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 2005-04-19

00:00:13micoonobody asked to join the team and try to port the gmini400?
00:00:27micoowhere in germany?
00:01:08 Join lostlogic [0] (
00:01:48gromit`jyp and strath (here) have been working on the gmini
00:02:03micooor the other ones?
00:02:07gromit`other ones
00:02:20gromit`maybe some common parts
00:02:22micoothe 400 is a total diff device
00:02:26Shagnarmicoo near stuttgart
00:02:34gromit`ya we know
00:03:03micoogood ot know shagnar :p
00:03:15Shagnarhehe :)
00:03:29Shagnarhave you ever been in germany?
00:04:00micooi've been in munich
00:04:09micoobut not for long
00:04:26micoojust passed there to go to austria
00:04:50ShagnarI've never been in munich yet ^^
00:04:58micoobut i've been to poland, italy, france..
00:05:01micooand mm
00:05:08micooaustria of course
00:05:25micooandi can't remember where else
00:06:18micoohmm i've just landed in munich
00:06:37micoobut i do remember it was damn scary in the airport
00:07:12micoothere were all these soilders on the second floor aiming their guns to the first floor
00:07:57micooi guess that's cause of what happened in the olympic games
00:08:02Shagnarsounds strange
00:08:15micooi'm sure its not the same now
00:08:19micooit was back in 94
00:08:43Shagnaroh, long time ago^^ 94 i was just 6 years old *gg*
00:09:07micoohmm it was guarded because of what happened in the olympic games in 70' something
00:09:21micoowhen 7 isralies were murdered
00:09:27micoodo you know anything about it?
00:10:09micoooh so you're 17 right?
00:10:09Shagnarnot really...
00:10:15Shagnarexactly :)
00:10:46Shagnarnot very old, i know ;o)
00:11:24micoowell im 17 wither
00:11:31micoocheck that out
00:14:54 Quit asdsd_ ("Trillian (")
00:16:14Shagnarhorrible thing
00:17:02micooi'm pretty sure that's why it was so guarded
00:18:05Shagnarbut i think better guarded and save than not guarded and perhaps a blood bath...
00:18:40micoosure do
00:19:33Shagnarthats very difficult sometimes, the only thing i hear about israel is in the news when some people were killed.... kinda strange
00:19:42 Join DangerousDan [0] (
00:20:22 Quit ashridah ("Leaving")
00:20:28micooyou mean
00:20:32micoothe bomb suicides
00:20:42Shagnarfor example...
00:21:04micooyea, it's been very scary lately
00:21:48micooalthough it's a bit quiet in the past few months
00:21:49 Quit matsl (Read error: 54 (Connection reset by peer))
00:22:16 Quit DangerousDan ("Miranda IM! Smaller, Faster, Easier.")
00:22:44Shagnaryep. hope will go on like that (no more bomb suicides)
00:23:49micoowe all do
00:25:38Shagnarof cours
00:32:25 Join XShocK [0] (
00:32:55rashumsafternoon XShocK
00:35:14 Quit Harpy (Read error: 60 (Operation timed out))
00:36:07 Join Tang [0] (
00:37:31micoogn guys
00:37:32XShocKhi, rashums
00:37:49 Nick rashums is now known as rasher (
00:38:15XShocKrasher, ah, not you look more familiar. :)
00:39:00 Nick stevenm|food is now known as stevenm (
00:39:00Shagnarmicoo: gn8 :)
00:39:11micoogn man :)
00:44:13 Quit Ka (Nick collision from services.)
00:44:51 Quit DMJC ("Leaving")
00:46:59 Join Ka_ [0] (
00:47:05stevenmhey rasher, can you do another time test for me?
00:47:08stevenmpleas ?
00:48:06stevenmrasher, this one is at half rate so it better take no more than 5 minutes
00:52:38Shagnarstevenm: my H just finished!
00:52:50stevenmShagnar, hmm ?
00:53:32stevenmShagnar, do you feel like testing a (somewhat) optimized 'new' version?
00:53:37Shagnarmidi2wave for the 1min file
00:53:48Shagnarabout 2h^^
00:53:51 Quit thegeek (Read error: 54 (Connection reset by peer))
00:53:52Shagnarwhy not :)
00:53:56stevenmI expect this to take about 5 minutes at 120Mhz
00:54:02stevenmhere lemme tar it up and send it your way
00:54:15Shagnarall right
00:56:11stevenmShagnar, okay.. grab this: stevenm/midi.tbz2">
00:56:26stevenmuntar that into rockbox code/apps/plugins/ and build
00:56:58stevenmput the midi2wav.rock file into .rockbox/viewers and there should also be a file, mi1.mid in apps/plugins/midi. fire that up.. see how long it takes
00:57:02stevenmmake sure iriver is set to 120Mhz
00:57:45stevenmShagnar, I am expecting 5 minutes, hoping for less than that
01:00:36HClwhats the goal?
01:00:43HCl(how many minutes?)
01:01:00stevenmHCl, 1:30
01:02:24Shagnarstarting now
01:02:58preglowstevenm: for it to be truly usable it'll have to be at least 200% realtime, really
01:03:01stevenmexcellent, thanks
01:03:11stevenmpreglow, yea.. don't remind me :(
01:03:19stevenmsadly you are very right
01:03:20preglowbut of course, there'll always be files these codecs can't take
01:03:51stevenmpreglow, now it's at 22.05k sampling rate btw. if we ever get it fast enough, trivial changing it back
01:04:12Shagnarshould be enough for midis?
01:04:26stevenmOh yea, just quality a bit worse
01:04:39stevenmShagnar, how much time elapsed? I had watch problem
01:04:53stevenmShagnar, it is 120Mhz right ?
01:04:59Shagnar2 minutes
01:05:19Shagnaryes i think. set it before, hope it's still set
01:05:40stevenmwell I don't see why it would go over 5 min. synth is doing half the work
01:07:29stevenmpreglow, well we can always leave it as a 'conversion tool' that generates files to be played by the other codecs..........
01:07:32Shagnarstill working (about 4 minutes)
01:07:37stevenmif you have 2 hours to kill that is
01:08:38preglowstevenm: sure, but i think we'll be able to make it useful
01:08:54 Quit Sucka ("a bird in the bush is worth two in your house")
01:09:02stevenmpreglow, I really really hope so. we put a lot of work into that code
01:09:25stevenmShagnar, wait a second ! !
01:09:36stevenmThat file I gave you. It could not have POSSIBLY worked
01:09:37Shagnarthe H is still working ;)
01:10:02stevenmShagnar, I made clean, but I do not remember running ./configure and changing the target to Normal
01:10:17preglowden indeed, it should have crashed the player
01:10:18stevenmdude yea it rebuilt the Sim again
01:10:20stevenmand I DID make clean
01:10:34Shagnarkk, i'll reset it ;)
01:10:39Shagnar(my H)
01:10:56 Quit gromit` (Read error: 60 (Operation timed out))
01:11:17Shagnaror shall i wait ?!
01:11:20stevenmShagnar, how the heck is it running at all ?
01:11:34stevenmShagnar, does it say START PLAYING ?
01:11:36Shagnar[START PLAYING] + working HDD
01:11:43stevenmShagnar, then you have the wrong file
01:12:02stevenmif you loaded the one that I just put up, it would have said Bad Plugin or something. It should have never ran- it's built for x86
01:12:21stevenmwow that was dumb of me
01:12:42Shagnari wondered because of the filesitze (the one on my H had abaout 16kb, the new one 106kb, but i copied it....)
01:12:57Shagnardunno whats going on then, i'll reset the player and have a look, one moment
01:12:58stevenmShagnar, did you copy it to .rockbox/viewers or to / ?
01:13:05Shagnarto /rocks
01:13:16Shagnari think.... i'll look
01:13:29stevenmnaw, it's gotta be viewers
01:13:36stevenmlemme put up a good one this time
01:14:38Shagnarsorry for copying it into the wrong folder, i thought .rock has to be copied in /rocks.... my fault ;)
01:14:48stevenmnah its ok. it wouldnt have worked anyway
01:14:53stevenmthere, that one is built for coldfire
01:14:59preglowit would have crashed your player
01:15:05preglowso just as well you didn't copy it right
01:15:15stevenmit should say "Starting Playback" instead of "START PLAYING" now, jic
01:16:18stevenmlemme know what happens ..
01:17:19Shagnarokay, this time i'll start a better timer to have the time
01:17:31Shagnarjust checking if the frequenzy is also the right one
01:18:28Shagnarjust started :)
01:18:39Shagnarmuc more hdd work :)
01:18:55stevenmthat's expected.. 1/2 the sample rate
01:19:05stevenmpreglow, what are the chances that at this point HD access is slowing it down?
01:20:14 Quit Seed (Read error: 110 (Connection timed out))
01:20:28Shagnar2min, still working :)
01:21:17Shagnarfor the MI1.MID
01:21:29stevenm42% realtime
01:21:38rasherwhat *DID* you do?
01:21:44Shagnaryeah :o)
01:21:56stevenmrasher, half the sampling rate, plus store a value that was shifted like a zillion times
01:22:12Shagnarbut stevenm
01:22:16stevenmrasher, wait a sec .. half the sampling rate should have been 4 minutes, all other things being equal
01:22:16Shagnarthe miditest.wav
01:22:21stevenmShagnar, yes ?
01:22:22Shagnardoenst seem to sound very nice...
01:22:29stevenmShagnar, eep !
01:22:35Shagnarshall i send it to you?
01:22:44stevenmlemme produce it here first
01:22:54stevenmit sounded fine on the soundcard..
01:23:13Shagnarthe length of the wave is abaout 43 seconds
01:23:19Shagnarthe original midi is 1:35...
01:23:29stevenmShagnar, oh I know why
01:23:35stevenmI changed the sampling rate of the SYNTH
01:23:49stevenmbut the but the sampling rate of the wave file is STILL set at 48Khz
01:23:58stevenmmy bad
01:24:14stevenmShagnar, it's okay though. Open it with goldwave and play it at say, half speed. it sbould be fine
01:24:35stevenmYou can send it to me if you wanna, I will make sure
01:25:03preglowstevenm: goes for telling you how much you can do in c
01:25:24stevenmpreglow, wow. again, keep in mind- half the sampling rate
01:25:29preglowstevenm: i know
01:25:31stevenmand there's still that weird issue at using it half
01:25:43 Quit Tang (Read error: 110 (Connection timed out))
01:26:01stevenmpreglow, I can still probably do some more.
01:26:03preglowxxx2wav will report the wrong realtime ratio if you don't set its sample rate correctly
01:26:07preglowof course you can
01:26:11preglowthere's probably tons more to be done
01:26:12stevenmpreglow, like, some IF statements OR two constants, for instance
01:26:17preglowthat's why i'm pretty optimistic yet
01:26:25Shagnarstevenm: just played it with 24000 but ... still these "scratching" sounds in the background.... don't think thats right
01:26:45 Join thegeek [0] (
01:26:47stevenmwait, I counted wrong. That's 60% realtime, not 42%
01:27:08stevenmShagnar, scratching? Could be an 8bit sample that I haven't added support for
01:27:11stevenmShagnar, send me ?
01:27:33Shagnaryes. what format? (zip/rar)
01:27:50stevenmi dont care
01:27:54stevenmzip I guess
01:28:40Shagnar7mb ... will take some minutes
01:29:13stevenmShagnar, are you referring to low-pitched buzzing on the left channel ?
01:30:29Shagnarboth, left and right
01:30:56stevenmI am hearing a low-pitched sound of like, a cicada flying, on the slight left
01:31:22stevenmI think I know what you are saying.. but I wanna see that file anyway.
01:31:36Shagnar3/7 mb upped
01:32:29stevenmah, cool
01:32:33stevenmShagnar, you on dialup ?
01:32:52Shagnarno, dsl but only 128kbit/s up ;)
01:33:09stevenmah. beats comcast- 32kb
01:34:30Shagnarabout 1mb to go
01:34:52stevenmAh. I think it's the flute patch doing it
01:35:08stevenmI smell a looping problem. I re-did that today
01:35:51Shagnareasy to solve?
01:36:12Shagnarstevenm -
01:37:33stevenmShagnar, thank you
01:37:43Shagnarno problem :)
01:37:51stevenmShagnar, hopefully. I know where it is approximately
01:38:04***Saving seen data "./dancer.seen"
01:38:08Shagnargood to hear :]
01:38:23stevenmnow we'll go hunting in synth.c
01:38:55stevenmShagnar, woah!! I hear WHITE NOISE
01:39:08stevenmShagnar, did you hear music, or was it just hissing ?
01:39:22Shagnarwell, not much of the music ;)
01:39:41stevenmOoooookay, what the HECK ?
01:40:03stevenmShagnar, that is not what I was getting on this end .. ..
01:40:16stevenmI just hear SSHHHH etc
01:40:33Shagnari had a miditest.wav on the player before and i didn't delete that, perhaps thats the problem? dunno if its deleted at the beginning
01:40:52stevenmthe one I found was a slight click .. and what sucks more is that I have a Sim build and it works fine here
01:41:06stevenmI would think it is deleted..
01:41:35stevenmthat is severely severely bad
01:42:07stevenmhey rasher, could you maybe try for me a fresh build of this ?
01:42:20stevenmmaybe it has to do with different versions of the rest of the program.. though unlikely
01:42:36Shagnari'm gonna make it a second time without an exisiting miditest.wav...
01:42:49stevenmShagnar, sure, try.. but I don't think that will help
01:43:57stevenmShagnar, that actually sounds like that it did when I had a sign mismatch within synth.c
01:44:21stevenmShagnar, but it sounds perfectly fine on sim
01:44:26stevenmthis is just great
01:44:33rasherstevenm: it's your lucky day.. url me
01:45:13Shagnarreally great? working on sim but not on the H ...
