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 2005-06-11

00:00:29 Nick StrathAFK is now known as Strath (
00:01:27Tangledingbye all :)
00:02:39Tangleding(i just DL the latest bleding edge)
00:05:29road_runneris the latest bleding edge now differs?
00:06:29Tangledingyes some bugfixes :)
00:06:37 Join thegeek [0] (
00:06:45 Nick Strath is now known as StrathAFK (
00:07:37Nuxatorok now it's compiling
00:13:36 Quit Tangleding ("Chatzilla 0.9.68a [Firefox 1.0.4/20050511]")
00:16:29road_runnerAHHH !! I'm flashing in a few minuts!! i really hate this part!!!
00:20:21road_runnerah. my stomach wants to flip.. disconnecting usb prot...
00:20:24 Quit Harpy (Read error: 60 (Operation timed out))
00:20:45road_runneririver fw is loading...
00:21:23road_runner>general>firmawre upgarde (ho please god please godDddD)
00:21:49road_runnerUPGRADING! do not touch during firmare upgrade...
00:21:51preglowgod has left you now!
00:21:54road_runnerh.d stops spinning....
00:22:08road_runnerheart is pounding!!!
00:22:28 Quit pbvas ("Leaving")
00:22:30preglowif this is what a firmware upgrade is for you, i'd hate to see how you handle life-threatening situations
00:22:42road_runnerplyaer is off.f.f... . pleasee!!! pressing PLAY
00:23:01road_runnerahhh... it nowoyhdfh
00:23:21road_runnerwhat is this ?!
00:23:30*preglow stands by for more tremor optimisations!
00:23:31road_runnerahhh it works!!
00:23:37road_runnerwait, now the iriver fw
00:23:50road_runnersystem off... now with the REC BUTTOn
00:23:58 Quit Jason ("CGI:IRC (Ping timeout)")
00:24:01road_runnerit boots!
00:24:05preglowthen hooray!
00:24:35road_runneroomg... it's all over. i'm all over it.
00:24:41road_runnernow to the story
00:27:46road_runnerho it's all in hebrew. I have a log file in hebrew of all my problems with it. twice it bricked after a bad firmware upgarde, one time I don't know why (used the right file), the other - they send me it with the firmware of the iHP 15 GB version (was not published all around the world). another time - the main screen of the unit was ugly broken (but system worked fine....) and another time - was burned by an electrical shock.
00:28:45road_runnerbut now it's unbelivebly works
00:28:45 Quit Nuxator ("CGI:IRC (EOF)")
00:33:24 Quit yngwi ("Chatzilla 0.9.68a [Firefox 1.0.4/20050511]")
00:34:05 Join silencer [0] (
00:37:47HClgotta hate flashing
00:41:07HClan iriver 15gb? o.o.
00:43:47HClwhat kind of specs do those have?
00:43:54HClbased on 110 or 120?
00:44:10MoosCamaropreglow: tremor optimisations comitted by Pedro :)
00:44:23preglowMoosCamaro: i know, just spoke to him about them
00:44:29amiconnHmm, that's strange. I believed H-115 was like H-110, 16 MB ram
00:45:09MoosCamaropreglow: :)
00:45:25HClamiconn: mhm... thats what i'm wondering.
00:45:55amiconnpreglow: Mhm, LINE_ATTR ;-)
00:46:08preglowamiconn: might be clever to stuff that LINE_ATTR thing he does in a main rockbox header somewhere, considering your movem discovery
00:46:12preglowahh, yes
00:46:48amiconnProbably we don't want to do that anywhere, might waste a fair amount of ram
00:47:15amiconnIt surely helps if the data structure is >= 16 bytes and accessed with movem
00:47:16pregloweverywhere, you mean? the ram waste would be negligable for the few places where we know it would matter
00:47:52preglowespecially considering the possible gain
00:48:16HClwe waste like what, 12 bytes?
00:48:28preglowworst case, yes
00:48:54preglowor in strange cases, 15 bytes
00:54:03HClheh, tremor guy knew about line attr all along?
00:54:12HClor did he do the improvements after reading?
00:54:34preglowhe asked me about it now
00:54:49preglowbut he suspected it, it seems
00:58:33road_runner115 was probebly based on 110 otherwise it shouldn't have crashed when they put me the wrong firmware un my player and i've updated it to the right one.
00:58:47HClthen rockbox shouldn't have booted on it though.
01:00:11preglowlemme see if i get this correctly
01:00:16preglowyou've just flashed rockbox on a 115?
01:00:49road_runnerrbx l00ks wow, AMAZIng. way the F*cking go! it has like everyting i ever dreamed of in a player, honest! it's totally amazing. really nice to think that after all that time this would really be one of the best possible players. will donate soon.
01:02:37road_runnerno. they once send me my unit flashed with a 115 fw. (you right it isn't based on a 110 by that..) then I tried to flash it with 120 content (They just told me to ship it back to them. didn't had a solution, their fault, they knew it's the wrong firmware, but still i tried.) and it crashed.
01:04:46amiconnCommit coming up...
01:12:28 Quit matsl (Remote closed the connection)
01:13:23 Join matsl [0] (
01:17:16preglowthe musepack struct is way too large for it to fit in iram in any way
01:18:15RoriI really hate my telco
01:18:24RoriNo free upgrade to 2mbit for me
01:19:01RoriAll because they did not even bother testing the line but rely on an out of date database that is incorrect. I now have to pay for a manual upgrade instead.
01:19:17RoriWhich means they HAVE to do a line test
01:19:45RoriSo unfair :P
01:20:43HClpreglow: split it up in substructs?
01:21:30preglowHCl: yeah, possibly, that or make some of the bigger arrays pointers that we assign somewhere in an init
01:23:30 Quit matsl (Remote closed the connection)
01:24:21 Join matsl [0] (
01:25:09preglowHCl: but yeah, splitting it to two subtructs where one contains the iram stuff might be clever
01:25:32amiconnHmm, I need to take care of the different chargers. The plugs of the iriver and the archos chargers are *very* similar and fit into the 'wrong' socket
01:25:57preglowput some tape on one of them
01:26:04preglowred tape, for the sake of irony
01:27:08amiconnIt might be fatal for the iriver to feed it unregulated 9 volts
01:27:31amiconnThe other way round is no problem, the archos just wouldn't charge
01:27:37amiconn(polarity is the same)
01:27:39preglows/might be/surely is/
01:27:47HClwhat preglow said
01:28:17 Quit t0mas (Read error: 60 (Operation timed out))
01:28:47preglowi'll perform shakespeares collected works in the nude publicly if your iriver survives that
01:28:59HClthats worth a shot
01:29:00HCl :P
01:29:10amiconnpreglow: I won't try that for sure
01:29:23preglowheh, i sure as hell won't force you, having made the claim i have
01:30:31amiconnThe iriver dc in isn't very well designed, methinks
01:30:33preglowh1x0 units seem to get sweating fits if they're as much as within a metre of a nokia charger
01:30:47HCl :p
01:30:54HCli'm glad to get rid of my nokia charger
01:30:57preglowthey start to wiggle away from it
01:31:26preglowi sure as hell keep mine as far away as i can
01:31:28*amiconn has a SonyEricsson...
01:31:50preglowamiconn: i'm sure it won't take any kinder to those chargers either, assuming the plug even fits
01:31:53amiconn...but 2 archoses otoh...
01:32:07amiconn...with a charger, that is
01:32:33amiconnpreglow: SonyEricsson has a totally different plug
01:34:14preglowbut yeah, if you value your h140, never ever plug in the wrong charger
01:34:19preglowit'll fry itself given half an excuse
01:34:38MoosCamarogood night all
01:34:56 Join bipak_ [0] (
01:35:00 Part MoosCamaro
01:35:01preglowmy, musepack math really is shifting hell
01:36:26***Saving seen data "./dancer.seen"
01:36:37preglowwith a bit of luck, this wont go well with the emac unit at all
01:36:39*preglow gives it a shot
01:37:17HClbad luck you mean?
01:37:34HClcan't you use times 2^shifting ?
01:37:47preglowno, that's not the problem
01:38:10preglowthe emac fractional mode has an implicit << 1 shift of the result
01:38:38preglowwhich might not go to well with musepack
01:41:22preglowwhen you cast a 64 bit int to a 32 bit int, you get the lower 32 bits, yeah?
01:41:53preglowwell, shit
01:42:33amiconnYou could use a union
01:42:43preglownah, problem is i can't provide a 64 bit int
01:42:53amiconnlong long?
01:43:18amiconnOr are you talking about emac?
01:43:23preglowemac can only provide top 32 bits, what musepack does is do a full 32x32 -> 64 bit mul
01:43:31preglowthen shift the result down quite a few bits and cast to 32 bit int
01:45:10crwlhum, is vorbis actually faster than mp3 on rockbox currently?
01:45:27preglowsince i can't provide any of the lower 32 bits (except the top lower 8, but at great penalty), i'll probably have to do prescaling
01:45:33preglowcrwl: sincerely doubt it
01:45:49preglowcrwl: might be with the wildly inefficient dither codecmpa uses
01:45:58crwlpreglow, vorbis2wav and mpa2wav behave differently then?
