie setting up a m68k cc toolchain with portage :)
19:21:00rasherthat is ugly!
19:21:00preglowi have
19:21:24preglowbut i just compiled the toolchain by hand
19:21:37preglowrasher: not lovely, no
19:22:29CoCoLUSwon't building my own gcc with m68k support break the existing gcc/portage?
19:22:53preglowthat is, i just put everything in a separate dir
19:23:00CoCoLUShow long did it take you?
19:23:07preglowlet's say ten minutes
19:23:13preglowperhaps a bit more
19:23:19CoCoLUSthe whole compiling process?
19:23:23preglowbinutils + gcc
19:23:39CoCoLUSraw power of ghz i think
19:23:49preglowmy box is a 1.2 ghz duron
19:23:56preglownothing to brag about
19:23:56CoCoLUShm... i'll just try :)=
19:24:16CoCoLUSyou found a guide or something or figured out everything on your own?
19:26:19preglowwiki has a guide
19:27:34CoCoLUSBuilding and Debugging ROCKbox using Eclipse
19:27:43CoCoLUSsounds nice, since i know eclipse fairly well
19:30:45preglowbut i'll go boot linux
19:30:47 Quit preglow ("leaving")
19:30:56*rasher punches the iRiver usb handling in the face
19:32:40CoCoLUSwon't help :)
19:33:03rasherIt'll help me :)
19:34:40CoCoLUSyou think the iriver firmware will bring a lot of attention to the rockbox project?
19:35:01rasherah, rockbox hard to say I guess
19:35:20Quelsarukamiconn: are you here?
19:35:44Quelsaruknow i get a splash message when i boot rockbox
19:36:01Quelsaruksomethhing like playlist buffer empty or something like that
19:36:06amiconnI know, have to investigate that
19:36:26Quelsarukso i don't need a goat to clean that
19:36:32amiconnIt is caused by the modified viewers.config, but doesn't seem to hurt
19:44:59Quelsarukrockboy is only black and white
19:45:02Quelsarukisn't it?
19:45:32CoCoLUSbecause of the iriver display i assume
19:46:05Quelsarukand i suppose that it's hard to make it somehow greyscale...
19:46:52CoCoLUSi own that damn player a year and can't tell without looking at it if it's grayscale or monochrome...
19:46:53amiconnThe iriver display is able to do 2 bit greyscale - an exact match for the gameboy display.
19:47:21Quelsarukamiconn: but archos one?
19:47:25amiconnHowever, this mode is not yet used in rockbox, and I guess it's not high priority to add this
19:47:40amiconnArchos does only b&w natively
19:47:43Quelsaruki think they want music before greyscale
19:48:04CoCoLUShowever, using the full capabilities of the display is quite important imho
19:48:06amiconnThere is my grayscale lib, but this would eat too much performance to also handle the gameboy emu
19:48:19Quelsarukyeah amiconn, but if you were able to "play" snow and cube in archos player...
19:48:33rasherYou are.
19:48:52rasherI read that wrong.
19:48:59Quelsarukrasher: :D
19:49:06rasherIGNORE ME
19:49:08rasherI AM NOT WELL
19:54:31amiconnHCl: r u there?
19:55:35Quelsarukamiconn: but your greyscale uses more than just 2 bit greyscale, isn't it?
19:55:56Quelsarukso maybe.. if it's only 2 bit greyscale?
19:56:23amiconnHowever, the amount of cpu power taken by the refresh does not depend on the number of greyscales.
19:56:38amiconnIt only depends on the size of the greyscale overlay.
19:57:10amiconnFullscreen greyscale takes about 50% cpu power for the refresh alone.
19:57:37 Join preglow [0] (
20:01:01izzyI have the rockblox in vertical mode but the line detection doesn't work :(
20:02:26rasherGood job!