01:45:34stevenmrasher, stevenm/midi.tbz2">
01:46:04stevenmrasher, just go to line 139 of midi2wav.c and change the 48000 to SAMPLE_RATE
01:48:31stevenmrasher, awesome
01:48:44Shagnarstill the same, as you said
01:48:47stevenmrasher, the ones that took 10, 8, minutes, etc, did those sound right at least ?
01:48:52rasherhow long can I expect it to take?
01:49:03stevenmrasher, this time it took 2:31
01:49:16rasherI think so.. didn't listen to all of them
01:49:20rashersome of them did
01:49:35stevenmrasher, but the time it did that, it generated UTTER crap
01:49:35stevenmrasher, well it was the right length.. and sounded remotely like it.
01:50:17stevenmrasher, this one sounds like there is a sign problem, and I hope to god that it's not a difference between how gcc generates code differently for iriver and x86
01:50:46stevenmpreglow, by any chance, is the endian order different between x86 and coldfire code ?
01:51:15stevenmpreglow, so if I read in an array of 16 bit data, and then cast it to short *
01:51:33preglowif it's correct for m68k, it should be wrong for i386
01:51:44stevenmpreglow, then I would have to byteswap on iriver but not on sim, right ?
01:51:56preglowwell, you said it was right for iriver not so long ago
01:52:07stevenmpreglow, it ran FINE on the sim
01:52:15stevenmpreglow, but we are getting NOISE out of iriver
01:52:51rasher2:20 roughly
01:52:53stevenmpreglow, I guess I have to have a #if statement in the gusloader to byteswap the gusloader to include byteswapping the waveforms if we're on an iriver build
01:52:59stevenmrasher, cool. does it sound like noise ?
01:53:07rasherwhat can I expect miditest.wav to be?
01:53:10preglowstevenm: then you have to byteswap
01:53:24rashercan I play it on the iriver?
01:53:35stevenmrasher, sure.. but you'll still get noise
01:53:44stevenmIm about to pate you a one line hopfeully fix
01:54:56Shagnari'll go to bed then. in 5 hours 've to get up for school :-/
01:55:10Shagnarhope you'll find the wrong line soon :)
01:55:15stevenmShagnar, thanks a lot for your help. I would have never bothered to check
01:55:23stevenmrasher, I am writing it. hang on a sec ..
01:55:38rasherline-noise indeed
01:55:48Shagnarcya guys!
01:56:25 Quit Shagnar ("( :: NoNameScript 3.81 :: )")
01:58:03stevenmrasher, I'll pm you the 3 lines
01:58:17stevenm((short *) wav->data)[a] = (((((short *) wav->data)[a]) & 0xFF)<<8) | (((((short *) wav->data)[a]) & 0xFF00)>>8);
01:58:26stevenmer I don't know if that worked
01:58:43stevenmfor(a=0; a<wav->wavSize/2; a++)
01:58:43stevenm((short *) wav->data)[a] = (((((short *) wav->data)[a]) & 0xFF)<<8) | (((((short *) wav->data)[a]) & 0xFF00)>>8);
01:58:43DBUGEnqueued KICK stevenm
01:59:09stevenmgoes as the first thing inside the if statement inside guspat.c .... line 96
01:59:26stevenmI have no idea if that's even right.
01:59:31stevenmno way to test
02:00:09stevenmI mean, I play it on my end and get noise, if that's indicative of anything
02:01:35rasherI'll try
02:02:02stevenmrasher, thanks
02:02:23stevenmrasher, there could be more noise.. I don't know if it should go before or after the unsigned->signed loop. but I figured before
02:02:46stevenmI am also very antsy about a paper I should have stgarted on an hour ago, so forgive the quick-n-dirtyness
02:02:56stevenmand bugs THRIVE in that
02:03:34rasherrunning now
02:03:40stevenmrasher, sweet
02:04:03stevenmrasher, uhh.. whoops wrong if statement !!
02:04:12stevenmit may get rid of HALF of it
02:04:18stevenmit should have gone before that
02:04:21CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
02:04:21*stevenm kills himself
02:04:53stevenmpreglow, does that even look like the right code to you ?
02:05:07rashertar something up for me? :)
02:05:15stevenmrasher, sure, one sec
02:06:36 Quit thegeek (Read error: 145 (Connection timed out))
02:07:14stevenmrasher, stevenm/midi.tbz2">
02:13:44rasheroops, building
02:14:07preglowstevenm: if you're accessing the individual bytes, you might as well stop casting it to short*
02:14:39preglowbut yes
02:14:41preglowlooks correct
02:14:43preglowbut i'm dead tired
02:14:45stevenmpreglow, I guess... I noticed that about 10 seconds after I put that up
02:14:45preglowand now i'll go to bed
02:14:47preglowlater, all
02:14:53stevenmgood night preglow
02:14:54preglowhappy hacking
02:15:04 Quit preglow ("murfs")
02:15:11stevenmrasher, luck ?
02:15:14rasherunning test
02:15:19stevenmrasher, sweet
02:16:33 Quit cYmen ("zZz")
02:17:33rashersame time..
02:17:48stevenmrasher, cool
02:17:48stevenmsame noise ?
02:17:53 Join Biptoria [0] (
02:18:22rasherlet me test...
02:18:24 Quit Biptoria (Client Quit)
02:18:41rashernot the same noise
02:19:23rasheris this 22khz?
02:19:29stevenmrasher, yes
02:19:34rasherbecause it sure sounds too fast
02:19:38rasherah, figures
02:19:50stevenmdidn't I change it?
02:19:55rasherI don't think "Audio test" plays anything but 44,1khz
02:19:57stevenmoh wait I changed it BACK to see if that was the problem
02:20:12stevenmrasher, does it sound clean ?
02:20:17rasherwant the soundfile?
02:20:21rasherit was interesting
02:20:21stevenmrasher, sure please
02:20:43stevenmrasher, I think the wave header is written as 48000 but the file runs at 22050
02:20:49stevenmI remember changing it back and forth
02:21:20stevenmrasher, how you gonna send ?"> - 6854kb
02:22:34stevenmrasher, thanks
02:22:36stevenmlessee here
02:23:31 Quit stevenm ("Leaving")
02:23:56 Join stevenm [0] (
02:27:48stevenmok, something is DEFINITELY wrong
02:28:53stevenmI don't know what to say.
02:29:23stevenmother than go back to the original method of getSample()
02:29:31stevenmi dont have that in me tonight
02:30:01stevenmI now want to say it is sign related crap
02:30:07stevenmbut I have NO IDEA what goes on there
02:31:13stevenmrasher, thank you for the testing
02:31:39rasheryou're welcome
02:31:43stevenmI have to go write a stupid paper now
02:32:08stevenmmy best guess would be to go back to the oritinal getSample() method of ORing stuff together and then seeing if that works
02:32:11stevenmthen going from there
02:32:29stevenmmaybe flip the sign conversion around
02:40:45DBUGEnqueued KICK silencer_
02:48:14 Join Seed [0] (
03:03:33 Nick tvelocity is now known as tvelocity[away] (
03:13:31 Join asdsd_ [0] (
03:17:31 Quit rasher ("CGI:IRC 0.5.4 (2004/01/29)")
03:20:43 Join amiconn_ [0] (
03:24:28 Quit stevenm ("Leaving")
03:29:50 Quit Ka_ (Read error: 110 (Connection timed out))
03:30:57 Join Ka_ [0] (
03:37:49 Join amiconn__ [0] (
03:38:05***Saving seen data "./dancer.seen"
03:39:16 Quit amiconn (Read error: 110 (Connection timed out))
03:39:16 Nick amiconn__ is now known as amiconn (
03:44:54 Join stevenm [0] (
03:47:17 Quit edx (Read error: 110 (Connection timed out))
03:56:20 Quit amiconn_ (Read error: 110 (Connection timed out))
04:04:18stevenmWell getSample() is back unfortunately. Can anyone test on target?
04:08:26 Quit stevenm ("Leaving")
04:16:53 Quit XShocK (" HydraIRC -> <- Leading Edge IRC")
04:49:33 Quit asdsd_ ("Trillian (")
05:08:51 Quit Aison ("( :: NoNameScript 3.72 :: )")
05:24:41 Join edx [0] (
05:33:55 Quit tvelocity[away] (Read error: 110 (Connection timed out))
05:34:56 Join tvelocity[away] [0] (
05:38:07***Saving seen data "./dancer.seen"
05:50:42 Join stevenm [0] (
05:55:04stevenmHello. Can someone please test my codec on target? I cant do any more optimizing before I know this existing code works
05:56:30 Quit Seed (Read error: 145 (Connection timed out))
06:03:20 Join bagawk [0] (Lee@bagawk.user)
06:15:16 Quit bagawk ("Leaving")
06:37:30 Join LinusN [0] (
06:39:22stevenmHello Linus
06:39:44stevenmQuick hardware question for you
06:40:16stevenmWe got the codec running 60 ish percent realtime but somewhere at some point some major endian badness happened and stuff came out very very distorted
06:40:33stevenmNow then, how does coldfire store its data.. and where does the sign bit go?
06:40:41LinusNit's big endian
06:40:58LinusNsign bit is always msb
06:41:29stevenmIt all worked fine on the Sim, but then on target things became wrong. This was an array of Shorts.. I tried swapping the bytes but to no avail. I suspect there is something up with the signs
06:42:05LinusNwhere's the code?
06:42:13stevenmbig endian.. so, it is sign bit followed by 7 bits of the most significant byte?
06:42:18LinusNwas that your hardware question?
06:42:24stevenmThat was it
06:42:36stevenmI guess not much of hardware as in circuits, but relating to coldfire itself
06:42:42LinusNsign bit is the leftmost bit
06:43:15stevenmI have an array of 16bit signed little-endian data in a file
06:43:30LinusNthen you should swap them
06:43:35stevenmif I load it into an array of unsigned chars, then cast it to short *, it works fine on Sim
06:43:51LinusNbecause the intel cpu is little endian
06:43:58stevenmon coldfire, you have to just swap the bytes, not worrying about the sign?
06:44:45stevenmSo swapping works for both signed and unsigned data ?
06:44:54LinusNcheck load_wave() in apps/debug_menu.c
06:44:59LinusNyes it does
06:45:08stevenmI see
06:45:22stevenmAnd then to go from unsigned to signed, you just subtract 32767 ?
06:45:39stevenm(if it's an unsigned waveform and I need it in signed form)
06:47:55LinusNyes, or add 32768 :-)
06:48:11stevenmAll right
06:48:28LinusNin fact, you should subtract 32768
06:48:40stevenmAh, all right
06:48:50stevenmThis was very strange that was happening
06:49:03stevenmAll worked on PC. No go on iriver. Made it swap, still no go.
06:49:35LinusNit's not strange that it worked on the pc but not on coldfire
06:49:50LinusNbut the swapping should have worked
06:49:51stevenmI have reverted the code to use the less efficient, but apparently working code for that part.. that basically ORs the bytes the way they really should be
06:49:55LinusNcare to show me the code?
06:50:14stevenmOh boy.. if I can dig it up.. so many changes, so much going back
06:50:34stevenm/#if !defined(SIMULATOR)
06:50:34stevenmfor(a=0; a<wav->wavSize; a+=2)
06:50:34DBUGEnqueued KICK stevenm
06:50:34stevenmunsigned char tmp;
06:50:34stevenmtmp = wav->data[2*a];
06:50:36stevenmwav->data[2*a] = wav->data[2*a+1];
06:50:38stevenmwav->data[2*a+1] = tmp;
06:51:00stevenmthen in the synth, wav->data was cast to a (short *) and a single 16-bit signed value was read from it
06:51:29LinusNthere is a swap macro in system.h
06:51:33stevenmworked beautifully on sim. Worked fast on coldfire (since swapping occurs only on load) but noise
06:51:58stevenmso now I am afraid - swapping should have worked.. could have some OTHER optimization broken it on iriver
06:52:16LinusN*p = SWAB16(*p);
06:52:38stevenmand p is a single 16bit value ?
06:52:45LinusNstevenm: sure, the endianness could have bitten you somewhere else in the code
06:52:49LinusNp is a short
06:52:53LinusNshort *
06:52:55stevenmah, all right
06:53:17LinusNshort data[];
06:53:29LinusNfor(i = 0;i < max;i++) {
06:53:33stevenmI didn't THINK it would have any effect elsewhere.. I did convert a bunch of unsigned chars to ints.. and moved a bunch of things into iram.. seemingly nothing should have gone wrong
06:53:43LinusNdata[i] = SWAB16(data[i]);
06:54:03stevenmah gotcha
06:54:22stevenmNow I have gone back to the original way of doing it (no more shorts, but get one byte, shift it, and or it with another, return result)
06:54:29LinusNis this in cvs?
06:54:48stevenmCVS is unoptimized but thankfully working right
06:55:03LinusNwhat other opts have you done?
06:55:26stevenmjust iram, changing unsigned chars to ints
06:55:32stevenmone WEIRD thing though happens
06:55:40stevenmthis is on an x86 by the way that this occurs
06:56:41stevenmthere are two variables, s1 and s2, and they are declared as short
06:56:53stevenmand they have a 16bit signed value assigned to them in the synthesis process
06:57:13stevenmeverything sounds fine if they are short. but if I make them int, I get horrible distortion
06:57:26LinusNshow me
06:57:43stevenmshort s1 IDATA_ATTR;
06:57:43stevenmshort s2 IDATA_ATTR;
06:58:00stevenmblah blah blah, looping code, unrelated. then
06:58:22stevenms2 = getSample((so->cp>>10)+1);
06:58:36stevenmand s1 gets assigned a similar call from somewhere lower.