01:46:16crwlwith vbr files with similar bitrates (~200-220) i seem to get ~130% with mp3 and ~170 with vorbis
01:47:06 Join t0mas [0] (
01:47:29preglowwell, yeah, mp3 uses a really slow dither at the moment
01:47:31preglowmight be that
01:47:37preglowvorbis has no dithering
01:47:48preglowdither should be an option, imho
01:47:56preglowi most certainly can't hear dithering at 16 bits
01:48:08crwlme neither
01:48:41preglowdithering should also probably be moved from the codecs to the rockbox core
01:49:14amiconnpreglow: Is the downshift after multiply constant?
01:49:56preglowamiconn: sometimes
01:50:12preglowamiconn: they've got a constant shift mul and a variable shift mul, i'm about to study the code to see what's most commong now
01:51:58preglowthe only mul that is really friendly is the FRACT mul, which does a full 32 bit shift down
01:52:02 Join webguest42 [0] (
01:53:20 Quit webguest42 (Client Quit)
01:54:17preglowfortunately, this is what the main synth loop uses
01:55:32crwlhow do i properly convert a font to rockbox format?
01:56:07crwlthere's convbdf which turns bdf fonts to .fnt files, but maybe my bdf font isn't in a proper format
01:56:13t0masgood night :)
01:58:10 Join ]RowaN[ [0] (
01:58:25]RowaN[guys, what is car adaptor mode? and does it work on iriver
01:58:27 Join stripwax_ [0] (
02:00:05preglowhello, stripwax
02:00:49crwlhmm, limiting the number of characters did the trick
02:01:15stripwax_preglow- hi! I actually just finally installed rockbox for first time on iriver.. awesome news to come back to after a vacation. i'm totally blown away.
02:01:43preglowagreed, heh
02:01:52stripwax_i'm hearing it but i'm not believing it :-) figured it was about time to donate to rockbox too
02:03:05]RowaN[i'll donate to rockbox, it would be nice if i could get crossfading to work mind you
02:03:43]RowaN[i donated to zsnes 2 years ago and i still havent been added to the donators list, meh
02:04:00preglowi really prefer to be anonymous when donating
02:04:02nobbehdoes rockbox have a donator list?
02:04:23preglowanyone know how one does a 64 bit mul with only 32 bit hardware multipliers?
02:04:53stripwax_preglow- do four 32-bit mults, shift, and adjust for the sign bit
02:05:11preglowfour 32 bit muls, yes
02:05:22preglowthat explains why musepack is so incredibly, amazingly slow
02:05:56preglowthis friggin requires an arm to be an easy port
02:07:24stripwax_- actually, you'd need to do a bit more than that, as you need to adjust the sign of each of the mults, and that requires more bits.. hmm..
02:07:43stripwax_four 33-bit mults, or four 32-bit mults and some extra stuff too
02:08:56stripwax_I'll see if I can find that section of my dissertation :)
02:12:29preglowi just sped libmusepack up by 100%
02:12:31*HCl still hasn't tried music playback on iriver :X goes to try
02:12:37]RowaN[im guessing car adaptor mode is a feature that stops the battery from charging when plugged in?
02:12:42preglowreplacing an innerloop with asm
02:12:52HClrealtime now?
02:13:00HCl :X
02:14:45]RowaN[is this the right channel for me to be asking rockbox questions or am i invisible?
02:14:55HClnope, thats the right channel
02:15:06HClcar adapter mode is for making it controlled by car controls, i think.
02:15:08HCli dunno
02:17:12stripwax_Oh, I remember. You need to use a combination of signed and unsigned multiplications: m*n = (a<<32 + b)*(c<<32 + d) where a,c are signed 32 bits, b,d are unsigned 32 bits. so compute bd + ad<<32 + bc<<32 + ac<<64 but of course bc (and ad) are signed*unsigned ..
02:17:46stripwax_dunno if you have that sort of flexibility on coldfire or not
02:18:53preglowgreat, musepack outputs white noise now
02:19:40preglowi don't know, gcc does the 64 bit muls for me
02:19:48preglowbut it's obvious a lot of work is involved
02:19:56preglowespecially since coldfire can't do cross register shifts
02:21:14amiconnYou could check how gcc does it
02:23:21preglowwell, it doesn't matter
02:23:25preglowit clearly is heavy
02:23:34preglowand explains why libmusepack is so extremely slow
02:24:11preglowthis doesn't look good
02:25:17preglowi'm really not very confident in gcc's ability to not screw up code around an inline asm bit
02:25:25 Nick StrathAFK is now known as Strath (
02:26:37preglowdo you know how to make objdump stop outputting mit syntax assembler?
02:26:58*HCl wonders what time it is where preglow lives..
02:27:07preglowoh, around 02:30
02:27:28HClsame here
02:27:38HClyou should go to bed and look at it tomorrow ;x
02:28:01preglowneed to get it work now!!
02:28:04HCl ;p
02:28:18HCli know how that goes.
02:28:23amiconngood night all
02:28:26stripwax_preglow - what do you mean by "cross register shifts"?
02:29:16preglowstripwax_: well, the 64 bit result obviously ends up in two 32 bit registers, and libmusepack does a lot of shifting on the 64 bit result
02:29:22preglowamiconn: nite
02:29:51HClwhy does musepack use 64bit ints so much
02:30:07preglowall codecs use 64 bit ints much
02:30:10preglowthey need the precisio
02:30:27preglowbut usually the top bits are the only ones of interest
02:30:33preglowbut musepack uses the ones in between as well
02:32:49 Nick DJDD_SLEEP is now known as DJ_Dooms_Day (
02:34:23*HCl goes to sleep
02:43:31 Quit matsl (Remote closed the connection)
02:43:39preglowwoot, recognisable output!
02:51:13preglowwoot, correct output!
02:51:52HClwhat speed?
02:52:07preglownearly 100% for some file
02:52:22preglowit's okish
02:52:25preglowi'll commit this then go to bed
02:52:41 Quit preglow ("foop")
02:56:16 Quit road_runner ("CGI:IRC (EOF)")
02:59:41 Part stripwax_
03:20:56 Join [solid] [0] (~solid@
03:31:31 Join amiconn_ [0] (
03:36:27***Saving seen data "./dancer.seen"
03:40:02 Join ashridah [0] (
03:40:03 Quit amiconn (Read error: 60 (Operation timed out))
03:40:03 Nick amiconn_ is now known as amiconn (
03:57:27 Join Lost-ash [0] (
04:05:26 Join QT_ [0] (as@area51.users.madwifi)
04:05:30 Quit [solid] (Read error: 145 (Connection timed out))
04:07:08 Quit ashridah (Nick collision from services.)
04:07:36 Nick Lost-ash is now known as ashridah (
04:08:48ashridahhrm. i'm getting skipping in one of my tracks when the system kicks back in to refill the buffer after a few minutes :(
04:09:10 Quit QT (Read error: 60 (Operation timed out))
04:16:54 Join stripwax_ [0] (
04:21:12 Join kergoth [0] (
04:22:39 Part kergoth ("Leaving")
04:24:45 Quit markun ()
04:54:04 Quit ehntoo ("Leaving")
04:56:09 Part stripwax_
05:28:36 Join DJ-suparappa [0] (
05:36:31***Saving seen data "./dancer.seen"
05:47:02 Quit gromit` (Remote closed the connection)
05:56:50 Join DMJC [0] (
06:33:38 Quit tvelocity ("Leaving")
06:33:48 Quit ashridah ("Leaving")
06:34:55 Quit DJ-suparappa ()
06:45:04 Join StrathAFK [0] (
07:10:34 Quit Strath (Read error: 110 (Connection timed out))
07:16:23 Join bumi [0] (
07:17:03bumihey everyone
07:17:07bumianyone here seen ziv latly?
07:17:21bumiam wondering about the hebrew display patch....
07:18:21bumianyone even here?
07:20:09bumiill take this as a no
07:20:10bumibye then
07:20:12 Quit bumi (Client Quit)
07:36:34***Saving seen data "./dancer.seen"
07:43:39 Join ashridah [0] (
07:59:41 Quit ashridah ("Leaving")
08:06:30 Join ashridah [0] (
08:11:54 Join webguest66 [0] (
08:12:54 Quit webguest66 (Client Quit)
09:00:44 Join Harpy [0] (
09:30:39 Join Strath [0] (
09:36:38***Saving seen data "./dancer.seen"
09:48:27 Join preglow [0] (
09:48:48 Join yngwi [0] (
09:49:14 Quit StrathAFK (Read error: 110 (Connection timed out))
09:52:56 Quit DJ_Dooms_Day (Read error: 110 (Connection timed out))
09:54:30 Quit bipak_ ("Lost terminal")
09:58:14t0masmorning :)
10:07:33 Join bipak [0] (
10:09:03 Quit DMJC ("Leaving")
10:10:43preglowmy, what buffer sizes musepack operates with
10:12:28 Join DMJC [0] (
10:23:39 Part DMJC ("Leaving")
10:57:05 Join stripwax_ [0] (
10:58:45 Quit yngwi ("Chatzilla 0.9.68a [Firefox 1.0.4/20050511]")
10:59:40 Join niobos [0] (
11:08:37 Join DomZ [0] (
11:08:47DomZ1218188 None 5 nobody Don't wrap-around when the button is autorepeated 2005-06-10 15:01
11:08:55DomZnew patch :)
11:09:30DomZjust in case you miss it
11:13:22niobosmorning all
11:14:34preglowDomZ: we know about it, linus liked the idea
11:16:00DomZoups ok :)
11:18:37preglowanyone know of any good ripping/encoding programs for linux?