20:02:28rasherwell, apart from the line-detection:)
20:03:48izzyit's not that much fun yet :)
20:06:46 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
20:12:50 Join pappou [0] (
20:14:45CoCoLUSnow setting up that buildchain works like a charm, really
20:14:45 Quit [IDC]Dragon ("CGI:IRC")
20:14:52CoCoLUSi can't even configure the binutils
20:15:04 Join amiconn_ [0] (
20:15:20 Quit amiconn (Nick collision from services.)
20:15:24 Nick amiconn_ is now known as amiconn (
20:18:43CoCoLUSgod hates me
20:19:40preglowgod didn't make autoconf
20:19:45preglowwhat's wrong?
20:21:56CoCoLUSgot it working now
20:22:02CoCoLUSat first it didn't find install-sh
20:22:07CoCoLUSwhich was there
20:22:33CoCoLUSthen i had to do a manual automake -a because some scripts were missing
20:22:37CoCoLUSand now ltconfig was missing
20:23:23CoCoLUSand now a compiler error...
20:23:34CoCoLUSdid god made gcc? :)
20:23:53CoCoLUS *** No rule to make target `../bfd/bfd.h', needed by `size.o'. Stop.
20:33:00 Quit Nibbler (Remote closed the connection)
20:33:48 Join Nibbler [0] (
20:33:49 Quit Nibbler (Remote closed the connection)
20:34:20 Join Nibbler [0] (
20:35:00CoCoLUSi'm missing a configure script
20:35:08 Quit cYmen (Read error: 113 (No route to host))
20:35:10CoCoLUSit's just not in the binutils cvs
20:36:18 Quit Nibbler (Remote closed the connection)
20:36:30 Join Nibbler [0] (
20:36:30 Join NibbIer [0] (
20:36:56 Quit NibbIer (Client Quit)
20:39:11 Quit Nibbler (Client Quit)
20:41:05HClamiconn: back
20:41:24 Join Nibbler [0] (
20:41:24 Join NibbIer [0] (
20:41:26 Join xNibbler [0] (
20:41:33amiconnRockboy now does >8 fps in mario title screen on archos :)
20:41:45 Quit Nibbler (Client Quit)
20:41:47 Quit xNibbler (Client Quit)
20:42:16preglowhot damn
20:42:19 Nick NibbIer is now known as Nibbler (
20:42:32preglowwhere'd you get the last two fps from? more lcd opt?
20:42:40amiconnHCl: I also have a new C version of vid_update that should run faster on iRiver too.
20:43:11amiconnpreglow: By not even calculating the dropped scanlines
20:45:39CoCoLUSis frameskip implemented?
20:45:49HCli was already not calculating dropped scanlines though
20:47:05amiconnI know. You already drop scanlines > 128
20:47:19HClsend me the new vid_update?
20:47:23amiconnNow I drop scanlines > 128 *and* all odd lines
20:47:32HCli'm curious to your asm routine
20:47:35HClwell, iriver does draw odd lines
20:47:42amiconnYes, of course
20:47:45HClok ok
20:47:55amiconnMy new C routine should be faster on iRiver too.
20:48:01HCl :)
20:48:17 Join cYmen [0] (
20:48:19HCli want to see your asm too, see if i can convert it to m68k
20:48:23amiconnI timed my C routine vs. yours on archos: 6.07 fps vs. 5.00 fps
20:48:38amiconnMy asm doesn't improve much on that - 6.16 fps
20:48:45amiconnThat was before dropping the odd lines
20:48:59amiconnOne mom - you'll need more than one file
20:49:10 Part pappou
20:50:55amiconnHmm. How do I find out what changes I did...
20:51:36CoCoLUSdiff :)
20:51:44amiconnAgainst what?
20:52:04amiconnHCl: Should I rename iriver.c to rockbox.c ?
20:52:16CoCoLUSthe older version you saved... ? :)
20:52:18HClyea, i guess
20:53:07amiconnHmm, that's a bad idea. Too confusing...
20:53:32amiconnSo how do we name it?
20:53:46CoCoLUSiriver_rockbox ?