06:58:49stevenmwell, like so. then:
06:58:52stevenms = s1 + ((signed)((s2 - s1) * (so->cp & 1023))>>10);
06:59:14stevenmand 's' is the output sample, declared as int
06:59:30stevenmthen some more operations on just 's'.. shifting, volume, etc. and it is returned
06:59:47stevenmall works fine when s1 and s2 are shorts, but uber noise if s1 and s2 are int
07:00:54stevenmI am still waiting to find out what broke it for iriver.. I undid all the (short *) stuff, but I haven't got an iriver to test it on
07:01:03LinusNmy guess is that (s2 - s1) depends on the 2's complement "wrap"
07:01:19stevenmLinusN, true, in some cases it does
07:01:25stevenmholy crap you are right
07:01:41stevenmYou don;t know which is bigger- it varies
07:01:48stevenmand we just learned 2's complement too
07:02:26LinusNif you really need them to be int's, AND with 0xffff
07:02:44stevenmI was told int would be faster
07:02:54LinusNit's often so, yes
07:03:18LinusNin many cases, you'd have to benchmark, or even read the assembly output from the compiler
07:03:32stevenmso I should AND the result with 0xFFF ?
07:03:34stevenmer FFFF
07:05:04stevenmLinusN, if you are not busy, do you think I can give you a file that you can test tell me if it works or not?
07:06:09LinusNunfortunately i am busy
07:06:38stevenmAll right. I will ask someone tomorrow then
07:07:14stevenmI basically spent the last 3 days working non-stop on this. Did the short* thing this morning.. never even imagined endianness would sneak up on me like that
07:07:45LinusNyou don't have an iriver?
07:07:53stevenmno :(
07:08:08stevenmI keep bugging others to test code for me. It must be irritating
07:08:21LinusNplanning on buying one?
07:08:32stevenmAn H3X0 when the support is there
07:08:48stevenmI have an ajbrec that I hardly even use
07:09:11stevenmbut if I were to get one, it would be for development and stuff
07:09:20stevenmYou know where I can get an old h1x0 for this ?
07:09:49LinusNthey sometimes appear on ebay
07:10:16LinusNjeff on misticriver has a few too
07:11:18stevenmhow much do they run for ?
07:11:30LinusNroughly $300
07:12:10LinusNthey are attractive
07:12:20stevenmI was going to by my friend's H3x0 at one point
07:12:27LinusNpartly because of rockbox, i believe
07:12:35stevenmoh wow
07:12:42stevenmthat's pretty cool
07:12:49LinusNsort of, yeah
07:12:54stevenmI was looking for an iriver 'emulator' to test this code on, out of sheer desperation
07:12:59stevenmand all I got were rockboy hits
07:13:10LinusNthere is none
07:13:48stevenmI could really use one right about now :)
07:13:56LinusNmy advice to you is to do the optimizations in steps, and commit to cs
07:14:08stevenmI see
07:14:10LinusNthen we can keep track of the changes
07:14:23LinusNand other people can test by checking out from cvs
07:14:25stevenmI was going to do that today after testing it
07:14:34stevenmonly we found out it was dead..
07:14:46stevenmhey.. I guess I can get previous versions out of CVS too
07:14:50stevenmit - the code
07:14:55stevenmit wasn't playing right
07:14:56LinusNyes, cvs has all versions
07:15:19 Quit hile ("leaving")
07:17:13stevenmhopefully tomorrow we can get the latest version tested and if it works, then I guess commit it
07:17:20stevenmand if not.. check out a working version and start again
07:38:09***Saving seen data "./dancer.seen"
07:40:10 Part LinusN
07:43:06 Quit stevenm ("Leaving")
07:49:03 Join Lynx0 [0] (HydraIRC@
07:50:02 Join Tang [0] (
07:50:12Tanghello :)
07:50:24 Join LinusN [0] (
07:50:32Tangwon't stay a while
07:51:21LinusNhi Tang
07:51:27Tanghi Linus
07:51:32TangHow are you :)
07:52:00LinusNfine i guess
07:52:12LinusNca va?
07:52:21TangCa va yes for me
07:52:25TangI'm in paris
07:52:41Tangto see family and friends
07:52:57LinusN(no, paris) *g*
07:53:01 Quit Lynx_ (Read error: 60 (Operation timed out))
07:53:01 Nick Lynx0 is now known as Lynx_ (HydraIRC@
07:53:14Tangno âris?
07:53:19Tangno paris?
07:53:24Tangdidn't understood
07:53:29LinusNbad joke
07:53:37Tangah lol
07:53:40Tangmaybe not bas
07:53:44LinusNnice is a town on cote d'azur
07:53:53Tanghard to understood for me
07:54:12TangAh okay
07:54:54TangAbout rbx, nice shot for the remote LCD
07:55:08LinusNyeah, that went well
07:55:27Tangit avoid we let the iHP ON under rockbox without knowing
07:55:57Tang(for those who use only remote)
07:56:18LinusNunfortunately, i have to release a new bootloader to fix that issue
07:56:29Tangthyat issue?
07:56:36Tangabout ogg?
07:57:03LinusNno, starting rockbox with the remote
07:57:21LinusNor did i misunderstand something
07:57:29Tangah i see whith short push of remote play button
07:57:54Tangi wasn't clear
07:58:30 Join Seed [0] (
07:58:35Tangi've seen the magasine scan with your interveiw
07:59:00LinusNdidn't know someone scanned it
07:59:01Tangi didn't understood texte of course
07:59:14Tanghum hard to remember who did
07:59:27Tangmaybe badger
07:59:47 Join gromit` [0] (~gromit`
08:00:16amiconn(morning, btx)
08:00:23LinusNamiconn: aha
08:00:25Tangin fact i didn't understood you were three swedish here
08:00:28Tanghi amiconn
08:01:39LinusNlousy picture
08:02:32Tanglol indeed
08:03:28Tanghave you talked with BobTHC yesterday?
08:03:37LinusNabout what?
08:04:12Tanghe suggested to me to create some wikipage in french
08:04:19Tangbut he had to ask you before
08:04:54Tangi guess he didn't had time yesterday
08:04:57LinusNwe have a few french pages already
08:05:16Tangi wasn't knowing
08:06:26Tangso no problems to make french version of the fornoobs page?
08:07:34Tang*anyway it's a bit early
08:07:41LinusNwe are not all that interested in wiki translations, mainly because the information easily gets unsynched
08:08:04Tangah okay
08:08:56LinusNstill, if you are prepared to update both the french and english versions, go ahead
08:09:13LinusNhere's one of the few french wiki pages btw:
08:09:29LinusNypou can easily see how old it is
08:09:34Tangi see
08:09:35LinusNonly 6 questions
08:09:39Tangindeed it make sense
08:09:54Tangthe two version have to be synced
08:10:26LinusNare french people still that bad at english?
08:10:36LinusNi thought that was a generation issue
08:10:43Tangmore even
08:11:00Tanglol gtuess no
08:11:29LinusNi once met an english teacher from tahiti
08:11:40LinusNi could barely understand her
08:11:48Tanglinus i'velol
08:12:06Tanglinus i've to go
08:12:12LinusN"she spook like zis"
08:12:18Tangmy zis?
08:12:22Tangah (this
08:12:49 Quit Seed (
08:12:49 Quit Tang (
08:12:49 Quit LinusN (
08:12:49 Quit pabs (
08:12:49 Quit _Lucretia_ (
08:13:16NJoinSeed [0] (
08:13:16NJoinLinusN [0] (
08:13:16NJoinTang [0] (
08:13:16NJoin_Lucretia_ [0] (
08:13:16NJoinpabs [0] (
08:13:30LinusNwhat a ride!
08:13:53Tang ;D
08:14:09TangLinus i go my friend come back
08:14:18LinusNhave a nice day in paris
08:14:37Tangthabks Linus
08:15:00Tangwill see when i come back at home for french fornoobs version :)
08:16:27 Quit ac_away (
08:16:27 Quit Tang (
08:16:27 Quit LinusN (
08:16:27 Quit pabs (
08:16:27 Quit Seed (
08:16:27 Quit _Lucretia_ (
08:16:27 Quit lostlogic (
08:16:27 Quit micoo (
08:16:27 Quit Nibbler (
08:16:27 Quit QT (
08:16:27 Quit courtc (
08:16:27 Quit edx (
08:16:27 Quit nozomiyume (
08:16:27 Quit HCl (
08:16:27 Quit Rob- (
08:16:27 Quit dwihno (
08:16:27 Quit ze (
08:16:27 Quit Rick (
08:16:27 Quit tvelocity[away] (
08:16:27 Quit amiconn (
08:16:27 Quit Ka_ (
08:16:27 Quit Strath (
08:16:27 Quit webmind_ (
08:16:27 Quit Hadaka (
08:16:27 Quit crwl (
08:16:27 Quit izzy (
08:16:27 Quit crash_ (
08:16:27 Quit Stryke` (
08:16:27 Quit elinenbe (
08:16:27 Quit t0mas (
08:16:27 Quit ferenczy (
08:16:27 Quit [zmaj] (
08:16:27 Quit mbr (
08:16:27 Quit ncrawla (
08:16:51NJoinpabs [0] (
08:16:51NJoin_Lucretia_ [0] (
08:16:51NJoinLinusN [0] (
08:16:51NJoinSeed [0] (
08:16:51NJointvelocity[away] [0] (
08:16:51NJoinedx [0] (
08:16:51NJoinamiconn [0] (
08:16:51NJoinKa_ [0] (
08:16:51NJoinlostlogic [0] (
08:16:51NJoinmicoo [0] (
08:16:51NJoinStrath [0] (
08:16:51NJoinac_away [0] (~austrianc@
08:16:51NJoinStryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
08:16:51NJoinnozomiyume [0] (
08:16:51NJoinHCl [0] (
08:16:51NJoinelinenbe [0] (~elinenbe_@
08:16:51NJoinNibbler [0] (
08:16:51NJoinRob- [0] (
08:16:51NJoinQT [0] (as@area51.users.madwifi)
08:16:51NJoincourtc [0] (
08:16:51NJoindwihno [0] (~dw@
08:16:51NJoinze [0] (
08:16:51NJoinRick [0] (rick@Rick.user)
08:16:51NJoinwebmind_ [0] (~webmind@
08:16:51NJoin[zmaj] [0] (
08:16:51NJoinmbr [0] (
08:16:51NJoinncrawla [0] (
08:16:51NJoincrash_ [0] (
08:16:51NJoinizzy [0] (
08:16:51NJoincrwl [0] (
08:16:51NJoinferenczy [0] (
08:16:51NJoinHadaka [0] (
08:16:51NJoint0mas [0] (
08:17:50***Server message 477: 'logbot #rockbox :[freenode-info] why register and identify? your IRC nick is how people know you.'
08:18:58 Quit ac_away (
08:18:58 Quit LinusN (
08:18:58 Quit pabs (
08:18:58 Quit Seed (
08:18:58 Quit _Lucretia_ (
08:18:58 Quit lostlogic (
08:18:58 Quit Nibbler (
08:18:58 Quit micoo (
08:18:58 Quit QT (
08:18:58 Quit courtc (
08:18:58 Quit dwihno (
08:18:58 Quit Rob- (
08:18:58 Quit HCl (
08:18:58 Quit nozomiyume (
08:18:58 Quit Rick (
08:18:58 Quit ze (
08:18:58 Quit edx (
08:18:58 Quit webmind_ (
08:18:58 Quit amiconn (
08:18:58 Quit tvelocity[away] (
08:18:58 Quit Strath (
08:18:58 Quit Ka_ (
08:18:58 Quit izzy (
08:18:58 Quit Hadaka (
08:18:58 Quit crwl (
08:18:58 Quit crash_ (
08:18:58 Quit t0mas (
08:18:58 Quit mbr (
08:18:58 Quit [zmaj] (
08:18:58 Quit elinenbe (
08:18:58 Quit Stryke` (
08:18:58 Quit ncrawla (
08:18:58 Quit ferenczy (
08:20:05NJoinpabs [0] (
08:20:05NJoin_Lucretia_ [0] (
08:20:05NJoinLinusN [0] (
08:20:05NJoinSeed [0] (
08:20:05NJointvelocity[away] [0] (
08:20:05NJoinedx [0] (
08:20:05NJoinamiconn [0] (
08:20:05NJoinKa_ [0] (
08:20:05NJoinlostlogic [0] (
08:20:05NJoinmicoo [0] (
08:20:05NJoinStrath [0] (
08:20:05NJoinac_away [0] (~austrianc@
08:20:05NJoinStryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
08:20:05NJoinnozomiyume [0] (
08:20:05NJoinHCl [0] (
08:20:05NJoinelinenbe [0] (~elinenbe_@
08:20:05NJoinNibbler [0] (
08:20:05NJoinRob- [0] (
08:20:05NJoinQT [0] (as@area51.users.madwifi)
08:20:05NJoincourtc [0] (
08:20:05NJoindwihno [0] (~dw@
08:20:05NJoinze [0] (
08:20:05NJoinRick [0] (rick@Rick.user)
08:20:05NJoinwebmind_ [0] (~webmind@
08:20:05NJoin[zmaj] [0] (
08:20:05NJoinmbr [0] (
08:20:05NJoinncrawla [0] (
08:20:05NJoincrash_ [0] (
08:20:05NJoinizzy [0] (
08:20:05NJoincrwl [0] (
08:20:05NJoinferenczy [0] (
08:20:05NJoinHadaka [0] (
08:20:05NJoint0mas [0] (
08:20:19 Quit Bagder (Killed by (Nick collision))
08:20:19 Quit silencer_ (Killed by (Nick collision))
08:20:19 Quit CoCoLUS (Killed by (Nick collision))
08:20:19***Nickname collision
08:20:19***Saving seen data "./dancer.seen"
08:20:19***Started Dancer V4.16
08:20:19***Connected to on port 6667
08:20:19***Logfile for #rockbox started
08:20:20***Server message 501: 'logbot :Unknown MODE flag'
08:20:20Mode"logbot :+i" by logbot
08:20:21 Join logbot [242] (
08:20:21 Join Lynx_ [0] (HydraIRC@
08:20:21 Join gromit` [0] (~gromit`
08:20:21 Join pabs [0] (
08:20:21 Join _Lucretia_ [0] (
08:20:21 Join LinusN [0] (
08:20:21 Join Seed [0] (
08:20:21 Join tvelocity[away] [0] (
08:20:21 Join edx [0] (
08:20:21 Join amiconn [0] (
08:20:21 Join Ka_ [0] (
08:20:21 Join lostlogic [0] (
08:20:21 Join micoo [0] (
08:20:21 Join Strath [0] (
08:20:21 Join ac_away [0] (~austrianc@
08:20:21 Join Stryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
08:20:21 Join nozomiyume [0] (
08:20:21 Join HCl [0] (
08:20:21 Join elinenbe [0] (~elinenbe_@
08:20:21 Join Nibbler [0] (
08:20:21 Join Rob- [0] (
08:20:21 Join QT [0] (as@area51.users.madwifi)
08:20:21 Join dwihno [0] (~dw@
08:20:21 Join ze [0] (
08:20:21 Join Rick [0] (rick@Rick.user)
08:20:21 Join t0mas [0] (
08:20:21 Join Hadaka [0] (
08:20:21 Join ferenczy [0] (
08:20:21 Join crwl [0] (
08:20:21 Join izzy [0] (
08:20:21 Join crash_ [0] (
08:20:21 Join ncrawla [0] (
08:20:21 Join mbr [0] (
08:20:21 Join [zmaj] [0] (
08:20:21 Join webmind_ [0] (~webmind@
08:20:21 Join courtc [0] (
08:20:30 Join CoCoLUS [0] (
08:20:31 Join Bagder [0] (
08:20:42 Quit CoCoLUS (
08:20:42 Quit Bagder (
08:20:42 Quit Lynx_ (
08:20:42 Quit gromit` (
08:20:42t0masarg.. freenode :P
08:20:42*t0mas has his birthday today :D
08:20:45 Join silencer [0] (
08:20:47t0masghehe :(
08:20:48t0maslogbot missed something
08:20:49*t0mas has his birthday today :D
08:20:58LinusNt0mas: congrats!