11:19:14 Join [solid] [0] (~solid@
11:19:35preglow[solid]: yo, did you do a vorbis battery test?
11:20:22stripwax_could someone quickly explain *when* we do the cpu boosting on iriver vorbis decode? is it just when the pcm buffer gets low?
11:20:46SlasheriI am now running mp3 battery test, unless i will do some development today and have to stop the test
11:20:49[solid]preglow: no... i was waiting for the skipping problem (which occured most intensively with vorbis) to be resolved
11:21:26[solid]preglow: i'll see if it's okay with current bleeding edge - if it is i'll do the test as soon as it fully charges :)
11:25:08preglowoh, what a bloody shifting hell libmusepack is
11:28:15niobospreglow: for linux I use abcde
11:28:17 Join gromit` [0] (~gromit`
11:28:18niobostogether with lame
11:28:29preglowdoes it support fetching cd info from freedb?
11:28:29niobosabcde is a frontend for cdparanoia
11:28:40preglowok, i'll give that a spin, then
11:28:55niobosdoes freedb, ripping, encoding and tagging
11:29:08niobosabcde = "a better cd encoder"
11:29:52*niobos is back to the study-books
11:29:57 Nick niobos is now known as niobos_study (
11:36:42***Saving seen data "./dancer.seen"
11:57:30 Quit Zoom2 ("CGI:IRC (EOF)")
12:05:24preglowSeed: do you know of any musepack coders that wanna port this? :/
12:07:33stripwax_preglow - how bad is it?
12:08:43preglowi'm not sure, it just know it isn't trivial
12:08:53preglowi've got it close to 100% realtime now, though
12:09:01linuxstbNot sure if anyone apart from me is interested, but I've just added a codeca52 (for A52/AC3) to CVS.
12:09:07stripwax_synth_filter.c, right?
12:09:13preglowstripwax_: yes, that's the major part
12:09:16linuxstbIt's already faster than realtime - it doesn't even boost the CPU all of the time.
12:09:21stripwax_linuxtsb - ooh, nice!
12:09:25preglowthey rely heavily on a full 64 bit mul being available
12:09:55linuxstbOf course, the problem with AC3 at the moment is that all my files are 48KHz.
12:10:07preglowlinuxstb: why not start working on that at once? :)
12:10:26linuxstbpreglow: I'm tempted. Would you be able to send me your resampling code?
12:10:55stripwax_q- why would it need to resample?
12:11:08Rorihas low bitrate mp3 playback been looked at yet?
12:11:11linuxstbThe iRiver hardware can't do native 48KHz.
12:11:18preglowRori: looked at? why? doesn't it work?
12:11:32linuxstbGotta go - back in 5 minutes...
12:11:35Roriwell audio books play back too fast
12:11:36preglowRori: who the hell wants low bitrate mp3 anyway? ;)
12:11:45preglowRori: that's a sample rate issue, not a bit rate issue
12:12:01stripwax_linuxtsb - really??!? blimey. I was fairly certain it could.
12:12:07preglowstripwax_: it can't
12:12:40Roriok sample rate issue then. 22khz 11khz :)
12:12:49preglowhmm, i guess i could make the resampler work in place, but then it gets degraded to just being a downsampler
12:13:58preglowlinuxstb: my code can't be used, it's floating point based
12:14:20stripwax_so does the stock iriver firmware resample 48khz to 44.1khz too? that blows!
12:14:29preglowit resamples everything to 44.1khz
12:14:43preglowand it'll have to be a cheap resampling as well
12:14:51preglowcurrently i'm planning on just doing linear interpolation
12:15:51Roriwhat about ac3 via toslink?
12:16:29preglowforget it
12:16:46[solid]linuxstb: maybe you should look at secret rabbit code?:)
12:16:48Rorinot a big deal really
12:17:13preglowsecret rabbit code is too expensive
12:17:22preglowwe can't do that kind of interpolation
12:17:51[solid]that's what i thought..
12:18:03preglowat least the sinc based resamplers are
12:18:13preglowbut apart from them, there's only linear interpolation, and that i can code myself
12:19:04stripwax_cubic interpolation's just a couple of multiplies, adds and shifst per sample. would that be too expensive?
12:22:14linuxstbI'm going to send off an email to the linux-audio-dev list - asking if anyone has a GPL'd fast integer resampler.
12:22:40stripwax_preglow - would it make sense to move A00, A01, ..B00,B01,.. etc in Calculate_New_V into IRAM?
12:23:23[solid]hmm... it seems bleeding edge is skipping no more
12:23:35[solid]so the vorbis test will take place today :)
12:24:19preglowstripwax_: they already are in iram
12:24:54crwli actually found no other odd problems with rockbox except that playlist queue thing
12:25:03stripwax_preglow ah ok. are stack local variables always in IRAM then?
12:25:14preglowthe stack is in iram, yes
12:25:31stripwax_preglow cool
12:25:36stripwax_didn't realise
12:27:52 Join MoosCamaro [0] (
12:31:18preglowis it possible to make a c macro depend on its parameters?
12:31:44preglowno, of course not
12:31:52preglowi'll have to do it with c code in the macro
12:32:27stripwax_preglow? how do you mean, 'depend'?
12:34:32preglowi need to output one asm instruction if a parameter is above a certain threshold
12:34:42preglowand another if it's below
12:35:09stripwax_could probably achieve that using template metaprogramming in C++ but.. not in C preprocessor I don't think
12:38:30stripwax_no, wait: could the parameter come from an arbitrary range or one of only a few likely values?
12:42:44stripwax_.. because if so, you could use ## to contenate the argument value into the macro name itself..
12:43:09preglowahh, sorry, busy for a bit
12:43:22preglowi can't, the i need to shift right for some values, and left for some others
12:45:08stripwax_that should still work.. what are the exact values and what is the threshold? (i'm thinking -32 < v < 32 and threshold is 0 but I'm probably thinking wrong)
12:46:38preglowvalues range from 32 to around 15 or something, i guess
12:47:07preglowonly 32 needs right shift, rest needs left shift
12:47:15stripwax_so something like SHIFT( a , n + 32 ) where #define SHIFT( VAL, NSHIFT ) SHIFT ## NSHIFT( VAL ) where #define SHIFT_0( X ) something, #define SHIFT_1( X ) something etc
12:47:51stripwax_oh.. silly question but are the values constant or not? if not, then obviously you'll need to use C code
12:47:58preglownot always constant
12:48:13preglowfuck this, i don't have time
12:51:07stripwax_preglow yeh. i can take a look if you like.
12:51:17preglowi have one more opt uncommited
12:51:34preglowand that's an asm version of the FRACT multiply macro
12:52:01stripwax_hmm.. will the compiler optimise away the Data var in Synthese_Filter_float_internal? seems to be just a copy of OutData
12:53:02stripwax_preglow cool
12:59:34stripwax_preglow wow, so MAKE_MPC_SAMPLE does a 64bit multiply by 1<<14 rather than just a couple of shifts?! is that the thing you're optimising right now?
13:06:20 Join yngwi [0] (
13:06:26preglowonly thing i've done that actually works is an asm version of MPC_MULTIPLY_FRACT
13:07:06stripwax_ok. math.h is pretty crazy
13:07:16preglowit is
13:11:05preglowbut we have serious problems anyway
13:11:50preglowMPC_MULTIPLY_NOTRUNCATE, for example, shifts the 64 bit answer 14 bits down
13:12:06preglowfor us, that quite simply translates to a sixteen bit loss of precision, if we want any semblance of speed
13:12:36 Quit edx (Read error: 110 (Connection timed out))
13:14:27stripwax_Could we do it all in emac and use an 8 bit fractional part instead of 14?
13:16:12preglowwell, i don't know anything about the precision needs of musepack
13:25:48nobbehhey, im setting up a cygwin development environment on windows XP. Im going through the installer now. Do i want the latest version of everything listed in step2 here: ???
13:26:10nobbehand do i need source as well as binarys for each of those?
13:26:44ashridahSlasheri: around?
13:27:14stripwax_nobbeh - you don't need source for those
13:27:36linuxstbI'm just starting a "codecmpc" - I just want to check that no-one else is working on it?
13:27:47nobbehand i want the latest versions offered, yes?
13:28:50Seedlinuxstb: what are your plans for it?
13:29:00preglowlinuxstb: musepack isn't realtime yet, btw
13:29:12stripwax_nobbeh yep
13:29:48linuxstbIt needs to be written at some point - is it very far from realtime?
13:36:43***Saving seen data "./dancer.seen"
13:37:27 Join xen` [0] (
13:38:04nobbehanother question: the last step of the instructions says to set the path variable to one of two things. which one do i want?
13:39:09stripwax_nobbeh - are you building for iriver or archos?
13:39:46stripwax_if you're building for both, add both: PATH=${PATH}:/opt/m68k/bin:/opt/sh1/bin
13:40:03stripwax_second one
13:40:07nobbehonly iriver
13:40:21nobbehwill i add that to the wiki then?
13:40:32nobbehto say what each is for?