20:54:07amiconnCoCoLUS: It's no longer iriver only, that's the point
20:54:28 Join XShocK [0] (
20:55:22amiconnI'll leave it as iriver.c for now
20:56:46***Saving seen data "./dancer.seen"
20:58:28amiconnHmm, dcc send doesn't like me :(
20:59:35 Quit markun ()
20:59:36CoCoLUSworking binutils
20:59:37CoCoLUSnow gcc
21:00:03HCleh, sorry, i was slow
21:00:12HClyou can use if its not working
21:01:36amiconnThere are 2 files...
21:03:38HClyea, i hadn't thought of hardcoding that stuff
21:03:56HCleliminates the bitshifts
21:05:27amiconnThe asm is a bit better, because it tests the bit, then shifts in the result (from carry bit)
21:06:34amiconnBtw: The new C routine also eliminates the extra scanline pass in lcd.c, because it tests bit 1 of the source directly
21:07:25amiconnWhat framerate do you get now (reference: Mario title screen. This seems to be little cpu-intensive, so it mainly tests lcd refresh performance)
21:07:50 Quit Quelsaruk ("KVIrc 'Realia'")
21:08:40 Join bg_ [0] (
21:09:51 Join einhirn [0] (
21:16:46HClhaven't tested
21:16:51HCli'm rather busy with a lot x.X
21:21:55 Join Tang [0] (
21:22:18TangHello rockbox :)
21:22:33Tangand all other people ;)
21:22:47TangWhy this smiley?
21:23:00HClhi o.o....;
21:23:10Tanghello of course HCl
21:23:37TangSo what's new about the iRiverport?
21:24:01TangThe Coldfire always run at 19MHz instead of 140MHz?
21:24:21Chamois11 mhz i tought
21:24:25Chamoishi tang
21:24:42CoCoLUSi got that toolchain working :)
21:24:49TangHi Chamois
21:28:17TangHi CoColus
21:28:54CoCoLUShi Tang :)
21:29:06TangiHP owner?
21:29:56amiconnHCl: What is on your ftp?
21:30:09HClthat was my first testcase
21:30:16HClhomebrew demo rom
21:30:29HClneeded something really basic to test rockboy in its early stages
21:31:04TangSo the CPU always run at 11MHz?
21:31:30HClat the moment, yes.
21:31:37HCllinus is working on it.
21:32:01CoCoLUSihp-overclocking, sounds nice
21:32:12amiconnNo overclocking.
21:32:23HClno overclocking planned
21:32:30HClbut its possible for people who are dumb.
21:32:35HCland seriously want to try it
21:33:20TangThanks HCl
21:33:26Tangi see
21:33:54Tangwas just wondering as sometimes some progress weren't in the Wiki
21:33:59CoCoLUSif the gameboy emu gets fast enough, we won't need overclocking ;)
21:34:08Tangand i've not access to IRC often
21:34:16HClgameboy dynarec should provide more than enough speed.
21:34:50Suckadouble speed mario!