08:21:32 Quit ac_away (
08:21:32 Quit izzy (
08:21:32 Quit Hadaka (
08:21:32 Quit crwl (
08:21:32 Quit crash_ (
08:21:32 Quit silencer (
08:21:32 Quit Stryke` (
08:21:32 Quit elinenbe (
08:21:32 Quit t0mas (
08:21:32 Quit mbr (
08:21:32 Quit [zmaj] (
08:21:32 Quit ncrawla (
08:21:32 Quit ferenczy (
08:21:38NJoinBagder [0] (
08:21:38NJoinCoCoLUS [0] (
08:21:38NJoingromit` [0] (~gromit`
08:21:38NJoinLynx_ [0] (HydraIRC@
08:21:58NJoinsilencer [0] (
08:21:58NJoinac_away [0] (~austrianc@
08:21:58NJoinStryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
08:21:58NJoinelinenbe [0] (~elinenbe_@
08:21:58NJoint0mas [0] (
08:21:58NJoinHadaka [0] (
08:21:58NJoinferenczy [0] (
08:21:58NJoincrwl [0] (
08:21:58NJoinizzy [0] (
08:21:58NJoincrash_ [0] (
08:21:58NJoinncrawla [0] (
08:21:58NJoinmbr [0] (
08:21:58NJoin[zmaj] [0] (
08:22:19 Quit ac_away (
08:22:19 Quit crash_ (
08:22:19 Quit izzy (
08:22:19 Quit crwl (
08:22:19 Quit Hadaka (
08:22:19 Quit [zmaj] (
08:22:19 Quit mbr (
08:22:19 Quit ferenczy (
08:22:19 Quit t0mas (
08:22:19 Quit elinenbe (
08:22:19 Quit Stryke` (
08:22:19 Quit silencer (
08:22:19 Quit ncrawla (
08:23:49 Join Harpy [0] (
08:24:42NJoinHadaka [0] (
08:24:42NJoincrwl [0] (
08:24:42NJoinizzy [0] (
08:24:42NJoincrash_ [0] (
08:25:15NJoinac_away [0] (~austrianc@
08:25:19 Join Tomas [0] (
08:25:20Tomasnice :D
08:25:29*Tomas just got a new nick
08:25:42Tomasit's registered
08:25:54 Nick Tomas is now known as t0mas|split (
08:26:25NJoinsilencer [0] (
08:26:25NJoinStryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
08:26:25NJoinelinenbe [0] (~elinenbe_@
08:26:25NJoint0mas [0] (
08:26:25NJoinferenczy [0] (
08:26:25NJoin[zmaj] [0] (
08:26:25NJoinmbr [0] (
08:26:25NJoinncrawla [0] (
08:26:41t0mas|splitI'm back :P
08:29:53 Quit t0mas|split ("ok, [08:28:10] <lilo> should be resolved")
08:36:32 Join Zagor [0] (
08:37:22 Join rasher [0] (
08:37:48rashercan't say I'm sad I missed that
08:41:56t0maswhat? everybody quitting a few times? :)
08:42:33amiconnmorning Zagor
08:43:16amiconnI've updated my fat.patch. Most of the old style is back, and it always reuses deleted entries now
08:43:43amiconnThe fat tests run fine
08:47:06amiconnSame place:
08:56:32amiconnArgh, this one has a bug :(
09:05:34 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
09:15:58 Join bobTHC [0] (
09:16:29bobTHCmornin' folks !
09:24:01Zagoramiconn: why do you clear the first sector of a new cluster separately from the rest?
09:33:05 Join B4gder [0] (
09:33:56B4gderthe web client is neat at times
09:36:05Zagorthe what?
09:37:46B4gderthe irc client on the web
09:39:52*Zagor performs server update. possible temporary service interruption.
09:40:46B4gdercoffee break time
09:42:46 Quit rasher ("CGI:IRC (EOF)")
09:42:46 Quit B4gder ("CGI:IRC (EOF)")
09:44:06 Join ashridah [0] (
09:59:15LinusNi can't believe all this memory copying in libmad...
09:59:43Ricklibmad is the mod player right?
09:59:46Rickor is that dumb?
10:04:01amiconnZagor: I clear it separately from the because the error condition is different
10:04:29Zagoryes, but why? a disk is not likely two have an odd cluster free
10:04:31amiconnIf writing the first cleared sector fails, I assume disk full
10:04:33Zagoroff sector, sorry
10:05:06amiconnIf writing the rest fails, the only possibility is that something else failed
10:05:20amiconn...because, as you said, space is reserved cluster-wise
10:05:26Zagorfair enough
10:05:34amiconnAnyway, my new version has a severe bug
10:05:51Zagorok i'll stop looking :-) that it doesn't scan for duplicate shortname all the way to the end
10:06:28amiconnI'll restrucure again in the evening (shouldn't do such things in a hurry)
10:06:45Zagorright, I saw that in a modified test case yday but didn't have time to follow up
10:09:28 Join webguest09 [0] (
10:17:40 Join webguest85 [0] (
10:19:56 Quit webguest09 ("CGI:IRC (EOF)")
10:20:21***Saving seen data "./dancer.seen"
10:32:54 Join cYmen [0] (
10:34:16 Join B4gder [0] (
10:43:03 Join preglow [0] (
10:44:52LinusNhi preglow
10:50:45preglowi wonder what i did to suddenly deserve all this good weather
10:51:05Rickyou joined #rockbox, no? :p
10:52:04LinusNgood weather is rockbox' worst enemy
10:52:22preglowfour days of consecutive good weather is not something you see often in trondheim
10:52:37LinusNsounds like a wonderful place :-)
10:53:54bobTHCOn Mon, 18 Apr 2005 17:09:54 -0700 Tanguy SIMON <> wrote:
10:53:57bobTHC>Voila je t'envoie le post de mise au point que je
10:53:58preglowweatherwise, it's my worst nightmare. not long until i move, fortunately
10:53:59bobTHC>comptais envoyer sur le topic iHp concernant la
10:54:00bobTHC>polémique du risque lié à Rockbox...
10:54:03bobTHC>J'ai pu discuter par voie privé sur IRC avec un membre
10:54:05bobTHC>de Rockbox qui souhaite garder l'anonymat.
10:54:06bobTHC>Je voulais avoir son avis par rapport à la polémique
10:54:07bobTHC>de ces derniers jours sur le port Rockbox iHP...
10:54:12*preglow claps
10:54:13bobTHC>D'un commun accord je vous livre l'essentiel de sa
10:54:23preglowindeed :P
10:54:34LinusNbobTHC: no, it's french :-)
10:54:39RickWhat does that say?
10:54:43bobTHCbig fuck ! and big sooory
10:54:45micoogm guys :p
10:55:03B4gderit looked Rockbox related at least ;-)
10:55:07Lynx_micoo: genetically modified guys? ;)
10:55:09bobTHCfor sure
10:55:41LinusNan anonymous guy who wants to know about the rockbox porting progress?
10:55:56bobTHCit's a paste of a french forum
10:56:27LinusNbobTHC: url?
10:57:22RickWhat a long URL
10:57:33bobTHCthe polemic is " flash or not flash"
10:57:49preglowcan't french people read the wiki?
10:58:01Rickpreglow: Do they know english?
10:58:17preglowi guess some of them do :P
10:58:26bobTHCthey have a lot of problem with english, particulary with "techie english"
10:58:52bobTHCand they want a clear infrench explaination
10:59:13preglowmust suck badly to not have a good grasp of english
11:00:44bobTHCi dont think it's only a french problem :)
11:01:46micoois it so hard to use babylon..
11:02:17LinusNmicoo: you mean babelfish?
11:02:20bobTHCit's worst to understand what he say ;)
11:02:33preglowhell no, just saying it's got to suck, i'm pretty dependent on english
11:02:45micoowho's he?
11:02:57 Join rasher [0] (
11:03:11bobTHCbabelfish (sorry it's a "it") ;)
11:03:37micookinda linusn
11:04:51micooit's a translator
11:05:08micoobut only for words
11:05:24micoosingle words..
11:08:13bobTHCfor single word i prefer my oldfashion robert & collins ;)
11:10:40LinusNlogbot: "les premiers lecteurs Archos, dont le firmware était réputé pour son pathétisme"
11:10:45LinusNLOL: "les premiers lecteurs Archos, dont le firmware était réputé pour son pathétisme"
11:12:11bobTHChehe u have the same autocomplete nick feature bug, i'm not alone ;)
11:13:01rasherwow, babelfish ate that nicely
11:15:19rasher"the first Archos readers, whose firmware was famous for its pathetic"
11:16:09rasherthat's pretty good (I think?)
11:18:44LinusNi really like this:">
11:19:20LinusNthe random quotes are really useless, but funny
11:19:59preglowi like dwihnos quote
11:20:52 Join amiconn_ [0] (
11:21:11rasheryes, that's a great quote
11:21:11 Quit webguest85 ("CGI:IRC (EOF)")
11:21:40rasher"last seen" is pretty nice
11:22:03rasherbut of course doesn't work for previous years
11:22:12rashernot quite sure what happens for those
11:22:35rasherI guess it's with regards to the last day of the period
11:23:15RickDon't you love how 'rockbox' is treated as a nickname?
11:24:12rasherthe only real nick is amiconn :-O
11:24:16rasherin that table
11:25:46RickI'm guessing someone has joined here under that nick
11:25:50Rickwhich is why it picks it up as one
11:26:23rashersounds likely
11:28:55preglowmidnight2k3 is the guy who said the most and also the one who was kicked the most? :P
11:29:43rasherLinusN said most words.. midnight2k3 just talked to himself
11:31:20 Nick ac_away is now known as austriancoder (~austrianc@
11:31:23austriancoderhi all
11:31:47austriancodershould the backlight of the lcd go on, if the hold button gets off?
11:32:06LinusNi would like that
11:32:48Rickyou mean *off*
11:32:51*Rick misread
11:33:05austriancoderok i will do it so
11:33:24Rickyes, it's a good idea
11:33:26Rickiriver firmware does it
11:37:17 Quit amiconn (Read error: 110 (Connection timed out))
11:37:17 Nick amiconn_ is now known as amiconn (
11:39:32*LinusN added a link to the irc stats on
11:39:49rasherguess I'll have to keep it updated now then :)
11:40:13LinusNyup :-)
11:40:50austriancoderLinusN: where is the place to add icon stuff for hold?
11:42:10austriancoderi mean, where should i add the check if hold is on -> draw lock-icon
11:42:23austriancoderstatus.c or wps.c
11:47:00austriancodericon works ;)
11:47:26austriancoderbut atm only at the main lcd..
11:47:35LinusNof course
11:47:53 Join MoosCamaro [0] (
11:48:08MoosCamarohi all
11:48:43MoosCamarohi austriancoder
11:50:50preglowLinusN: the instruction decoder can fetch 32 bits from the iram in one cycles, yes?
11:51:45preglowif it required two 16 bit instructions, it would just fetch the same 32 bits two times and extract different bits, or?