13:40:45nobbehoh wait
13:40:47nobbehno need
13:40:51nobbehi see now
13:41:09stripwax_I guess, if it's not clear.. m68K == coldfire == iriver so if you installed the m68k-elf-... packages it should be clear
13:41:28nobbehi'll add a line clarifying that in the wiki page
13:41:43stripwax_but by all means update it, yep, to say "e.g. for iRiver builds"
13:46:19linuxstbAnyone know why Rockbox crashes if I add more than 16 file extensions in the filetypes array in tree.c?
13:48:38preglowsomeone really should make a codecs.config
13:49:03linuxstbpreglow: Yep.
13:50:27preglowi'll have a look at it later if i get extremely bored
13:50:40linuxstbIncreasing MAX_FILETYPES from 32 to 64 fixes it, but I don't know why.
13:50:53*nobbeh hopes for extreme-preglow-boredom :p
13:51:13preglowlinuxstb: remember that the array gets extended by viewers.config as well
13:51:23preglowlinuxstb: which has quite a few entries by now
13:51:25linuxstbAh, OK then.
13:51:36linuxstbI am happy with the world again.
13:54:26nobbehwhat text editors do you guys use in windows? i'm sick of notepad's lack of unix line-end-character-thing support
13:54:47crwlanything is better than notepad
13:55:27stripwax_wordpad? cygwin? :)
13:55:33stripwax_msdev at a pinch too!
13:56:18Seednobbeh: try the trial period of EditPlus
13:56:58linuxstbnobbeh: I know Windows people who use either textpad or jedit
13:58:00stripwax_nobbeh - now that you have cygwin installed, you could use nano, joe/jmacs or emacs
13:58:38ashridahnobbeh: putty onto my gateway + vim :)
13:59:00ashridahmanaged to hook a couple of my windows using friends onto gvim. that was fun :)
14:00:23Slasheri6h done and 65% battery left (mp3 battery test)
14:01:14ashridahSlasheri: i'm getting some stuttering during the second phase of buffering the playlist
14:01:36Slasheriashridah: Hmm, with mp3s?
14:02:02ashridahone mp3 in particular. havne't found any others that do it yet
14:02:11Slasheriok, what bitrate it is?
14:02:27ashridah192. other files with 192kbps seem fine tho
14:02:56SlasheriHmm. Could send that file? I think that some adjustments to the buffering should fix that
14:03:06crwlcan the dithering be turned off easily from codecmpa?
14:03:16ashridahyeah, it'll take me a while darned thing is 6.7 megs
14:03:48Slashericrwl: hmm, why would you like to do that?
14:04:00preglowSeed: know of any musepack devs that would love to port musepack to coldfire? :>
14:04:27crwlSlasheri, because i hear it sucks because the current implementation eats a lot of cpu power :)
14:04:47preglowSlasheri: bitrate?
14:05:03Slasheriah :) you can edit the source code if you wish, but there is currently no other method to turn it off
14:05:18Slasheripreglow: Hmm?
14:05:36 Join bumi [0] (
14:05:40bumihey everyone
14:05:52bumiwhats up?
14:06:02preglowSlasheri: for your mp3 testing
14:07:27Slasheripreglow: most of the files are 128, few of them 192
14:07:59Slasherishuffle is on and buggy crossfade also that might eat some cpu
14:08:46bumianyone here heard from gadi cohen?
14:08:53 Quit t0mas (Remote closed the connection)
14:08:57Slasherivolume was ~2 hours 100% while i was listening to the unit from HT system, now it's 50%
14:09:02bumianyone did any battery benchmark?
14:09:34 Join Lost-ash [0] (
14:10:02 Quit ashridah (Nick collision from services.)
14:10:05 Nick Lost-ash is now known as ashridah (
14:10:21Slasheribumi: initial mp3 benchmark is just going :)
14:10:36Seedpreglow: no
14:12:06Slasheripreglow: Hmm, that test will not be very accurate because now it's playing some 256k files
14:13:22preglowSlasheri: well, it's still nice to get a figure
14:13:56Slasheriyep :) i hope the test will finish this evening/night
14:13:57 Quit yngwi (Read error: 54 (Connection reset by peer))
14:15:19stripwax_i get an internal gcc compiler error building last nights source tarball. am I using wrong versions of gcc or something? gcc 3.4.4 (cygming special) and m68k-elf-gcc 3.3.5 fwiw
14:16:38preglowdon't use gcc 3.3
14:16:50preglowdumb will choke it
14:17:05 Join DaKi][er [0] (
14:17:11stripwax_preglow - arg. ok
14:17:37 Join yngwi [0] (
14:18:01SlasheriHmm, if the battery percent indicator is linear (and i strongly doubt it's not), the mp3 test would take ~17 hours
14:18:09stripwax_preglow - yep, i'm dumb :-) thx
14:18:28 Join LinusN [0] (
14:20:16crwli think that it would be almost hilarious if rockbox did, at the current state of optimization, better than the iriver firmware
14:20:28thegeekdoes anyone know if the sb live 5.1 with the kx drivers can give bitperfect output?
14:20:34preglowyes, it would, but don't expect that :P
14:20:36preglowhi, linus
14:20:41crwlyes, i'm not expecting :)
14:20:51crwlthegeek, i think the hardware always resamples to 48 kHz no matter what
14:20:58LinusNSlasheri: i alse get the stuttering on the second load
14:21:02thegeekI resample using foobar
14:21:15thegeekso it's already resampled before It is sent to the card
14:21:28thegeekif it can give bitperfect digital
14:21:33thegeekI am tempted to just use it
14:21:34SlasheriLinusN: ok, that might be caused due to too small file chunk size or too low filebuffer watermark level
14:21:39thegeeksince the kx drivers rock
14:22:02crwlthe hardware sucks if it doesn't
14:22:20ashridahLinusN: yeah, i'm uploading an mp3 that it occurs on atm.
14:22:21crwlbut then, it's sb live! :)
14:22:51thegeekas long as the output is bitperfect it does not really matter
14:22:57thegeeksome claim "jitter" can affect the sound
14:23:00preglowthegeek: you need to use 48khz for no internal processing
14:23:04thegeekI know
14:23:08thegeekI am doing that right now
14:23:12crwlthegeek, you could try playing some of those .wavs that actually contain dts or dolby digital data and see if your home theater amplifier (if any) recognizes the signal
14:23:19thegeekI know
14:23:32thegeekbut then I need a amp that has dts support
14:23:39thegeekthat means I would have to borrow my fathers;)
14:23:50thegeekand he's not too happy about that
14:23:56crwli have one, and i even had sb live once, but gave it away and never tried its digital out
14:23:59thegeeksince last time I did I held onto it for a bit too long;)
14:24:08thegeekoh well
14:24:16thegeekI guess I'll just have to test it myself;)
14:24:18crwlyou could suppose it's ok
14:24:23thegeekI could
14:24:39thegeekI'm going for the ultimage low budget hi-end headphone setup
14:24:41thegeekgood dac
14:24:44thegeekgood amp
14:24:51thegeekcheap digital bitperfect source
14:25:08preglowthegeek: i don't think you'd want to call the dacs on live very good
14:25:11crwlbuy a spdif <-> toslink converter box and record with iriver optical in and see what happens?
14:25:15thegeekI don't
14:25:17thegeekthats' the point
14:25:21thegeekI'm using the digital out
14:25:22crwloh, iriver didn't record 48 kHz, did it :)
14:25:23thegeekon the live
14:25:27thegeekand then using an external dac
14:25:40thegeekmhm crwl
14:25:42thegeekand then
14:25:51thegeekperhaps the iriver does something funky too
14:25:57*LinusN is impressed with the speed that new codecs are added
14:25:58thegeekI'm not about to trust the iriver fw;)
14:26:20preglowhaving linuxstb back does speed up things, yes ;)
14:26:43ashridahLinusN: heh. all it takes is for someone to make the first small hole in the dam
14:26:58ashridahthen it quickly breaks into a torrent
14:27:41preglowwe'll have musepack soon as well
14:27:42nobbehnow someone do .SID :P
14:27:46preglowbut it's not realtime yet
14:28:09nobbehi think the wiki says it is :S
14:28:14preglowno, it doesn't
14:28:23preglowit is realtime if i add this remaining opt
14:28:25nobbehoh, my bad
14:28:35nobbehwas thinking of wavpack
14:29:31preglowwavpack is by far the fastest codec we have, so yes, that's realtime ,)
14:29:35preglow;) <-
14:30:05linuxstbI've written codecmpc, but it doesn't work - the decoding fails immediately. But unless anyone objects, I'll add it to CVS so others can look at it.
14:30:14preglowdavid bryant's doing a great job on that one
14:30:49 Quit bumi ("( :: NoNameScript 3.81 :: )")
14:32:23preglowlinuxstb: sure, why not
14:32:52preglowi'll commit this last musepack thing so it at least has a modest chance of playing with some semblance of realtime speed
14:32:54LinusNdavid has cvs access now btw
14:33:03preglowLinusN: excellent
14:33:11preglowthen we'll probably see wavpack codec support soon
14:33:26linuxstbfiletypes.c currently uses #if CONFIG_HWCODEC == MASNONE to decide when to increase the number of extensions/filetypes - anyone have any better suggestions?
14:33:30preglowwavpack is by far the best lossless codec for portables
14:33:33preglowthanks to its hybrid mode
14:34:02LinusNlinuxstb: not at the moment
14:38:34 Part LinusN
14:38:50preglow101% realtime.....
14:41:10linuxstbOK, codecmpc is in CVS now.