21:35:04CoCoLUSone magic mushroom too much :)
21:35:35preglowTang: clocking the cpu correctly takes alot of time, i'm pretty certain the wiki will be updated once it's done
21:35:44preglowhaha, i'd hardly call fly agarics "magic"
21:35:51preglowdangerous, perhaps
21:36:10preglowit's a wonder mario's not foaming at the mouth and raging around like a mad man
21:36:15preglowhe must have built up a tolerance
21:36:22TangAh okay thanks Preglow
21:36:31Tangno misunderstanding
21:36:38Tangi'm not making pression
21:37:05CoCoLUSbest thing would be playing mario while listening to some songs, gapless, ogg
21:37:13Suckamario has moved onto ketamines ;)
21:37:20Tangi do know an open source project is based on vonlunteerness and free hours
21:39:36preglowSucka: he'll be doing smack before long
21:39:49preglowit's a wonder he still lives to feature in gamecube games
21:39:50Suckamario mario mario....what happened
21:40:33Suckayou know its all going pearshaped when you start hallucinating giant eggs with green spots
21:42:29bg_hehe... it would be funny if someone photoshopped up an enquirer-esque picture of mario looking tired and confused from drug use
21:42:53preglowagainst a psychedelic background
21:42:57preglowshrooms coming out of his pockets
21:43:13bg_the guy does it all
21:44:02bg_speaking of mario, this is hilarious
21:44:27Suckaawesome flash
21:44:49CoCoLUS :)
21:45:21preglowCoCoLUS: ahahha, i'd love that shirt
21:46:17CoCoLUSbuy it
21:46:29 Quit Chamois ("CGI:IRC")
21:47:57preglowcan't bloody well find it
21:56:13 Quit bg_ ("Chatzilla 0.9.67 [Firefox 1.0+/20050201]")
21:59:01 Join xen` [0] (
22:07:04Tangi go
22:07:11Tangbye folks
22:07:11 Quit Tang ("Chatzilla 0.9.66 [Mozilla rv:1.7.5/20041108]")
22:18:06 Join Quelsaruk [0] (~kvirc@
22:18:09Quelsarukhi again
22:23:16 Quit preglow ("TYKJI")
22:25:31 Join preglow [0] (
22:31:30 Join Patr3ck [0] (
22:35:56 Join ashridah [0] (
22:46:09*Bagder sneaks in
22:46:17Quelsaruka Bagder
22:48:29 Join FlexH [0] (
22:49:28FlexHthere is a lot of people here :)
22:49:49Bagderbut most of them are idle atm
22:50:04*amiconn is not idle
22:50:13*amiconn tries to understand make...
22:50:44FlexHi had a JB recorder, now i have an iriver H320 :)
22:51:14FlexHi'd like to help rockbox team but i don't kwow how
22:54:06Bagderamiconn: anything I can do to help?
22:56:50***Saving seen data "./dancer.seen"
22:56:52amiconnI don't get what this depfile thing means, and how it is supposed to work
22:57:11amiconnI'm trying to get the rockboy Makefile sorted.
22:57:28amiconnI need a flexible method.
22:58:13amiconnFor iriver, rockboy.rock should be the plugin itself
22:58:20Bagderthe depfile stuff is the's method of generating a file with dependencies
22:59:57amiconnI thought that it contains dependencies, but how are the dependencies defined, and what are they used for? It seems some Makefiles work without a depfile
23:00:34Bagderthe dependencies are used so that when you change a header file, all C files that use that header get recompiled
23:00:58preglowa very good arrangement to have, god knows how many times i've had bugs turn up over that
23:01:21amiconnFor archos, rockboy.rock should be a little loader plugin that loads the main rockboy code blob at a fixed address into the mp3 buffer after grabbbing it.
23:02:27Bagder.rock files use (and depends on) includes too
23:03:26amiconnHow could I define conditional rules, i.e. rules valid for one target, but not for another?
23:04:13Bagderit only makes deps for files in SOURCES that survived a preprocess
23:04:53amiconnHmm. How could I explain...
23:05:20amiconnI need a separate liinkfile for rockboy, which will reside in apps/plugins/rockboy
23:05:50 Join coob [0] (
23:05:51amiconnThis links rockboy_main to the plugin start address for iriver, and to the special fixed address for archos
23:06:06Bagderyou mean the plain .rock build rule is not good enough for it?
23:06:27amiconnNo, since rockboy is more than one source file
23:06:54coobto anyone working on the libflac stuff: someone working on ipod linux has optimised mpd's flac stuff quite a bit, so might be worth checking to save yerselves some time...
23:06:54amiconnplus I need to create rockboy.rock differently depending on target
23:07:04Bagderright, the build stuff for plugins assume a single file...