11:52:34preglowthe imdct_l innerloop will require 16 bit address offsets for all mac instructions, but don't think i can avoid it anyway
11:52:51preglowthen i remember we don't use iram for code, bah
11:53:49LinusNpreglow: i have a suggestion for you regarding mpa2wav.c
11:53:58preglowgo ahead
11:54:47LinusNmad_fixed_t mad_frame_overlap[2][32][18]; ATTR_IDATA
11:55:00LinusNunsigned char mad_main_data[MAD_BUFFER_MDLEN] ATTR_IDATA;
11:55:10LinusNdeclare these globally
11:55:48LinusNthen after the mad_timer_reset() call:
11:55:58LinusN /* We do this so libmad doesn't try to call codec_calloc() */
11:55:58LinusN memset(mad_frame_overlap, 0, sizeof(mad_frame_overlap));
11:56:04LinusN frame.overlap = &mad_frame_overlap;
11:56:04LinusN stream.main_data = &mad_main_data;
11:56:35LinusNthat's the only two things that are malloc()ed by libmad
11:56:39preglowhok, i'll do it now
11:57:13preglowi somewhat doubt there's room enough for it in iram, though
11:57:14LinusNin mpa2wav.c, it's Stream and Frame, though
11:58:27LinusNshould be about 6-8kbytes
11:59:04 Quit Nibbler (Read error: 113 (No route to host))
11:59:07preglowi'll give the result a spin
11:59:51LinusNshould give *some* boost at least
12:00:55 Join Nibbler [0] (
12:00:59preglowa handsome boost
12:01:09preglowit's up to 220% realtime on iriver.mp3
12:01:28LinusNwhat was it before?
12:01:31preglowthis is without wav writing or dithering
12:01:40rasherand that's 320kbps cbr..
12:01:44rasherwhich cpu sped?
12:01:48preglow120mhz, of course
12:02:03preglowit's not THAT fast yet, heh
12:02:18preglowthe realtime counter increases all the time
12:02:22rasher:) wasn't sure.. hadn't been paying too much attention
12:02:22preglowi don't get why it does that for cbr files
12:02:32preglowthe only variable should be the huffman decoder
12:02:42preglowit's up to 90% realtime in 42 mhz
12:02:44preglowalmost 90
12:02:58Zagorhow much difference is there between say 128kbit and 320?
12:03:07preglowi'll check a 112 kbps file
12:03:11preglowi haven't got a 128kbps file handy
12:03:36preglowstarts out at 300% realtime
12:03:56preglowthen dithers around 297%
12:04:28bobTHCmiss nothing to be able to decode mp3 without using cpu_boost(true) ?
12:04:57LinusNthere is a lot of memory copying going on when decoding mp3
12:05:30preglow128 kbps probably is well above realtime at 42 mhz
12:05:42preglow48 :)
12:06:07bobTHCgood news for power consuption
12:06:16preglowgetting to this point with all the codecs is going to take a lot of time at the current rate...
12:07:07 Join sox [0] (
12:07:26bobTHC:/ we need more bitwizard
12:07:42soxhoy LinusN and preglow, for chrissake, commit that durn code and make us rock our boxes...!
12:07:50preglowsox: most of it IS commited
12:08:28preglowi'll have to stop this chattering and finish imdct_l before my mind forgets its asm mode
12:08:31preglowbut first, food
12:10:14soxpreglow: ok i understand, so it's still the api that's not ready to get the data flowing?
12:10:38soxotherwise we could all be listening to stuttering mp3s i guess ;-)
12:11:45LinusNi am working on this
12:12:41soxallright! good news are raining down on us
12:13:27sox"a stuttering rockbox-played mp3 is better than a working iriver-played mp3"
12:15:13rasherI wouldn't go that far.. but it induces more hope :)
12:15:49soxand more stress on you brave souls, im afraid
12:16:02preglowsend beer
12:16:10soxi will, what brand?
12:16:55preglowsomething english
12:17:09LinusNa nice ale would be nice
12:17:19preglowoh, indeed
12:17:28soxok, so this explains where the donations end up...
12:17:46soxswedish systembolaget...
12:18:27preglowi've been a steady contributor to norwegian vinmonopolet after i developed a taste for beer that can't be gotten in stores as well
12:19:09*preglow laments his wallet
12:19:11soxoff topic: poor norwegian, i hear prices on beer and wine over there are insane
12:19:21preglowoh, quite
12:19:47preglowmost things are pricey here
12:20:17preglowwe pilfer the swedes for cheap goods every now and then
12:20:23***Saving seen data "./dancer.seen"
12:20:29preglowearning ourselves a bad reputation in the process ;)
12:21:29soxah, you are so welcome :-)
12:22:06rasherthe swedes?!
12:22:20rasherthe same swedes who come to denmark to buy cheap alcohol?
12:23:00soxthe global alcohol market , makes people travel
12:23:49preglowoh, don't worry, we pilfer danes as well, but it's too far away for regular plundering
12:23:59preglowand you're so hard to understand :P
12:26:40rasherare we worse than the swedes?
12:26:44rasherapparently so
12:26:47preglowoh yes
12:26:51preglowno trouble understanding swedes
12:26:55preglowunderstanding danes = impossible
12:27:09preglowand i do mean speaking your native language here
12:27:10rasherah, wasn't quite sure how those stacked up, but I'd have guessed that as well
12:27:37preglowbut last slice of bread gone, i'll hack away at some assembler again
12:27:48rashergo go go!
12:39:44LinusNpreglow: do you have time to try a patch for me?
12:40:16preglowLinusN: sure
12:41:49LinusNthis is the IDATA_ATTR fix we tried earlier, plus i removed the memcpy() for the mp3 indata
12:43:19LinusNshould give us more real-life performance values
12:43:52*LinusN forgot his earphones, so he can't listen to the results
12:46:10preglowit peeks at 230% realtime now
12:46:16preglowi'll enable the disk writing again, heh
12:46:59preglownice! my sound playback skips when i connect the h120
12:47:26LinusNon the pc?
12:48:06preglowthe scheduler in the kernel i use seem to be less than optimal
12:48:14preglowit also skips when i do a heavy grep
12:48:23rasherI suggest you get a real OS.
12:48:33LinusNrasher: like?
12:48:39rasherI don't know :)
12:49:03preglowLinusN: sounds great
12:49:07rasher(long time joke for me :).. whenever something's wrong, the correct response is: "get a real os")
12:49:12LinusNpreglow: good
12:49:22preglowi already run a proper os
12:49:31LinusNpreglow: i'll commit it then
12:49:35preglowdo so
12:49:43preglowcut away a great portion of mpa2wav, i see
12:50:18LinusNsilly stuff
12:50:36austriancoderi think a HAVE_HOLD_BUTTON is a good thing.... but he
12:50:39preglowLinusN: remove the unused variables before you commit
12:50:42preglowwarnings are bad
12:51:11LinusNsilly me
12:52:01austriancodergone to eat something
12:52:25 Join xen` [0] (~xen@
12:53:11LinusNaustriancoder: looks ok to me
12:53:19LinusNlunch time for me too
12:55:36B4gderthe build times really vary a lot on the server
12:58:06B4gderI thought about making the build script wait for 3 minutes of non-commits before it proceeds to build
12:58:19B4gderbut I haven't bothered
12:58:49preglownot a bad idea
12:59:13B4gdersince now, it might do the build in the middle of a commit series
13:00:35rasherthat'd be amusing
13:00:54rasherin the same way that blowing up toy cars is amusing
13:02:44B4gderbut at least, if this would happen, it would build again immediately afterwards
13:06:16HCllonghorn is gonna be crap... with its tcpa restricted stuff..
13:06:31HClheh @ pic :p
13:06:50preglowlonghorn is going to have palladium stuff?
13:06:55HClas far as i know, yea.
13:07:20preglowi'm getting quite comfortable with linux anyway
13:07:24HClit'll prolly be the day that i quit accepting updates from microsoft for winxp, and switch to linux.
13:07:43HCli tolerated windows so far because its not restricting me too much yet.
13:08:40 Join Aison [0] (
13:10:41HCltime to go to class
13:10:57*B4gder stuffs his kernel with printk
13:13:06micoo(14:08:29) (HCl) it'll prolly be the day that i quit accepting updates from microsoft for winxp, and switch to linux.
13:13:06micoo(14:08:49) (HCl) i tolerated windows so far because its not restricting me too much yet.
13:13:13micoosame case with me
13:13:25micooand i need to go either
13:14:33austriancoderi want to add me to the credits, but i dont know which file i must edit
13:14:46rasheryou're already added iirc
13:15:28B4gderdocs/CREDITS it is anyway
13:15:34austriancoderah :) fine thanks
13:15:47rasheror maybe not
13:16:17austriancoderhmm.. seems not
13:16:28austriancoderbut i will add me.. with the hold-button stuff
13:16:36B4gderof course!
13:16:50B4gdereveryone who contributes are welcome there
13:18:21austriancodernext thing for me, will be to finish the remote lcd stuff
13:19:00austriancoderor make remote buttons work
13:19:02austriancoderlets see
13:30:11austriancoderLinusN: is this commit ok for you?
13:32:10LinusNwhy int data = 0;?
13:32:47austriancoder+g+ dont know..
13:32:53austriancoderwill change it
13:33:24austriancoderthe rest is ok for you?
13:33:58austriancoderthen i will commit it now
13:34:42austriancoderhave a lot of fun ;)
13:40:15austriancoderthere are in the driver of the main lcd functions like: lcd_getstringsize
13:40:51austriancoderi could now add them to the remote driver, but i think its not so good, to have same functions in more then one file
13:41:10austriancoderor shouldn't i care?
13:41:15amiconnYes... that's a function which should be moved to font.c. Linus?
13:41:32amiconnIt's not really lcd related
13:41:43LinusNi agree
13:43:19austriancoderwhats with the new graphics api...
13:43:32austriancoderwhats the progress of it?
13:43:35amiconnBack then I copied that for the grayscale lib as well... I should remove it
13:43:51amiconnaustriancoder: Not much... :(
13:44:24austriancoderamiconn: not much time? or is the api not ok?
13:44:37amiconnNot much progress yet
13:45:00amiconnImho this api suggestion is ok... since it's my suggestion :)
13:47:27austriancoderamiconn: i could and would help you
13:49:41austriancoderi only need to know, where to start
13:51:50 Quit Zagor (Remote closed the connection)
13:59:41austriancoderamiconn: so.. what we are doing now with the new api.. shall we wait?
14:03:04amiconnThe new api should get implemented for all graphics 'packages', i.e. lcd-recorder.c, lcd-h100.c, grayscale lib, player gfx lib etc
14:03:17austriancoderi know
14:03:50austriancoderit is a big change in rockbox, but its needed
14:04:14austriancoderi could do the h1xx part
14:06:39austriancoderyou tell what i should do, and i will do my best
14:08:25 Quit lostlogic (
14:08:42NJoinlostlogic [0] (
14:09:06amiconnThe problem is that I have some ideas for optimisation that are very difficult to explain without at least having a code snippet
14:09:49amiconnThen I have multiple other things I'd like to get done as well... like variable blocksize support in the MMC driver
14:10:05amiconnI want to be able to actually use my 2 GB MMC
14:12:12 Quit lostlogic ("Going to the moon")
14:12:39 Join Shagnar [0] (
14:12:44austriancoderok.. then i will work on the remote control stuff for the iriver and will ask you in some weeks, if you have time :)
14:14:12austriancoderamiconn: have you moved the lcd_getstringsize to fonts.c, or can i do it?
14:20:27***Saving seen data "./dancer.seen"
14:23:58amiconnI didn't do that yet. If you move it, you should probably rename it as well. font_getstringsize() sounds good to me.
14:26:35austriancoderi will do the move
14:28:41amiconnThis may require changes in a number of places
14:29:10austriancoderi have seen it
14:29:21austriancoderbut thats not a porblem
14:29:55austriancoderis it ok to have static int curfont = FONT_SYSFIXED; in font.c and lcd-h100.c?
14:31:17amiconnHmm, that variable should only exist once.... perhaps that's why the function is part of the lcd code
14:31:54amiconnxx_getstringsize() of course has to use the current font, which may be different per graphics driver
14:32:50austriancoderso what shall i do now?
14:33:13amiconnOn the other hand, that function isn't the shortest, so duplicating it isn't the best solution either
14:34:39amiconnMy idea would be to put the getstringsize code itself into font.c, with an additional parameter (the font number)
14:35:07austriancoderah, that sounds good
14:35:26amiconnThen each graphics driver could have its own xx_getstringsize() function, which would simply call the shared code in fonts.c with the current font number for that driver
14:35:51 Join F1^Aison [0] (
14:35:53austriancoderint font_getstringsize(const unsigned char *str, int *w, int *h, int fontnumber);
14:36:49amiconnyes, exactly
14:37:05austriancoderok.. will do changes now that it compiles and works
14:37:16amiconnlcd_getstringsize(const unsigned char *str, int *w, int *h)
14:37:37amiconnreturn font_getstringsize(str, w, h, curfont);
14:37:58austriancodernice trick
14:38:17amiconnThen we could do similar in all other graphics drivers
14:38:59amiconnNo code changes needed except in lcd-*.c and font.c
14:40:00preglowi do get the feeling the person who factorized the c imdct to the arm imdct has some years of programming experience on his back
14:41:42 Quit Aison (Read error: 60 (Operation timed out))
14:41:52 Join lolo-laptop [0] (~lostlogic@
14:46:06MoosCamarogood team for graphics api
14:46:17MoosCamarogo!go! guys
14:47:32austriancoderamiconn has no time
14:47:49austriancoderamiconn: getstringsize move successfully done
14:48:18austriancoderon h1xx
14:48:31austriancoderneed to look at other platforms
14:50:58 Quit Shagnar ("( :: NoNameScript 3.81 :: )")
14:54:10austriancoderwill do now some reallife stuff.. see yeh
14:54:15 Nick austriancoder is now known as ac_away (~austrianc@
14:56:10 Join thegeek [0] (
15:01:55 Join DangerousDan [0] (
15:02:29 Nick webmind_ is now known as webmind (~webmind@
15:20:23 Join Shagnar [0] (
15:25:35B4gderthere's a gradient square on my screen
15:26:39*preglow looks forward to testing four hundred lines of asm code he's certain contains bugs
15:27:15LinusNpreglow: happy hacking
15:28:22 Part LinusN
15:30:44 Join einhirn [0] (
15:33:35 Join tedboer [0] (
15:36:34tedboeri will be receiving my iRiver H320 somewhere next week, and I am very interested in running Rockbox on it.