14:45:56preglowdidn't audiobuffer_insert have a limit on 32kb?
14:46:14ashridahSlasheri: still around?
14:46:21linuxstbpreglow: I tried reducing the buffer to 4K, and it still failed.
14:47:13linuxstbBut I'm sure it's just something simple like that that's preventing it working.
14:54:00 Join edx [0] (
14:54:00stripwax_reckon i'll be ok charging my iriver from a 6v adapter instead of a 5v adapter? my other adapter is at the office and i never got round to making the usb charger :)
14:54:30crwli have used four AA batteries that should make up about 6 volts, at least when full, and that has worked ok
14:54:34crwlbut don't quote me on that
14:55:04preglowstripwax_: never ever play around with adapters
14:55:29preglowstripwax_: use the original one at all times
14:55:43crwland you also need to check the polarity
14:55:49crwlso yes, maybe don't play around with them :P
14:55:50preglowh1x0 will fry itself given half an excuse
14:56:14DaKi][erdepends how the charger circuit for the lithium polymer battery works
14:56:33stripwax_yeh, it's 6v/2a/correct polarity. seemed to work.
14:56:44crwli have also successfully used a 4.5V charger that came with iriver iMP-150
14:57:58stripwax_maybe i'll just cannibalise the charger and cut the plug off .. it's not a charger i need or use anyway
14:58:30preglowlinuxstb: have you checked out if logf says anything interesting while loading codecmpc, btw?
14:59:59linuxstbI don't know how to use logf...
15:00:57preglowthe codec loader uses it
15:01:08preglowjust enable it in your build
15:02:11linuxstbBuild-type Devel, Options L ?
15:04:10linuxstbWhere does the logf output go?
15:04:25preglowyou can also dump it to a file
15:04:50linuxstbI'll better go and find it then... How do I dump to a file?
15:05:59preglowdebug menu somewhere, methinks
15:06:02preglowive just tried the remote support
15:07:33*ashridah prods Slasheri
15:08:33ashridahDaKi][er: plugging in a 9V nokia phone charger has been known to nerf H1x0 players
15:10:31ashridahHCl: break, bugger, screw up. nerf.
15:10:47 Join webguest39 [0] (
15:10:59*HCl only knows nerf from something that pinky usually says from pinky and the brain o.o
15:11:14stripwax_nerf. chuckle. 9v clearly way too high.. still wondering if 6v would be ok. saw some threads on misticriver with folk using 4x1.5V batteries which seems ok, so I guess a 6v charger would be ok too?
15:11:34*ashridah just played a 16KHz mp3
15:11:36bipakis there anthing like a changelog for cvs? (not the webstatuspage) :)
15:11:44ashridahbahaha. chipmunk!
15:11:48stripwax_ashridah - hehehe
15:11:59webguest39the voltage limits for the h140 according to the IC specs are 4.5 ~ 6.5 volts
15:12:18stripwax_bipak you mean like the daily changelog file?
15:12:31webguest39center pin psoitive
15:12:39bipakstripwax_: thats it :>
15:12:56stripwax_webguest39 IC specs? what about the rest of the components
15:13:05stripwax_e.g. battery charging circuit
15:13:33webguest39look for yourself
15:14:10webguest39the IC feeds the battery according to the schematic
15:14:39yngwidoes anyone know if the battery indicator of the original firmware is linear??
15:14:43stripwax_webguest ah, didn't know which IC you were referring to
15:15:02webguest39texas Instruments power supply and control IC
15:15:45Slasheriashridah: here :)
15:16:07ashridahSlasheri:"> let me know when you've got it.
15:16:12DaKi][erhaving a stock value of 5v, enough headroom to run 6v should be designed into it for 'just in case' times
15:16:16Slasheriashridah: ok, thx :)
15:17:24Slasherinow mp3 test has gone over 7 hours and still 62% battery left
15:17:52crwlexpect it to drop to zero soon :P
15:18:01linuxstbHow long does the iRiver firmware play for?
15:18:12Slasherii haven't tested that
15:18:21preglow16 hours at 128kbps, they say
15:18:32crwli get maybe 8-10 hours with some track switching etc
15:18:40preglowyeah, me too
15:18:52crwlthough i mostly have oggs, it's maybe even less
15:19:25crwli think i've never run out of battery and i rechage almost always when there's about 50-75% left so i don't know exactly
15:19:29yngwii do a benchmark right now, been playing at 128kbps, vol 20 and eq normal for 4.5 hours and still all bars full
15:19:32crwlnever in normal use, i mean
15:20:00linuxstbCan I use logf from inside a codec plugin?
15:20:00bipakwould be great if rockbox gets longer playback time :)
15:20:21yngwibipack: i'm counting on it :-)
15:20:24crwlcurrently ogg vorbis is faster than mp3, at least according to my test
15:20:25SlasheriHmm, there has been few oggs too in the test (in one folder there was vorbis files and i didn't remembered that). mp3/vorbis switch caused some buffering problems but test has been running continuously without bigger troubles.
15:20:26bipakyngwi: :D
15:20:30preglownot very realistic
15:20:34crwlbut that's that dithering thing
15:27:07SlasheriInteresting. Now battery dropped from 62% -> 50%
15:27:15 Join lodesi [0] (~moi@
15:30:01 Join asdsd____ [0] (
15:30:08 Part asdsd____
15:30:32 Quit linuxstb ("Leaving")
15:30:51Slasheriashridah: ah, btw, i got it :)
15:31:23bipakhow is the battery status measured? with an ADC or some other way?
15:32:23preglowadc, yes
15:32:23Slasherii think it's with adc because there is no battery current measurement capability in the unit
15:32:37preglowdoesn't make any sense to measure current anyway
15:33:34bipakyeah :)
15:33:55Rorianyone know anything about ATX form factors on motherboards?
15:34:58bipakyes, google:
15:36:48***Saving seen data "./dancer.seen"
15:39:17preglowanyone got an ac3 file i can have?
15:40:28 Nick QT_ is now known as QT (as@area51.users.madwifi)
15:42:51 Join linuxstb [0] (
15:43:05Rorihmm. that does not tell me much. does anyone know if an ATX 2.2 PSU will work on a AMD Socket-A KT400 motherboard?
15:43:12linuxstbpreglow: Are you after an AC3 file?
15:44:36 Quit DomZ ("CGI:IRC (Ping timeout)")
15:45:13stripwax_Rori - yes
15:45:28preglowlinuxstb: yeah
15:45:43preglowlinuxstb: trying to see if a few quick opts in ac3 will make it a bit faster
15:45:43linuxstbCheck my ftp server (I think I emailed you the address the other day)
15:45:44 Join markun [0] (
15:45:52preglowthat mail is deleted, i'm afraid
15:45:55stripwax_Rori - unless that particular powersupply has *only* a P4-style power connector
15:45:57preglowjust msg me it
15:46:49stripwax_hm, does iriver still require patching for greyscale?
15:46:57*preglow produlates markun
15:47:05markunoops.. :)
15:47:37markunpreglow: I also have a white stripes and 2 aphex twin ac3, do you want them?
15:48:19linuxstbIf you use Linux, then "mplayer -dumpaudio" is your friend - run it on a DVD.
15:48:50markunlinuxstb: That's how I produced my test files.
15:49:07preglowmarkun: i'll have the aphex twin ones, please :P
15:49:12linuxstbmplayer never ceases to amaze me.
15:49:18Roriwell I have a Gigabyte KT400 K7 Triton motherboard with an AMD XP 2400+ on it. The PSU I am looking at is this one
15:50:35markunpreglow: They are music videos from a Spike Jonze DVD, so a bit different from the CD tracks.
15:51:11preglowsure you don't mean chris cunningham videos?
15:51:17stripwax_Rori - yeah, looks fine
15:51:19markunAh, yes :)
15:51:25yngwi now its 5 hours playback with orig. firmware and still all bars left... :-)
15:52:07markunyngwi: you put in a bigger battery?
15:52:49linuxstbpreglow: I think the musepack problem may be due to libmusepack wanting to seek in the file.
15:52:52bipakyngwi: i thought you are testing with rockbox
15:53:02stripwax_Rori - it has the old-style 20 pin connector you need
15:53:13 Quit Hadaka ("leaving")
15:53:16bipaktests with the orignal firmware are allready done :)
15:53:26yngwino, my player is new and i wanted to try if the battery works ok :-)
15:53:32bipakah *g*
15:53:53bipakseems like it does
15:53:54yngwiit seems it works even better then expected... until now
15:54:27yngwiI'm a little paranoid concerning batteries
15:54:41Roristripwax ok thx
15:54:45bipakyngwi: why?
15:55:01bipakyngwi: you dont believe iriver? ;))
15:55:02yngwii always have the feeling that the lifetime ist too short
15:55:20markunpreglow: Did you already write assembly versions of MUL, MUL_C and MUL_L for liba52?
15:55:26yngwihas nothing to do with iriver... its the same for every kind of device
15:55:45preglowmarkun: yes, did it now
15:56:04bipakyngwi: well i'm very happy with the lifetime of the ihp batterie
15:56:26yngwirockbox benchmark comes maybe tomorrow (isn't slasheri doing one right now??)
15:56:26dwihnoHow hard is it to get a replacement battery for the IHP-1xx-series? I am getting closer and closer to a purchase.