23:07:20Bagderamiconn: then you add a separate $(OBJDIR)/rockboy.rock rule
23:07:28amiconnFor iriver, rockboy.rock would be a simple copy of the linked & stripped rockboy_main.bin
23:08:03Bagdercoob: interesting
23:08:10amiconnFor archos, rockboy.rock would be the usual plugin creation path, and I would have to create an additional file
23:08:49amiconnIm not 100% decided about the name, rockboy.ovl or something like that
23:09:08coobheh is rockboy in cvs yet? we need to pilfer that
23:09:26coobbe a bit shitty to control on an ipod tho
23:09:31amiconncoob: Guess why I'm fiddling with that make stuff? :)
23:09:39coobah :)
23:10:07coobi take it most of the modifications were libc stuff and the sys/iriver thing?
23:10:15Bagderthose ipodlinux guys should go Rockbox
23:10:20amiconnBagder: Is it possible to use some if...else... construct for defining the rules?
23:10:28Bagderamiconn: yes it is
23:10:37ashridahhrm. what resolution does an ipod's screen have?
23:10:42 Quit Quelsaruk ("formatting the computer")
23:10:45coobsame as iriver..
23:10:54HCldoes rockbox work on ipod?
23:11:17BagderHCl: no, but it could probably be done fairly "easy"
23:11:25HCljust curious.
23:11:38rashersince there's already an alternative system on it
23:11:43HCli'm kind of wanting to get rockboy up running to full speed on iriver.
23:12:17rasherand open
23:12:30coobyeah all the hard work's done for you :)
23:12:46Bagderfor us?
23:12:48 Quit ashridah ("damnit.")
23:12:53amiconnBagder: One thing I really need to understand: How does make in apps/plugins/ detect whether something changed in apps/plugins/lib/ and the plugin lib needs rebuilding?
23:12:56 Join muesli- [0] (
23:13:00coobstill any more work on the hw stuff would be invaluable... we basically have 1/2 people who are skilled enough/have the time for the hw side of things
23:13:33HCli'm not much of a hardware person
23:13:38Bagdercoob: that would be another benefit of going Rockbox ;-)
23:14:12Bagderamiconn: it always runs make in that dir to check
23:14:20coobwell now you're porting to iriver i guess a lot more code can be shared, especially on the codec side of things
23:14:20Diwayhum, how do I get the 'codecs' dir with cvs ? I can't compile rockbox without it :/
23:14:42amiconnBagder: How?
23:14:44Bagdercoob: yes. we're porting to the Gmini as well
23:15:20Bagderamiconn: the "all:" line says it depends on the libplugin, and then on the libplugin rule it runs 'make -C lib'
23:15:39coobhow's ram latency with the iriver? tremor runs like shit cos the ipod's only fast ram is the 96k of iram..
23:15:41amiconnAh cool, was just about to ask that :)
23:16:13Bagderram latency?
23:16:34rasherwouldn't the lowmem tremor fit in that?
23:16:42coobthe arm stuff in the ipod there's a lowmem tremor?
23:16:53coobthere's a lowmem tremor?
23:17:03 Join ashridah [0] (
23:17:16coobyeah the arm stuff has no cache
23:17:59coobheh quake on a gmini would be fun.
23:18:04coobah cool thanks
23:18:34Bagdermakes me suspect that is called for by the Neuros guys
23:19:02Bagderbut I won't complain ;-)
23:19:26rasherany news about the removal of 64 bit ints from flac?
23:19:53 Quit Cassandra_ (Read error: 110 (Connection timed out))
23:20:15rashernothing in that neuros thread
23:20:38 Join Cassandra_ [0] (~christi@
23:24:29FlexHhave a good night :)
23:24:45 Quit FlexH ("Si un jour, vous entendez une huitre chanter la marseillaise, fuyez ! C'est une huitre nationnaliste !")