15:36:45tedboerhm next month that it
15:37:17tedboernow, from the wiki i understand currently Rockbox runs on the 1XX series only
15:37:54tedboerso i'd be interested in helping with the 3XX development.
15:38:38bobTHCu're welcome
15:39:47bobTHCbut for the moment the effort is focus on the port on HP1XX serie.
15:40:15tedboeri see. so i better wait?
15:40:30preglowno idea when h3x0 will be supported
15:40:57tedboerare the HP1XX and the H3XX similar in hardware design?
15:41:06preglowyes, very
15:41:29bobTHCu can start to dig about the hardware diff ;)
15:41:44tedboerso to get started i could read the HP1XX docs on the Rockbox website...
15:42:08tedboerare similar docs available for the H3XX ?
15:42:26preglowthey share many of the same components
15:42:33preglowfinding the specs on the lcd is important, i guess
15:42:56tedboerthey use the same processor?
15:43:17B4gderthe lcd is the biggest diff to start with
15:43:43tedboeri see.
15:43:51bobTHCu can collect missing docs for hardware and board scans
15:44:08tedboerwhere did the docs for the HP1XX come from?
15:44:15preglowall around
15:44:19preglowmanufactorers websites, etfc
15:44:34tedboerhas iRiver been cooperating in any way?
15:44:37preglowthe lcd was one of the hard parts, though
15:44:49 Quit ashridah ("Leaving")
15:44:55tedboerno surprise...
15:45:04preglowwell, we haven't asked for any cooperation
15:45:10preglowand they probably wont cooperate anyway
15:45:13preglowthey have no reason to
15:45:17 Join ashridah [0] (
15:45:20bobTHCbut a lot docs are already there :
15:46:15tedboerwell, neuros is providing a lot of open source support for their players. that might be an argument to persuade iriver to do the same...
15:46:36B4gderwell, neuros is desperate
15:46:37tedboerok, thanks for the link. i did not find that reading through the wiki
15:46:49preglowthe h120 isn't even produced anymore
15:46:51 Join Bager [0] (~Bager@
15:46:56preglowand neuros haven't got much to lose
15:46:57bobTHCuse the wiki index
15:47:24preglowiriver wants to sell newer players, not support a player people will have to jump through hoops to buy anyway
15:47:50Bagerfor H3x0
15:48:05Bagerin fact, we need nearly only info for LCD
15:48:34tedboerok, all this sounds rather optimistic.
15:48:51Bagerbut i think that this part is the most difficult ...
15:49:34tedboeri am an experienced c/c++ programmer, with expertise in music software. i don't have much hardware skills though..
15:49:50Bageralso, we don't have much info about the PMU ( )
15:50:20Bagerbut i'm not sure that it's necessary
15:51:02B4gderredboer: the slightly higher level code will be identical on the h3xx as on the h1xx
15:52:01tedboerok. i guess i would be more useful working on the higher level code...
15:52:10bobTHC:) and we need codec optimization
15:52:24preglowyes, that we do
15:52:44*preglow peels sticky asm off of his figners
15:52:45B4gdertedboer: you could get your hands wet by setting up a build environment and compile a h1xx version
15:52:50Bagerpreglow how is ~400 lines asm ;)
15:53:15preglowi'm working up the courage needed to finish it
15:53:36tedboeri use linux exclusively. would that be a problem?
15:53:50B4gdertedboer: on the contrary :-)
15:53:54preglowit's for the better
15:53:59tedboerthought so :-)
15:54:16tedboeryou use gcc?
15:54:43tedboer(oh well, this is FAQ, i should just start downloading :-) )
15:55:19tedboeris there an emulator?
15:55:28B4gderthere's simulator
15:55:34B4gder a
15:55:48B4gderthat allows you to test UI stuff on the PC
15:56:12tedboerok. so i could get started with that as well.
15:56:34tedboerthe docs for setting up a build environment are on the rockbox website?
15:56:36B4gderthat of course uses your native gcc
15:56:51tedboergot it.
15:57:35tedboerthanks. i am really excited about this. i was afraid i'd be stuck with closed and limited software.
15:58:07 Quit thegeek (Read error: 54 (Connection reset by peer))
15:58:47tedboerhey, do i see that the cpu is m68k ?
15:59:09Bagerin fact, it's coldfire v2 core
15:59:23Bagerthere are some missing instructions in it
15:59:28tedboerit has been years, but i've written plenty of m68k assembly on atari st and macintosh
16:01:29Bagertedboer see also
16:02:18 Join Sucka [0] (
16:02:45tedboeri'll set up a build env, subscribe to the mailinglist, and come back here regularly.
16:03:12tedboerthanks for the info.
16:03:48 Join webguest18 [0] (
16:03:52Bageru're welcome ;)
16:06:05Bageralso, the CPU has some DSP additions
16:06:34Bagerfor more info ask preglow
16:08:15Bageri mean, there is info about them in datasheets, but preglow has experience in using them
16:09:23bobTHC3 badgers on the same chan, we are lucky today ;)
16:09:51Bageri'm not BagDer
16:10:16Bagermaybe i should change my nick
16:12:13webguest18has any one made any progres on the audio api
16:14:09preglowwebguest18: yes
16:14:29preglowwebguest18: it'll be posted on the site when it's ready for use
16:14:59Bagerthe DAY
16:15:34webguest18I was wondering how it had been done in the end does it ues two predefined slots or just one and load as many codecs as are nesasery?
16:15:52preglowwebguest18: we don't know, linus is the one fiddling with this at the moment
16:17:47webguest18the reason I ask is I have had a go at it but as I only have a h120 I was wondering how the speaking menues worked. i.e. if you file brouse while playing a track is are the menues said over the top of the music or is a pause introduced while the item is said?
16:18:35 Quit Shagnar (Remote closed the connection)
16:19:42B4gderon the archos, it is either music or speach
16:20:01B4gderbut on iRiver we should be able to do both at once
16:20:28***Saving seen data "./dancer.seen"
16:21:29 Quit ashridah ("Leaving")
16:21:33webguest18so the speach is not layed on top of the music. ok I have a api witch is flaky that can do that however I am working on laying the speatch on top
16:21:44preglowon the iriver, it should be laid on top
16:21:45webguest18at the moment
16:23:47webguest18so do you know if linus has gone with the two block aproch?
16:24:09B4gderho he hasn't
16:24:11preglowon arm, add a, b, c, lsr x means a = b + (c << x), yes?
16:24:32B4gderthere's no loadable codec at all yet in his work, afaik
16:26:50webguest18ok well I have to admit that I have been cheating for codec usage by uing dummy codecks i.e. wav files with .ogg or .mp3 and going that way for the time being
16:29:00 Quit B4gder ("CGI:IRC")
16:31:50 Quit tedboer ("Leaving")
16:33:23webguest18when is linus usaly on line
16:35:16 Join thegeek [0] (
16:35:24bobTHCwebguest18 >">
16:38:40crash_ah Linus and all you guys rock :) i just saw iriver remote lcd pic. sweet!
16:41:35 Quit Sucka ("a bird in the bush is worth two in your house")
16:44:10 Quit sox ("CGI:IRC")
16:46:23 Join Stryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
16:46:36 Quit webguest18 ("CGI:IRC")
16:46:36t0maspreglow? do I have to compile something for every archos platfor separate?
16:46:53preglowi don't do archos stuff
16:47:00t0masI know
16:47:09t0masBut I have to test my patch...
16:47:15preglowand hence i don't compile for them either :P
16:47:26pregloweverything i've done so far has been coldfire specific
16:47:34t0masyou don't test your code for archos?
16:47:40pregloweverything i've done so far has been coldfire specific <-
16:47:51preglowso the archos targets never see my code
16:47:59preglowbut of course
16:48:12preglowif your code is to be used on archos as well, you should test it
16:48:19preglowon at least one of the targets
16:48:27t0masyeah it works on player and recorder
16:48:47t0masbut I didn't test gmini and recorder V2
16:48:51t0masthings like that
16:49:49*t0mas starts writing a script to compile all targets sequentially..
16:53:11t0mashmz... I don't even have a Gmini compiler..
16:53:28 Join stevenm [0] (
16:53:38stevenmHello people
16:53:45amiconnt0mas: Talking about your bmp loader?
16:53:59amiconnIf yes, you certainly don't need to test on the player
16:54:09t0masI know
16:54:09amiconnThe player lcd is charcell based
16:54:14t0masit doesn't have an LCD
16:54:29amiconnIt does have an lcd
16:54:35t0masah ok
16:55:00amiconnHAVE_LCD_CHARCELLS instead
16:55:08t0masand Gmini? should that be tested?
16:55:37amiconnThe gmini does have a graphical lcd, but getting the compiler to work might be very tricky
16:55:55amiconncalmrisc gcc is cutting edge stuff
16:56:30amiconnThe bad thing is that you can compile for archos, but can't actually test
16:56:30t0mashm... it compiles for all recorder versions...
16:56:38t0masand player (but that was expected)
16:56:58t0masyeah, but as long as it doesn't break compiling for them... other people can test it...
16:57:15amiconnIf you compile graphical stuff, it's sufficient to compile for either one of (recorder, fm, v2, ondiosp, ondiofm). They all share the same lcd
16:57:38t0masok, I did recorder dan ondio
16:57:44t0masboth compiled...
16:58:08 Join Zagor [0] (
16:58:13amiconnAs I said, testing one of them is sufficient. Of course trying more than one doesn't hurt
16:58:48t0masok :)
16:59:00t0masthen I declare it working :D
16:59:24amiconnI could test your code in the evening
16:59:35amiconn...on Recorder V1 and Ondio SP
17:00:41stevenmyo rasher, can you possibly run a code test for me?
17:03:08preglowstevenm: yo, use ints whenever you can, i see even moves are slower with shorts and bytes
17:03:42stevenmpreglow, all right
17:03:52 Join Sucka [0] (
17:04:14stevenmpreglow, I am seeing a bit of strangeness in one place where it stops working when you go from short to int
17:04:53stevenmbut right now I need to see if the current code works at all. got rid of all the endian crap, see if it works now
17:10:22*HCl yawns
17:11:54 Quit Nibbler (Remote closed the connection)
17:16:04preglowimdct_l doesn't work
17:16:11preglowsurprisingly enough
17:19:31*preglow moans loudly
17:20:14stevenmpreglow, I hear ya
17:20:43preglowdebugging assembler means reading the code over and over until i have found all the errors...
17:21:29stevenmwell at least you have a device to check your code on
17:26:37 Join Shagnar [0] (
17:27:04Shagnarhello :)
17:27:22stevenmShagnar, hello
17:28:17stevenmShagnar, I think I have a .rock that works now. Wanna try it out ?
17:28:36Shagnarof course. just wanted to ask if you could find the bug alredy :)
17:28:52stevenmyea, it's endian stuff I HOPE. If not, I am rather lost
17:29:20stevenmI rolled a bunch of code back... this is with only iram optimization, nothing else
17:30:05stevenmhere: stevenm/midi2wav.rock">
17:30:29stevenmShagnar, goes into /.rockbox/viewers. Make sure you're at 120Mhz.. expect it to take 4 - 5 minutes.
17:31:45stevenm(with my test file, that is)
17:32:20Shagnarall right
17:33:18stevenmShagnar, please let me know what happens .. i hope that actually works
17:33:36Shagnarjust started :)
17:34:04bobTHCstevenm> WTFin' questionnaire ? : stevenm/IMGP0706.jpg">
17:35:21stevenmbobTHC, yea, that was the university life survey they gave people
17:35:25Rickstevenm: Arn't those kinds of questionnaires illegal? :p
17:35:31preglowargh, and there i ran out of ira,
17:35:33Rickfor jobs anyway
17:35:47stevenmWell.. I don't know.. I should have put transgender
17:35:55stevenmI mean, they're practically asking for it
17:36:07bobTHCit's fuckin' segregationist
17:36:42stevenmmaybe they were hoping we'd put random options so they get more diversity funding
17:37:40Shagnarstevenm took about 3 minutes
17:37:46stevenmShagnar, awesome!
17:37:49stevenmShagnar, does it sound right ?
17:38:49Shagnarsounds great
17:38:57Shagnarfantastic :o)
17:38:59stevenmShagnar, it works? ACTUALLY no noise ?
17:39:02stevenmAWESOME !
17:39:05Shagnarabsolutely clear!
17:39:15stevenmmost. relief. ever.
17:39:17Shagnaryeah :)
17:39:24Shagnargratulations ^^
17:39:28rasherBagder: digest fell over again..
17:39:49stevenmShagnar, do you think I could sneak in a shift optimization and you could test it again ?
17:39:58Shagnarof course :)
17:40:56stevenmHere, lemme do it, rebuild, etc
17:43:45stevenmShagnar, stevenm/midi2wav.rock">
17:44:17stevenmShagnar, there you go. that's only with shifting optimized. Should still work, I hope
17:44:47Shagnarall right. just one moment, tried to midi2wav another midi file ;)
17:45:15stevenmShagnar, should take a bit under 3 min
17:45:48preglowrasher: the thing hasn't learnt to get up on its own yet?
17:46:36stevenmShagnar, all right
17:46:59stevenmpreglow, woah.. eh ?