15:56:46yngwibipack: i guess, i will also be happy :-)
15:57:00bipaki hope so :)
15:57:04markundwihno: You can buy an ipod 1st of 2nd gen battery and change the polarity.
15:57:05ashridahdwihno: there are replacements for the ipod that fit in, and you can find li-poly ones that also fit
15:57:12preglowvery, very modest speedup
15:57:17ashridahebay has them, plus a few other suppliers
15:57:18preglowmakes me wonder if i've done something wrong
15:58:03dwihnoashridah, markun: so it won't be impossible to get "new" batteries in say, 1, 2, years?
15:58:29ashridahdwihno: nope.
15:58:47ashridahand you'd definently want a new one, not one that's been sitting on a shelf
15:59:18webguest39larger capacity "Ionity" batteries are pnp from dap store
15:59:18markunashridah: Does that matter much with li-poly?
15:59:27yngwii read, the ipod batteries work pretty good
15:59:46markunYes, I have a 2200 mAh now, works very nice.
15:59:52ashridahmarkun: so far as i know, li-poly degrade over time too
16:00:07yngwimarkun: does it fit in the case easy??
16:00:07ashridahi don't know if it's as fast as lithium ion
16:00:30markunyngwi: It's very tight, but it fits.
16:00:53markunmaybe 1 mm thicker.
16:01:21yngwithats ok... 1mm is not much
16:02:00crwlwhat about larger hard drives, are there any >20G 1,8" drives that would still fit into the H120 case
16:02:10crwl(the H140 is a few mm's thicker)
16:02:49yngwicrwl: i remember a few threads at misticriver on that topic
16:03:09preglowi did do something wrong, big boost
16:03:14yngwii think in a h140 someone put a 60GB HD
16:03:16preglowfrom 113% realtime to 140% realtime
16:03:42preglowit also outputs white noise :PPP
16:03:46markuncrwl: right now there is a 30GB drive that fits in the H120, later this year toshiba will release a 40GB.
16:03:57dwihnomarkun: is the playback time extreme with 22mah?
16:04:13markundwihno: 29 hours mp3 with the original firmware.
16:04:32preglowi am a fool
16:04:33markundwihno: But I did not do much testing.
16:04:36yngwiwow, thats amazing
16:04:40crwlmarkun, ok
16:04:44markun18 hours ogg
16:04:56Slasheri8 hours of playback and 50% battery left :)
16:05:04dwihnomarkun: quite impressive! I guess rockbox will extend that by some time
16:05:08stripwax_30GB fits in H120 case
16:05:31preglowshould i commit this?
16:05:34preglowit's just a quick hack
16:05:47bipakSlasheri: but the battery power doesnt go down lineary, right?
16:06:01Slasheribipak: that's right, i am not sure high linear it it
16:06:26crwlmarkun, quite expensive, maybe i'll wait :)
16:06:41bipakoh nice link :>
16:07:04CoCoLUSthose are outdated, though?
16:07:15markunCoCoLUS: Are they?
16:07:17 Join Naked [0] (
16:07:57CoCoLUSdon't know, thats why i'm asking :)
16:08:21CoCoLUSthe date is 27.3.
16:09:03markunI think the graphs are still valid, but you get a different scale if you draw less power.
16:09:27preglowwill we ever need to instatiate to copies of the same decoder?
16:09:30dwihnoWoo! Remote LCD!
16:09:34dwihnoDamn it looks cool!
16:10:07yngwidoes the remote use much power??
16:10:38stripwax_preglow - maybe for crossfading?
16:11:59preglowstripwax_: we never run two codecs simultaneously
16:12:16preglowstripwax_: we run first one until it's squeezed dry of data, then the other
16:12:47stripwax_preglow do we have a larger pcm buffer to do that? or is it a very quick crossfade?
16:13:07preglowmy, what a nice little codec
16:13:09preglow163% realtime
16:13:16stripwax_i should probably shut up anyway, what do i know :_/
16:13:17preglowstripwax_: crossfading is already done like that
16:13:27preglowstripwax_: it's one secondish
16:13:37stripwax_preglow cool
16:13:52preglowbut ok
16:14:00linuxstbpreglow: If you've optimised liba52, then please commit - I have one high bitrate 6-channel file that won't decode in realtime.
16:14:02preglowi'm removing a malloc in liba52 and declaring the main struct static
16:14:30preglowso we can't instantiate two liba52 at once
16:15:17ashridahSlasheri: hm, was that last patch to playback.c supposed to address the issue? it's still skipping here
16:16:35Slasheriashridah: No, i am unable to fix that issue before i can test it with the player (i can't do it until battery testing is finished)
16:16:48SlasheriI did the latest commit without real testing
16:17:48ashridahSlasheri: ah.
16:18:21preglowlinuxstb: ok, it's over 100% faster now
16:18:31preglowshould decode the new file just fine
16:18:32preglowi'll commit
16:19:39 Quit DaKi][er ()
16:23:53linuxstbI've found a couple of AC3 test files here -
16:24:13linuxstbThey also have DTS files - the AC3 files are the last two - labeled "DD" (Dolby Digital).
16:26:03linuxstbpreglow: Nice improvement. liba52 is a _very_ fast codec.
16:27:17preglow240% realtime
16:27:53linuxstbPlease commit...
16:30:00preglowstill working on it
16:30:36preglowthis is strange
16:30:44preglowi put something in iram and actually get worse performance
16:30:55preglowoh well, i'll leave it at this
16:32:18preglowa52towav has no iram support, that explains a lot
16:35:27preglowcould you add that to codeca52 in the meantime?
16:35:41 Join theebag [0] (
16:36:12linuxstbpreglow: I already use the IRAM for the only buffer.
16:36:28linuxstbSorry, I don't - there's another one I can move.
16:36:52 Join DomZ [0] (
16:37:22theebagFirst I just want to say that Rockbox is really amazing
16:37:54theebagI just installed it for the first time,
16:39:15preglowlinuxstb: i use 21500 bytes of iram now
16:39:27preglowlinuxstb: that's about every bloody table i could find
16:40:27theebagIs the peak meter working at the moment ? ( Firmware 20050611 )
16:40:38linuxstbpreglow: codea52.c is now using 4864 bytes of iram.
16:41:11preglowtheebag: nope
16:41:14preglowtheebag: not for ivier
16:41:19preglowiriver, even
16:41:23linuxstbpreglow: yes.
16:41:47theebagFound a small bug ( probably know already )
16:42:26linuxstbWhat's the bug?
16:42:40theebagWhen listening, if you press the stop button, the sound fades away but when it enters the tracklist,
16:43:06theebagyou hear the song again for a brief moment
16:43:10theebagjust a pop
16:43:16preglowtheebag: known bug
16:45:08theebagI've just used it for 5 minutes and I allready like it
16:45:56preglowlinuxstb: commited
16:46:48linuxstbpreglow: I'm listening already :-)
16:47:20linuxstbIt's fixed my stuttering track - 448kbps 5-channel AC3.
16:47:24preglowdidn't need more improvement than that, no?
16:47:43preglowwell, 150% performance improvement doesn't hurt
16:47:54linuxstbThe CPU is boosted on my 5-channel track....
16:48:39linuxstbBut it's not boosted all of the time :-).
16:49:12linuxstbSo yes, I think that's all we will ever need.
16:49:22linuxstbIt will be hard to get all the other codecs that fast.
16:49:24preglowdoes it decode all five channels before mixing down?
16:49:40linuxstbI think so, but the downmixing happens inside liba52.
16:50:03preglowyeah, i know
16:50:43preglowsomeone should try djbfft + sorting to speed up the imdct in tremor
16:50:55SlasheriHmm, that ac3 passthrough via toslink would be nice feature but perhaps it's impossible to do due to hardware limitations?
16:51:26preglowdon't think it's possible no, but i'm just guessing
16:52:31preglowbut shop, later
16:53:32thegeekwhat happened to that midi codec?
17:03:24preglowstevenmn is busy with other stuff, i guess
17:05:53preglowlinuxstb: please do check if you can hear any quality differences, btw, it should also be more accurate now, but i'm not sure
17:06:04webguest39is lame mp3 gapless ?
17:07:05preglownot yet
17:07:50*Rori waits with bated breath
17:08:04Roridon't mention gapless when I am around :)
17:08:09webguest39will it be gapless ?
17:09:12webguest39otherwise I'll have to transcode my 180 gigs of mp3 to ogg
17:09:28preglowit will be
17:09:31stripwax_webguest - no point transcoding!
17:09:45preglowtranscoding wont get you any closer to gapless
17:09:49preglowor wait, it will!
17:09:53preglowbut no
17:09:57preglowit will be gapless
17:10:09webguest39I get the picture preglow
17:10:14stripwax_:-) it might be gapless, if it's rtr
17:10:30linuxstbpreglow: It's hard to judge the quality with 48KHz files, but yes, they did seem better.
17:10:31stripwax_^transcoded using something that understands the lame tags
17:11:26 Join phil- [0] (
17:11:50preglowlinuxstb: as long as they're not clearly worse i'll add a 'hooray!' to that
17:12:05 Join Sucka [0] (
17:12:11phil-Hi all I would like to point out if i may, this topic on the RockBox forum:
17:12:16linuxstbpreglow: No, they are definitely not worse.
17:12:19phil-looking for feedback
17:15:08preglowphil-: does it patch the bootloader itself?