23:26:30coobok yeah going through the ipl archives and stuff using lowmem didn't have much an effect
23:27:15preglowi don't think we'll need lowmem
23:27:21 Quit muesli- (Read error: 104 (Connection reset by peer))
23:27:35 Join muesli_- [0] (
23:28:00preglowlist of memory requirements there
23:28:15rasherah, that's the one
23:28:46preglowmight want to wiki it if it's useful
23:29:33rashercould be for the gmini
23:29:51rasherif I got that right
23:31:37Bagdermuch larger than 96K anyway
23:31:52rasheroh well
23:31:55rasheradded to the wiki
23:32:04amiconnBagder: Where do all those tool definitions for the Makefile (CC, OBJCOPY etc.) come from?
23:32:29Bagderamiconn: the root Makefile, in the build dir. it 'exports' them so they are globally accessible
23:32:51Bagderamiconn: that's a GNU make feature not globally available
23:33:27amiconnI need 'cp'
23:33:51Bagderwell, cp is gonna be cp on all builds
23:34:13BagderI think we can consider the msvc build to be history
23:34:20 Join hubble [0] (
23:34:28amiconnIt does look like that, yes.
23:35:04amiconnMost probably the msvc make will not work any more since I added the multiple background pictures to the win32 sim
23:35:34Bagderand my upcoming "unification" will be very hard to adapt to msvc
23:36:04amiconnYes, most probably.
23:36:28Bagderif msvc should be used again, we should still use the regular makefiles and just setup the proper CC and flags etc
23:37:44amiconnI don't know how msvc handles the build process at all. I never used it.
23:38:45hubblehum.. not sure about interrupt levels, are int 0 higher prio than int 7 on the mcf5249?
23:38:48Bagderthere's a Makefile.vc6 for it
23:39:31*Bagder hasn't read up much on the 5249, so he doesn't know
23:46:12Bagder"getting rid of the 64-bit stuff is going to be *really* hard" / - Josh Coalson - FLAC developer -
23:47:21 Join LinusN [0] (
23:47:22hubbleBagder: but it should be possible to decompose 64-bit operations with 32 ops.. like one 64-bit mul = 4 x 32-bit muls ?
23:47:53Bagderhe addressed that as well
23:47:58Bagder"the actual math functions are the easy part, what's hard is hooking it up everywhere."
23:48:09hubbleok =)
23:48:19 Quit Nibbler ("blubber")
23:48:33hubbleadding 64-bit intrinsic functions to gcc? :D
23:48:43Bagderon the forum, URL on the SoundCodecs wiki
23:49:13amiconnWhat's the point of simple decomposing? Isn't it exactly what the gcc 64 bit math library functions do?
23:49:15Bagderthey referred to this:
23:49:30 Join toolmanwv [0] (
23:49:31LinusNint7 is the highest
23:50:04Bagder better url
23:50:27hubbleLinusN: hum.. i'm trying with dma and interrupts to transfer data to PDOR3.. how to select int level of the DMA0 int and also what is "Autovector"?
23:51:38LinusNautovector is using the 7 vectors for each interrupt level instead of passing the vector to the core
23:51:58 Join skav [0] (
23:53:24LinusNthe dma0 priority is set in the ICR6 register
23:53:45hubbleLinusN: hum.. so i should not use Autovectoring?
23:54:01LinusNno, use the vector in the dma controller
23:54:20LinusNor rather, program the dma controller to use the vector i chose in system.c
23:55:35LinusN54, iirc
23:56:26LinusNhave you got the dma running?
23:56:38hubbleLinusN: not yet =)
23:57:04LinusNi wish i had time to play with that
23:58:19 Join amiconn_ [0] (
23:58:19 Join Digital007 [0] (
23:58:30XShocKDMA moves data to pdor3, but I can't understand what should i write to DIVR, the interrupt
23:58:30 Quit amiconn (Nick collision from services.)
23:58:30 Nick amiconn_ is now known as amiconn (
23:58:37Digital007how is the sound progress for iriverbox coming on?
23:58:46hubbleXShocK: DIVR is the interrupt number the DMA generates
23:58:47 Quit midk ("Leaving")

Previous day | Next day