17:47:30Shagnarstevenm just started
17:49:03amiconnstevenm etc. I have a rather complex midi file. Might be a good test
17:50:22stevenmamiconn, sure, send it over if you wanna
17:50:59 Join Aison [0] (
17:51:31Shagnarstevenm 2:55 (abaout 15 seconds less)
17:51:41stevenmShagnar, sweet. it work?
17:52:13Shagnarall right
17:52:18stevenmShagnar, excellent
17:52:40stevenmI expect good improvement (30 sec or so, like last time) out of the getsample optimization
17:52:41amiconnstevenm: Bleh, can't access my other box atm (samba hanging or similar)
17:52:54stevenmamiconn, that's all right. any time you have it working would be cool
17:53:20stevenmbut that's the trickiest thing.. could cause most breakage. I'm going to commit this working version to CVS, then go from there
17:53:31stevenmShagnar, thanks for the testing
17:53:31Shagnarnow i'll try the midi file which took about 2h ;)
17:53:35Shagnarnp :)
17:53:42preglownot only does imdct_l now produce sounds that's a bit buggy
17:53:50preglowbit it's actually slower than the other ugly imdct_l
17:54:05preglowand i refuse to believe it's actually slower
17:57:45stevenmok there, it's committed
17:57:58stevenm50% realtime, working, at the moment
17:58:09 Quit F1^Aison (Read error: 148 (No route to host))
17:59:19stevenmShagnar, I'll probably see if I can get the endian magic working later.. right now gotta go to class. Will also ask enee professor about it- he used cf at some point
17:59:53Shagnarwould be great :)
18:00:34stevenmonce I code that right, we'll get about 60%. Then, more stuff from there
18:02:35stevenmnow class time.. and resisting the urge to beat motorola in the head with a stick
18:02:36bobTHCdo u think it's possible to aim realtime for this plugin ?
18:02:45stevenmbobTHC, it is possible
18:03:03stevenmwe got 60% yesterday, only all the signs were mangled. I'll try that again tonight
18:03:15stevenmPlus there may be other thing that can be done in ASM
18:03:35Shagnaryeah ASM... most complicated thing ever
18:04:04bobTHCand in the same time the most powerfull :)
18:04:44stevenmwell you can get realtime right now- just cut the sampling rate a lot
18:09:27 Quit Aison (Connection timed out)
18:13:34 Quit stevenm ("Leaving")
18:16:29 Join amiconn_ [0] (
18:17:40 Quit Shagnar ("( :: NoNameScript 3.81 :: )")
18:20:32***Saving seen data "./dancer.seen"
18:21:39 Join muesli- [0] (
18:22:24 Join Shagnar [0] (
18:22:31muesli-hi Shagnar
18:22:37Shagnarhi muesli :D
18:22:58muesli-a new pope-ware was released ;)
18:23:25muesli-about 20mins remaining until official kick off
18:24:00preglowyou may watch me proceeding not to care
18:24:45muesli-neither i do
18:25:01muesli-doenst touch me but its interesting though
18:25:07Shagnar:( still some kind of bug in midi2wav ....
18:25:18 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
18:25:33Shagnarsome instruments suddenly don't "play" anymore
18:25:47Shagnar... well, have to wait 'till stevenm is back again..
18:31:46 Quit amiconn (Read error: 110 (Connection timed out))
18:31:46 Nick amiconn_ is now known as amiconn (
18:44:20 Join Aison [0] (
18:58:14 Quit bobTHC ("Smoke Weed Every Day !")
19:17:00 Quit muesli- (Read error: 113 (No route to host))
19:26:28 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:34:29 Nick Sucka is now known as Sucka`away (
19:41:22 Join muesli- [0] (
19:42:33 Join muesli_ [0] (
19:42:46Shagnaryeah, 2 mueslis at once
19:54:52MoosCamarohey habemus papa
19:55:05MoosCamaroa german papa
19:56:03Shagnardoes anybody know when stevenm wanted to return?
19:56:52MoosCamaroprobably tonight
19:59:49 Quit DangerousDan (Read error: 110 (Connection timed out))
20:02:14rasherpreglow: well that's a bummer :-\ (new imdct_l being slower).. still acting up?
20:02:32preglowit might not be slower once i have room for more data in iram
20:02:42preglowi don't have time to code more on it today
20:02:48preglowwill have to see tomorrow
20:03:12preglowi refuse to believe it is slower
20:04:22 Quit muesli- (Read error: 110 (Connection timed out))
20:04:44 Join muesli- [0] (
20:08:55 Quit muesli_ (Read error: 60 (Operation timed out))
20:20:33***Saving seen data "./dancer.seen"
20:21:10 Quit rasher ("CGI:IRC (EOF)")
20:22:05 Quit Shagnar ("( :: NoNameScript 3.81 :: )")
20:29:49 Quit Aison (Read error: 111 (Connection refused))
20:30:13 Quit muesli- (Read error: 148 (No route to host))
20:39:22 Join DangerousDan [0] (
20:40:57 Join Shagnar [0] (
20:48:21amiconnZagor: r u there?
20:48:26 Join matsl [0] (
20:49:05 Quit DangerousDan ("Miranda IM! Smaller, Faster, Easier.")
20:49:49 Join DangerousDan [0] (
20:52:10 Quit DangerousDan (Client Quit)
20:53:19 Join DangerousDan [0] (
20:54:13 Join LinusN [0] (
20:57:39amiconnhi LinusN
21:00:28preglowLinusN: can i increase the iram size allowed for plugins easily? just change the value in the linker script?
21:01:14LinusNbut the iram left for the application would be very low
21:01:46preglowindeed, but does it need much? :P
21:02:04preglowwith your latest patch, there is no longer any room for windows and transform coefs in iram
21:02:45amiconnLinusN: I'm currently reading the FAT specs. It seems that we do some more things incorrectly / incomplete
21:02:53 Join Aison [0] (
21:03:23amiconn(1) The 0xE5 -> 0x05 replacement should only take place for the first char of a shortname
21:04:07amiconn(2) When extending a dir, we need to take care to not extend beyond 65535 _entries_
21:06:06webmindevening... any people here going to whatthehack ?
21:06:22LinusNamiconn: i'm sure you will find even more deviations from the spec
21:06:25 Quit Aison (Client Quit)
21:07:40amiconnLinusN: Yes, maybe. The questions is that while I am at it, should I fix these? (1, 2) should be easy to fix, so I think the answer is yes...
21:08:11 Join Aison [0] (
21:08:37amiconnAnother associated question - should I put the FAT doc into the wiki or may that cause problems?
21:09:02amiconnIt seems that it is free, however, there's a lot of license blahblah in it
21:09:23LinusNdo that
21:09:35LinusNand fix the other issues as well
21:11:50preglowbut what does/will rockbox use iram for?
21:13:58 Join asdsd_ [0] (
21:14:30 Part asdsd_
21:15:13 Nick Sucka`away is now known as Sucka (
21:15:28amiconnLinusN: Which 'do that', did you refer to fixing the spec deviations, putting the fat specs in the wiki, or both?
21:18:03LinusNpreglow: for other time critical stuff, like graphics routines, mixing waveforms, filters etc
21:18:34preglowi kind of finished the imdct_l, but to my raging surprise it was actually slower than the huge ugly blob
21:18:42preglowi hope this is due to not having any data in iram
21:19:05preglowas i can't possibly imagine how this can be slower
21:19:38LinusNno loops?
21:19:47preglowthis has a great big loop
21:19:51preglowthe old one does not
21:20:09LinusNso how cache friendly is it?
21:20:17preglowi'd say it's pretty cache friendly
21:20:20LinusNcan i see it?
21:20:51preglowthere's a lot of uncommented arm asm in it yet, it contains errors, and i want to spot them as easily as possible
21:22:07preglowthe loop is in the bottom
21:22:31preglowthe old ugly thing had no loops at all, which again makes me wonder how the hell this can be slower
21:23:23LinusNare the constants in iram?
21:23:39preglowthere's no rom for anything in iram
21:23:48preglowit's pretty close to 32kb
21:24:15LinusNi guess having the constants in iram would speed up things
21:24:19preglowyou can see the entire constant table is supposed to be in iram, but i've commented it out
21:25:40LinusNso libmad uses up 32k of iram
21:26:18LinusNmaybe you should remove one of my opts in mpa2wav then, at least temporarily
21:26:34LinusNor increase the iram space for plugins
21:26:37preglow32344 bytes
21:27:04LinusNhow big is the imdct routine, in bytes?
21:29:18preglowit seems
21:29:53preglowthe loop being 124 bytes of that :/
21:29:57preglowbut it is iterated some 12 times
21:31:59LinusNan impressive chunk of assembler code
21:32:58preglowi was hoping for a significant gain
21:33:11preglowthe old code is ugly, has no loops, and uses code space for constants
21:33:25LinusNwhat exactly is the loop condition?
21:34:26preglowit loops until it the lsb of the last const of each row is 1
21:34:33preglowi'm planning to fix that, i've got enough registers to spare
21:34:55preglowif you look at the const table at the top, you'll see the last constant has 1 as lsb
21:35:34LinusNi see
21:35:54LinusNyou really need to put the constants in iram
21:36:16preglowdone it know, tried it five seconds ago, my player hung :>
21:36:55preglowit works fine with consts in ram, so i've done some other wrong in the meantime
21:37:09LinusNcheck the mapfile
21:37:58preglowin the process
21:38:07pregloweverything seems ok
21:38:21LinusNis this the only changed file?
21:38:43 Join StrathAFK [0] (
21:38:50preglowno, i changed layer3.c to remove imdct36 from that
21:38:55preglowand just leave a decl in
21:39:19LinusNi'm not sure you should use @progbits for data
21:41:10preglowi think i checked that
21:41:20preglowbut it's too long ago to be sure
21:41:42preglowit hangs with ordinary data as well, now
21:41:48preglowgod knows why it worked an hour ago
21:41:57preglowthe sound even glitched, so i know it was the new code ;)
21:42:22LinusNdid you change the link control file?
21:42:36preglowwhat for?
21:42:46 Join Bippy [0] (
21:42:53preglowi uncommented IDATA_ATTR in mpa2wav for now
21:43:00preglowso there's room enough
21:43:14BippyAnything new worth testing ?
21:43:51BippyWhen will there be likley somthing new to test
21:43:58LinusNwhen it's done
21:44:19*LinusN goes for something to drink
21:44:20BippyCodecs i assume will be added one by one to see if they work in daily builds ?
21:44:47preglowwe've added a heap of codecs
21:44:58BippyIn daily builds ?
21:45:28BippyIf Mp3 is one thats worth testing :|
21:45:56preglowthere are a ton of codecs in the dailies, but they can't be used for listening to music directly yet
21:45:59preglowand most are too slow
21:46:09BippyOkeh pokeh
21:46:18BippyIll check back another day
21:46:20BippyBye all
21:46:24 Quit Bippy (Client Quit)
21:48:21preglowi don't fucking get this, it doesn't even load the file
21:48:28preglowi have made no changes that should merit this
21:48:35preglowapart from a cvs update, at least
21:50:08 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
21:50:08Shagnarpreglow: perhaps little bit strange question, but at the moment, ogg=>wave e.g. runs at 64% realtime, but later playing will work with high quality? or might that be complicated?
21:50:39preglowShagnar: what does the realtime rate have to do with quality?
21:51:17Shagnardunno, but thought (e.g. midi now runs also ~50% realtime, afeter decreasing the quality (48khz=>22khz)
21:51:44 Join muesli- [0] (muesli_tv@
21:51:49preglowvorbis runs at full quality
21:51:50preglowand yes
21:52:02preglowsample rate reduction is not an alternative for faster codecs, if you ask me
21:52:12preglowand i'm planning to work on vorbis once i have time
21:52:21preglowthat will almost certainly not be until summer
21:53:00 Nick t0mas is now known as t0mas_ (
21:53:17 Join t0mas [0] (
21:54:22*t0mas_ is going to shut down this pc :)
21:54:27 Quit t0mas (Client Quit)
21:54:42preglowoh wait...
21:54:51preglowhas anyone updated the plugin api or something?
21:54:53 Quit Strath (Read error: 110 (Connection timed out))
21:55:03t0mas_I did in my local copy...
21:55:11t0mas_added read_bmp_file()
21:55:16preglowit's got the be that
21:55:21 Join t0mas [0] (
21:55:43t0mas_I didn't submit it yet ;)
21:55:47preglowno, not you
21:55:50preglowbut someone
21:55:53t0mas_oh ok
21:55:54preglowi can't think of any other reason
21:56:54t0mas_ok, good night :)
21:57:02t0mas_my other client will take over :P
21:57:03 Quit t0mas_ ()
21:58:33preglowthat was it
21:58:40preglowLinusN: 20% gain with this asm version
21:58:49preglowLinusN: that is + 20%, not 20% overall
22:00:16 Join webguest82 [0] (
22:02:03 Quit webguest82 (Client Quit)
22:02:41amiconnpreglow: Starting from where?
22:02:53preglowamiconn: from 180% to a bit over 200% realtime
22:02:58amiconnNice :)
22:03:10preglowwhich isn't very much, but the code is more thoroughly done this time
22:03:10amiconn320 kbps I presume?
22:07:06 Quit izzy ("...")
22:09:21 Quit muesli- (Read error: 131 (Connection reset by peer))
22:09:55 Join muesli- [0] (muesli_tv@
22:20:36***Saving seen data "./dancer.seen"
22:36:48 Quit Harpy (Read error: 145 (Connection timed out))
22:41:39preglowbut bah
22:41:46preglowi also have to hunt down a couple of bugs in it
22:49:49 Join izzy [0] (
22:51:07 Join tedboer [0] (
22:51:24*Bagder appears
22:51:28 Quit Zagor (Remote closed the connection)
22:51:57tedboeri just build the rockbox uisimulator, just to give it a look, but i'm not sure what i can do with it
22:52:23LinusNtedboer: mainly develop ui stuff
22:52:29tedboer"No .rockbox directory","Installation incomplete"
22:52:42Bagdermake zip
22:52:44Bagdermake install
22:53:11Bagderwill install all the plugins etc
22:53:53tedboerok, thanks. is this documented somewhere?