17:15:23preglowphil-: patch the firmware, i mean
17:15:42phil-it comes with a patched bootloader, there are two versions for h120 and h140 at this early stage
17:15:50phil-the file size is 1.95 meg in total
17:16:23preglowdistributing a patcher firmware is really shady
17:16:30preglowiriver has copyright on the firmware
17:16:41*preglow sighs and corrects himself: 'patched'
17:16:50ashridahyou realise, of course, the only reason rockbox themselves don't distribute a pre-patched firmware is because they don't have distribution rights, don't you?
17:16:56phil-thats a point which i forgot to consider until now
17:17:00phil-I remember reading about it
17:17:05preglowthat's the entire reason why i made fwpatcher
17:17:12preglowwhy can't distribute a ready patched firmware
17:17:15ashridahwhich is why someone made the fwpatcher
17:17:23preglowgod what is wrong with my brain
17:17:30preglowit's not compatible with writing
17:17:36phil-well I could easily modify it, to run your patcher instead
17:17:47preglowwell, sure, but my patcher also is a gui tool ;)
17:17:54preglowso you need to wrap it somehow
17:18:02Suckayou could make your prog automatically download the firmware and then patch it?
17:18:24phil-i can do that, ive got it to autodownload files and run them
17:18:43phil-so let me get this right for myself
17:18:45phil-if u dont mind
17:18:59phil-i get it to download the firmware, then run fwpatcher?
17:19:12phil-if you wouldnt mind me distrubuting it with the software preglow
17:19:34preglowyou can do whatever you like with it
17:19:44preglowbut it's got its own gui, so it wont be as automatic as you like
17:20:07preglowwhat do you program in?
17:20:43phil-i was thinking, i could include it but the user would have to find the .hex file themselves using the software
17:20:55preglowyes, exactly
17:21:02preglowbat what language are you programming in?
17:21:12preglowif you use C or C++, you can just steal my patcher routine
17:21:23preglowfor other languages i have no idea
17:21:25phil-at the moment im using nsis to basically make it all, its very versitile as I lack the skills for C/C++ yet
17:21:48phil-the shame :(
17:22:12preglowbut yeah, please don't distribute ready patched firmwares, that could put us in a bad light
17:22:25preglowand reveal us like the nasty buggers we are
17:22:25phil-yes that was my own fault, sorry
17:22:44stripwax_maybe i'll tweak fwpatcher to accept cmdline args?
17:23:02phil-that could work?
17:23:17phil-because if i download the firmware to the same directroy as where fwpatcher isntalls
17:23:31phil-im sure it would be quite easy to do it automatically
17:24:17phil-so do you belive the concept of what im doing is worthwile? because i found all the relveant info for rockbox on the site eventually, but it could be quite daunting
17:24:21phil-for a new user
17:24:49stripwax_i quite like the idea, especially if it's automatically downloading the latest versions of everything.
17:24:53preglowSlasheri: around?
17:24:57Slasheripreglow: yes
17:25:13phil-yeah it does work, i have a finshed version here
17:25:18phil-just without documentation at this time
17:25:22preglowphil-: if enjoy what you're doing, just go ahead, if it's usable, someone will use it
17:25:33preglowSlasheri: how do you do mp3 gapless at the moment?
17:25:37phil-thanks preglow
17:25:57preglowyou can use fwpatcher for whatever you like
17:26:14Slasheripreglow: i am not really sure.. :) i look for some headers and then cut all silence at beginning of tracks
17:26:45Slasherii hope someone will do a better gapless handling for mp3
17:27:04preglowSlasheri: i'm looking at foobar2000 mp3 source code at the moment, and it seems to do two things:
17:27:19Slasherii haven't managed to implement lame gapless right way at the moment
17:27:34preglowskip an amount equal to the delay field in the lame info header plus a constant encoder delay amount at the start
17:27:50linuxstbI think the first thing we need to do is to switch to using samples in elapsed_time and "length". All the codecs work at the sample-level, so we lose a lot by converting to ms.
17:28:17preglowand cut an amount equal to the padding amount specified in the lame header minus the constant encoder delay
17:28:22SlasheriHmm. Do you know how to parse those fields from mp3 file? I guess that current id3 structure has not all of those fields
17:29:17 Quit Sucka ("a bird in the bush is worth two in your house")
17:29:17preglowit's not in id3
17:29:27preglowi thought you had made a lame info header parser?
17:29:33preglowor do i remember incorrectly?
17:29:34Slasherithere is also mp3info structure, it has at least padding
17:29:43Slasherii haven't :)
17:29:45preglowwhere in the source is the reader?
17:29:55preglowfor mp3info
17:30:01Slasheriin playback.c
17:31:03phil-preglow: is it possible if you have any free time to make a new fwpatcher which supports commands? would be grateful
17:31:14preglowphil-: sorry, i'm out of rockbox time for today
17:31:20preglowphil-: and i'm not in windows anyway, can't test it
17:31:46phil-ahh i see sorry to bother
17:31:48preglowi think stripwax said he could do it ;)
17:31:58phil-sorry to bother you*
17:32:03Slasheripreglow: and that mp3info structure can be found from firmware/export/mp3data.h
17:32:34preglowi see
17:32:42preglowit's just skipped at the moment
17:34:03stripwax_preglow/phil :-) sure, if I get a minute. I can probably hack it, actually...
17:34:17stripwax_must get food
17:36:50***Saving seen data "./dancer.seen"
17:37:07phil-sorry can i just clarify something, can i get the program to download an already patched bootloader from my webserver for example or is that still wrong?
17:37:48 Join Twc_12 [0] (
17:38:03Twc_12can you help me please ?
17:38:19preglowphil-: still wrong
17:38:33phil-so it has to be patched on the users computer for it to be right?
17:38:36preglowphil-: you need to download an ordinary firmware from iriver, then patch that with the bootloader
17:38:47phil-right thank you
17:38:48Twc_12somebody speak french here ?
17:39:07RoriI only speak gibberish
17:39:30Twc_12okay ... can you say me if the firmware of rockbox has a synthetical voice ?
17:39:48 Join bagawk [0] (~Lee@bagawk.user)
17:39:51Twc_12I'm blind and I search for a synthetical voice for iriver
17:39:59preglowno, it has not got that
17:40:03preglowit might in the future
17:40:45preglowthere is a lot of work ongoing in the area of sound right now
17:40:55bagawkIt should be somewhat easy now, concidering they already can decode mp3s
17:41:13bagawk(and play)
17:41:23preglowbagawk: on iriver we might even be able to use a complete voice synthesizer
17:41:40bagawkThat would be neat :)
17:42:03preglowyes it would
17:42:42ashridahTwc_12: unfortunately, the heavy development at the moment means that support for voice menus hasn't been addressed yet. i imagine it will get done when most of the basic audio playback features are stabilised and thoroughly debugged, tho. From what i know, it's an important topic for the rockbox developers.
17:43:40Twc_12okay ... thank you very much
17:44:07Twc_12thank you for all informations ! bye
17:44:15 Part Twc_12
17:44:50preglowthe potential for speaking menus is even greater on iriver
17:44:53 Quit yngwi ("Chatzilla 0.9.68a [Firefox 1.0.4/20050511]")
17:44:59preglowmight use different codecs, synthesizers, etc
17:47:40preglowi'll code a quick enc_delay and enc_padding extraction
17:47:58 Quit DomZ (
17:47:58 Quit webguest39 (
17:47:58 Quit xen` (
17:47:58 Quit MoosCamaro (
17:47:58 Quit niobos_study (
17:47:58 Quit Harpy (
17:47:58 Quit silencer (
17:47:58 Quit thegeek (
17:47:58 Quit bipak (
17:47:58 Quit elinenbe (
17:47:58 Quit Rick (
17:47:58 Quit C-Keen (
17:47:58 Quit bill20r3 (
17:47:59 Quit nobbeh (
17:47:59 Quit Bagder (
17:49:54 Quit phil- ()
17:50:12NJoinDomZ [0] (
17:50:12NJoinxen` [0] (
17:50:12NJoinMoosCamaro [0] (
17:50:12NJoinniobos_study [0] (
17:50:12NJoinbipak [0] (
17:50:12NJoinHarpy [0] (
17:50:12NJoinsilencer [0] (
17:50:12NJointhegeek [0] (
17:50:12NJoinelinenbe [0] (~elinenbe_@
17:50:12 Join Rick [0] (rick@Rick.user)
17:50:12NJoinC-Keen [0] (
17:50:12NJoinbill20r3 [0] (
17:50:12NJoinnobbeh [0] (
17:50:12NJoinBagder [0] (
17:52:05 Quit ashridah ("Leaving")
17:52:05 Quit Bagder (
17:52:05 Quit nobbeh (
17:52:05 Quit bill20r3 (
17:52:05 Quit elinenbe (
17:52:05 Quit silencer (
17:52:05 Quit niobos_study (
17:52:05 Quit MoosCamaro (
17:52:05 Quit xen` (
17:52:05 Quit DomZ (
17:52:05 Quit C-Keen (
17:52:05 Quit bipak (
17:52:05 Quit Harpy (
17:52:08 Quit Rick (
17:52:08 Quit thegeek (
17:53:49NJoinDomZ [0] (
17:53:49NJoinxen` [0] (
17:53:49NJoinMoosCamaro [0] (
17:53:49NJoinniobos_study [0] (
17:53:49NJoinbipak [0] (
17:53:49NJoinHarpy [0] (
17:53:49NJoinsilencer [0] (
17:53:49NJointhegeek [0] (
17:53:49NJoinelinenbe [0] (~elinenbe_@
17:53:49NJoinRick [0] (rick@Rick.user)
17:53:49NJoinC-Keen [0] (
17:53:49NJoinbill20r3 [0] (
17:53:49NJoinnobbeh [0] (
17:53:49NJoinBagder [0] (
17:54:48 Join Sucka [0] (
17:55:15 Quit markun (
17:59:20bagawknet split...