22:54:12 Join einhirn [0] (
22:54:42tedboerit should be added in docs/UISIMULATOR
22:55:20LinusNtedboer: indeed
22:55:38tedboeri will post a patch
22:55:45LinusNtedboer: great!
22:55:48*LinusN has to go
22:55:54preglowBagder: digest has fallen and can't get up
22:55:56 Part LinusN
22:56:13*Bagder is puzzled about that
23:00:20 Join stevenm [0] (
23:00:21tedboeri am amazed how well thought out rockbox development is. patch tracker!
23:00:23stevenmHello people
23:01:09stevenmShagnar, did you say you found another bug in midi2wav ?
23:01:19Shagnarah steven
23:01:21Shagnaryes i did
23:01:31stevenmWhat happens ?
23:01:32tedboerto test rockboxui, how do i create a database?
23:01:58Shagnari think a certain instrument makes a whole group of instruments get "stuck"
23:02:03Bagdertedboer: rockboxui needs no database
23:02:10Shagnari upped the files for you, one moment
23:02:15stevenmShagnar, that is quite possible
23:02:33tedboerBadger: so how do i play with it, if there are no files?
23:02:35stevenmI severely redid that part. Can you send me a .mid file that does this?
23:02:54Bagdertedboer: the 'archos' dir simulates the root directory
23:03:00Bagderyou can fill it with any files you want
23:03:23 Quit muesli- (Read error: 113 (No route to host))
23:03:23Bagdertedboer: try browsing plugins and run them
23:03:27stevenmShagnar, ah thanks
23:03:29Shagnarthe ogg file is the result of midi2wave
23:03:39Shagnarposition 02:10 it happens
23:04:23tedboeri created 1000 mp3 files, and i get a dir buffer is full... is that expected?
23:04:55Bagdertedboer: if you really want that many in a single dir, you need to change the setting that allows it
23:05:27tedboerok. just testing.
23:05:36tedboerthe plugins are the .rock files?
23:05:58tedboersorry if i'm asking the obvious
23:06:14Bagderno worries, we're happy to get another person aboard the ship
23:06:31preglowyou're on 68k assembly duty for every obvious question you ask
23:06:50Bagderpreglow: you've must been asking quite a few ones ;-)
23:06:57stevenmShagnar, hmm, no voice overflows.. let's look at the file
23:07:19tedboerhmm, i copied all .rock files to archos, but i don't see any plugins
23:07:19Shagnarif you want so, i can you send the wave, too (14megs)
23:07:21preglowi kept asking if you supported ipod
23:07:31preglowit needs to be in rocks/
23:07:39preglowand viewers/, if they're viewers
23:07:45Shagnarstevenm i also tried another midi, worked fine without stopping
23:07:54Bagderand you need "view files" set to "all" to see them
23:08:00stevenmShagnar, I just ran that MIDI thru it
23:08:14Shagnarah okay :)
23:08:18preglowdoesn't "make zip" make a zip with the correct dirs?
23:08:24Bagderyes it does
23:08:32stevenmShagnar, no voice overflows, so nothing gets 'stuck' in a sense of the synth screwing up. Could be a sequencer bug then
23:08:37Bagderand make install unpacks the zip properly
23:08:52tedboerwell, i did that..
23:09:06tedboerbut i don't see any plguins
23:09:19Bagderwhen how?
23:10:12tedboermkdir build; cd build; ../tools/configure ; make ; make zip; make install; ./rockboxui
23:10:25stevenmShagnar, did you say it happens at 2:10 ?
23:10:35Shagnar02:10-02:10 something about that
23:10:39Shagnari ment
23:10:48stevenmShagnar, what specifically happens that should not?
23:11:15Bagdertedboer: press INS
23:11:38tedboerINS does nothing
23:11:42Shagnar? stevenm, all instruments (except the drums) stop playing
23:11:52tedboerreceived ev 65456
23:11:53BagderINS on the numerical keyboard I should add
23:11:57 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
23:11:58Shagnardidn't listen to the ogg file? stevenm
23:12:04Bagderor .
23:12:06tedboeryes, that INS
23:12:10stevenmShagnar, I see what you are saying. I just opened up the file in a MIDI editor
23:12:24stevenmShagnar, looks like the file ends at 2:10.
23:12:38stevenmShagnar, now the question arises - why don't the drums stop playing with everything else ?
23:12:43tedboeri get an empty screen with the 4 icons
23:12:45Bagderwell, INS on the numerical is supposed to pop up the menu
23:12:57tedboerit doesn't...
23:13:03Bagderah, switch off numlock
23:13:05tedboerthat's CVS checkout
23:13:29Shagnarstevenm yes you're right... just didn't think of THAT case
23:13:42tedboera warning would be nice :-)
23:13:52Shagnarso the drums are what is wrong
23:14:05Bagdertedboer: feel free to fix!
23:14:10tedboeri will!
23:14:23amiconnstevenm: If you want a complex MIDI file:"> It's classical music...
23:14:41Bagdertedboer: btw, in the first screen you'd see all files you have in the archos/ dir if you had any
23:14:45stevenmShagnar, It is strange for some reason. The drums keep going for a few more seconds. Maybe there is some sort of STOP event that I am ignoring, or maybe the tracker gets offset
23:14:50Bagderand dirs too of course
23:15:09 Join rasher [0] (
23:15:11Shagnarcould be
23:15:12stevenmamiconn, thanks.. will look
23:15:23stevenmShagnar, well... lets have a look at the events
23:15:31tedboerdoes the simulator try to run in realtime?
23:15:37stevenmfunny thing.. for a STRANGE ass program, this editor works very well on wine
23:15:39Shagnarshall i test some more midi files, so we can perhaps find out whats wroing? (perhaps only this midi file is corrupted?)
23:15:43preglowthe simulator runs as fast as it can, afaik
23:15:47stevenmShagnar, cool
23:16:21tedboerpreglow: w~
23:18:04tedboerpreglow: but what kind of hardware would be needed to simulate correctly?
23:18:05stevenmamiconn, dang that's long
23:18:57preglowtedboer: almost anything will do
23:19:10stevenmamiconn, is this thing really 10 minutes long, or is something very wrong ?
23:19:24stevenmamiconn, sounds right to the end ..
23:19:25preglowif it's classical, it might very well be 10 minutes
23:19:29tedboeran amd athlon xp @ 1100 mhz/
23:19:36stevenmpreglow, yea, very true
23:19:52Shagnarshit :-/ forgot to set the clock speed at 120mhz
23:20:23amiconnstevenm: This length sound about right to me.
23:20:27stevenmpreglow, woah if we can get amiconn's file running realtime ... ... ... THAT would be sweet.
23:20:42amiconnBah, my UPS screams overload and I don't know why :(
23:20:45tedboerok, thanks for the help. see you later.
23:20:48 Quit tedboer ("Leaving")
23:21:50stevenmso I talked to my enee professor about endianness and coldfire.. apparently some people in our class too want to beat motorola in the head with a stick
23:22:41stevenmer endian
23:22:48preglowbeat intel for using little endian, that's what i say
23:22:54preglowmotorola uses the sane endianess
23:22:58stevenmwell there was a group of people who wanted to beat manufacturers in heads with sticks
23:23:05stevenmhalf intel, half motorola
23:23:09preglowsure, i too would love to browbeat motorola
23:23:15preglowbut not for their choice of endianess
23:23:36preglowmore for them not using a harvard architecture, or having a data cache
23:24:08stevenmpreglow, in any case, I wanna try loading the file 1 byte at a time, reconstruct the actual byte by shift and or, then friggin ASSIGN the value to a location in the array (cast it to short*) one at a time
23:24:25amiconnI'd really would go to beat them if they chose a harvard arch
23:24:25stevenmand let the compiler figure out how it wants to store it.. as long as it's always consistent and I don't screw with the array directly
23:24:39preglowamiconn: why?
23:24:57preglowamiconn: sharing the code and data bus doesn't exactly do this processor any favours
23:25:40preglowa harvard arch would make rockbox's job harder, though
23:26:17 Join amiconn_ [0] (
23:26:38preglowstevenm: you have to deal with endianess yourself, the compiler will never do it for you
23:26:55 Quit amiconn (Nick collision from services.)
23:26:55 Nick amiconn_ is now known as amiconn (
23:27:04amiconnGah, guess what happened :(
23:27:10amiconnDarn UPS
23:27:18preglowstevenm: i'd just read the samples in to their short array, then plain and simple do a SWAP16 on every short
23:27:26preglowamiconn: doesn't sound very uninterruptible to me
23:27:38amiconnhaha, yes
23:28:26*rasher looks at o.O
23:28:34stevenmpreglow, hmmm.... all right, will try that. only was afraid that last night we swapped and still got garbage. but I will try again
23:28:35amiconnI wonder why that happened though. I didn't change the connected devices recently, and it's the very first time it screamed overload...
23:28:45stevenmmaybe I didn't swap it right or something.
23:28:51preglow23:24 < amiconn> I'd really would go to beat them if they chose a harvard arch
23:29:14preglowrasher: ouch
23:29:18preglowwhat's happened
23:29:33amiconnThat would make rockbox's job way harder, and certain things almost impossible
23:29:47preglowamiconn: yes, i did mention that
23:29:53preglowbut you didn't see that, no
23:30:12amiconnLook at the gmini. Calmrisc16 is harvard, and I hardly can think of a way to make plugins work
23:30:31amiconn(No RAM in instruction memory space)
23:30:37preglowwell, it's not possible
23:30:45preglowsince code memory bus is always flash
23:30:47preglowmore or less always
23:31:09rasherpreglow: no idea.. maybe sf hiccuped
23:31:22Bagdermost likely
23:31:36Bagderthe script runs twice per day
23:31:39Bagderlet's wait and see
23:31:48preglowi vote for PANICING
23:31:58rasherDON'T PANIC
23:31:59*preglow runs around, hands in the air
23:32:45preglowi guess i'd better find the bugs in imdct_l before i forget how it works
23:32:48stevenmwhy panic? what happen?
23:32:48*preglow clenches teeth
23:33:02rasherstevenm: patches page is blank
23:33:13stevenmrasher, well that's no good......
23:33:21rasherno biggie
23:33:27rasherjust surprising
23:33:34stevenmdidn't know there WAS patches page though... sourceforge being dumb ?
23:33:47rasherMost likely
23:33:58Shagnarstevenm tried another midi, worked fine. now i'll try to test one with more drums ;)
23:34:00rasher < that page was probably broken at some point
23:34:11stevenmShagnar, awesome
23:34:35Bagderproject admins on sf have a "download project as a huuuuge XML" file
23:34:43stevenmShagnar, I'm considering messing with the getsample optimization some more.. that means endian crap. feel like changing rocks sometime soon ?
23:34:44Bagderwhich my script downloads
23:34:55Shagnarof course :)
23:35:04rasherBagder: ah
23:35:08stevenmShagnar, awesome.. once I figure out where to begin ..
23:35:27stevenmShagnar, at least we have semi optimized code in CVS this time, so I can clobber this all I want
23:35:50preglowstevenm: ehh?
23:35:57preglowstevenm: you really should get rid of getsample
23:36:04stevenmpreglow, I know.. I am about to.
23:36:09preglowstevenm: a function call for something that should be that basic is bad
23:36:35stevenmpreglow, yes.. It is what actually did the sign conversion and bit conversion
23:36:57preglowthat should be precalc
23:37:02stevenmpreglow, I wanted to keep things localized, just to get it working.. last night I tried getting rid of it and something went wrong. Now will try again
23:37:30stevenmI moved that stuff into guspat.c to run once (no point, its all the same ops over and over). It became FASTER but somewhere life got messed up
23:39:40 Join ashridah [0] (
23:45:50stevenmpreglow, what is the argument to SWAP16 again? * (short) ?
23:46:07preglownever used it, look it up in the headers
23:46:15stevenmpreglow, all right
23:48:18Shagnarstevenm tried another one with drums again, also without any error. so i'll wait for a new midi2wave :)
23:48:44stevenmShagnar, yea... I am having some issues nwith SWAP16.. namely it segfaulting and me not being able to see where it's defined
23:49:07 Join LinusN [0] (
23:49:17LinusNShagnar: it's SWAB16, not SWAP16
23:49:55preglownow there's service for you
23:50:02Bagderit was stevenm who asked
23:50:12stevenmoo, Linus comes to rescue. Thanks!
23:50:21LinusNsilly me, i meant stevenm
23:50:35stevenmwhat is weirder that SWAP16 actually compiled.. must be defined somewhere in standard c or something
23:50:53preglowc is weird
23:51:27 Quit edx (Read error: 110 (Connection timed out))
23:55:02stevenmpreglow, STRANGE. As soon as I have swab16, it segfaults.
23:55:09amiconnI think SWAP16 is just warned about not being defined, and evaluates to nothing
23:55:31stevenmit's defined as unsigned short swab16(unsigned short)
23:55:39stevenmamiconn, well that's not very useful..
23:55:39amiconnstevenm: You really should check for compiler warnings. Your latest commit added 3 of them
23:55:57amiconnand it's SAWB16 not swab16
23:55:59Bagder4 actually
23:56:04amiconnerr, SWAB16
23:56:14stevenmamiconn, true.. most of it is signed vs unsigned crap. I clean that up once it's basically working.. yea I used SWAB16
23:57:04amiconnMrf, FAT code causes me headaches :-/
23:57:04stevenmah there it goes, that aint defined either.. need system.h
23:58:25stevenmWhere the heck is this thing defined ?

Previous day | Next day