18:10:23 Quit theebag ()
18:10:57stripwax_strange startup problem on iriver. when it turn on it asks me "resume? play for yes" etc. I press play, the harddrive spins up and down and stops. I press play again, and then things get going properly. is it meant to be like that?
18:11:43RoriIt asks resume on everything I noticed
18:11:55Rorieven if you had previously stopped playback before shutdown
18:12:00crwlyou can turn it off
18:12:11RoriI know but it should be a little more intelligent
18:13:06 Join markun [0] (
18:14:45markunStrange.. I can only connect to freenet using ipv6
18:15:12preglowSlasheri: ok, i'm able to fish out padding and delay
18:15:21preglowSlasheri: where do i do the cutting and slicing? :P
18:16:12markunstripwax_: Try pressing the joystick instead of play.
18:16:58Rorihas anyone agreed on button functions yet? especially for the remote
18:17:00stripwax_markun - to turn it on??
18:17:12markunstripwax_: to resume
18:17:22Slasheripreglow: Oh, that's great :)
18:17:28Slasheripreglow: you should do it in codecmpa.c
18:17:45stripwax_markun - ah-hah! i kinda assumed play meant the button marked Play :-)
18:17:49Slasheribattery test still running, 13% left. I think it will fail soon
18:18:07Roribattery test on what format?
18:18:19markunSlasheri: Howmany hours till now?
18:18:23linuxstbpreglow: I don't know if it's important, but the "length" field (in ms) may need adjusting. I don't know if it's used for anything that needs that amount of accuracy though.
18:18:41Slasherimarkun: over 10h
18:19:10Slasheribut i think that test was not very reliable because some ogg files was played
18:19:43 Join elinenbe_ [0] (
18:19:46crwldid you try that queue thing already? ;)
18:19:55Slashericurrently oggs might lock up the cpu_boost because elapsed information is still missing
18:20:03Slashericrwl: i try it soon :)
18:20:50Slasheriyep, vorbis made the test very unreliable..
18:21:12linuxstbSlasheri: I'm working on Ogg metadata handling now. Still can't work out the total duration of the track yet though.
18:21:23Slasherii think battery should last longer with pure mp3 files only
18:21:30Slasherilinuxstb: good :)
18:21:49linuxstbWe should move this stuff out of playback.c though.
18:22:10Slasheriyes we should
18:22:35linuxstbMaybe something like apps/codecs/metadata.c ?
18:22:38Slasheri3% battery left and still playing ;)
18:22:40preglowSlasheri: can i read the mp3info struct from codecmpa?
18:22:45Slasherilinuxstb: Hmm, maybe
18:22:46crwli feel that battery should last longer with pure ogg files :)
18:23:15Slashericrwl: Currently it wont if crossfade is being used
18:23:28Slasherihehe, 0% battery :D
18:23:40preglowhow long has it been going?
18:23:50Slasheri10 hours, 25 minutes
18:24:02preglowand mostly 128kbps?
18:24:08Slasheribut it has played many oggs with cpu full boosted all the time
18:24:43Slasherisomebody should do a better test.. Maybe i will do someday :)
18:24:44preglowi wonder if the hardware keeps us from draining the battery too far
18:24:51stripwax_linuxtsb - does the source for ogginfo help to figure out total duration?
18:24:54crwlSlasheri, i don't use any damn crossfade :)
18:25:18preglowdraing a li-poly battery too far will reduce it's lifetime
18:25:38preglowSlasheri: but, yo, can i read the mp3info struct from codecmpa?
18:26:00Slasheripreglow: yes, just check the codec_api structure
18:26:12nobbehpreglow: if it runs down fully in usb mode, it WONT start at all after. Most stuff will power on for a few seconds...
18:26:28linuxstbpreglow: Do you mean ci->mp3data ?
18:26:36Slasheripreglow: that battery draining is not a problem because i will order ionity battery as soon as 1900 mAh model will be available
18:26:47linuxstbCheck apps/playback.h for the ci structure.
18:26:54linuxstb(codec_api structure)
18:26:58preglowSlasheri: i'm meaning generally, if the hardware doesn't keep us from drawing too much, we have to restrict it
18:27:07Slasheripreglow: ah, that's true
18:27:13Slasherihardware will drain it too empty
18:27:28preglowa li-poly cell should never be drawn below 3 volts
18:27:33preglowat that point the lifetime is reduced
18:27:57stripwax_ionity 1900 mAh already available, no?
18:28:17preglowbut i guess it never drops that far
18:28:49HCli have 2300mah in my iriver
18:29:02nobbehi have 10000000mah in my iriver!
18:29:02thegeekwhere did you buy it?
18:29:13markunHCl: hi!
18:29:16thegeekand does it fit in the h120?
18:29:22HClhey markun
18:29:32HClthegeek: ask markun, and ebay
18:29:33stripwax_thegeek - sell the 1900 mAh . eBay sell the 2100 mAh :-)
18:29:55stripwax_HCI - 2300???
18:30:27Slasheripreglow: battery monitor says it's 3.13V currently (0%)
18:32:12preglowi wonder if the reading is accurate
18:32:15preglowi bet linus knows that
18:32:21Slasheripreglow: 2.5V, hdd failed to spin up and then hardware turned off
18:32:56HClyes. 2300
18:33:00HClebay sells 2300 too.
18:33:56stripwax_hm.. where is bootloader/config.h ?
18:36:13 Join webguest65 [0] (
18:36:54markunIn config-h100.h: #define BATTERY_SCALE_FACTOR 16665 /* FIX: this value is picked at random */
18:37:35stripwax_hm, also, which cygwin package installs windres?
18:37:49markunSlasheri: so I don't think the 3.13V was accurate.
18:38:35Slasherimarkun: ah, ok
18:40:21preglowSlasheri: is there a gapless attempt in codecmpa.c that i should be aware of?
18:40:46Slasheripreglow: not really. Just remove everything unnecessary stuff
18:41:12Slasheriand you can also comment out the detect silence hack if it causes troubles
18:41:59preglowso you actually use id3 tag to determine song length?
18:42:08Slasheriyes :D
18:42:22Slasherii didn't find a better field to use
18:43:15preglowhere goes
18:43:20*preglow dives int code
18:45:00SlasheriHmm, maybe i will order now the ionity battery.. I hope they get them in stock soon
18:47:09crwlhow much they cost?
18:47:49 Join ShockerEngr [0] (
18:47:50stripwax_you can get 2200 mah (not ionity) for £7 from ebay.
18:49:29Slashericrwl: 1900 mAh ionity 39e. Quite much but it does include battery protection module and should fit in iriver well
18:50:14crwlincluding shipping?
18:50:22Slasherii am afraid not :)
18:50:58markunThe 2200 is EUR 13,70 including shipping
18:51:05 Join Shagnar [0] (
18:51:10crwlwhere from?
18:51:13Slasherimarkun: Hmm, do you have any specs available?
18:51:14crwland does it fit h120 too
18:51:26markunhong kong. no, don't have any specs.
18:52:23markunI would have to open up the iriver again to read what's written on the battery, but I don't have the right srewdriver here
18:52:29crwlthere's nothing wrong with my battery, though...
18:52:36crwlbut i could order a spare one :P
18:52:41Shagnarwhat is liba52 exactly?
18:52:48preglowShagnar: ac3
18:52:51crwlis the iriver easy to open and the battery easy to change?
18:52:56preglowSlasheri: there are still bufferin bugs :PP
18:53:21Slasheripreglow: Hmm, what kind of bugs?
18:53:35crwlthat queue thing at least ;D
18:53:57 Quit bagawk (Read error: 110 (Connection timed out))
18:54:28Slasheripreglow: You mean the buffer skipping bug at some bitrate files? I will fix that now :)
18:54:39preglowit just skipped an entire song here
18:54:45Slasherireally :D
18:54:46preglowhere's what i did
18:54:54preglowi loaded a track, then fast forwarded to the end
18:54:57preglowthen let the track change
18:55:05Slasheriah, seeking bug
18:55:09preglowthen pressed back until i was at the previous track
18:55:24preglowthen, when the track was supposed to change again, it just skipped right into track three
18:55:29stripwax_crwl yep. go search on . H-series mods section
18:55:36stripwax_you'll need a torx T-5 screwdriver
18:55:39crwli don't like searching misticriver :D
18:55:40preglownot the start of the track or anything
18:55:46preglowmore like into the middle of track three
18:56:19stripwax_crwl -
18:58:03crwlARGH, is there a way to disable all those nasty signature pictures and avatars and in fact everything except the text
18:58:32preglowSlasheri: is the sample count accurate across a seek? doesn't look like it to me
18:58:51preglowcrwl: yes, i believe most forums have a stripped down version meant for google, heh
18:59:00Slasheripreglow: there is at least one known bug with seeking, i will try to fix it soon
18:59:10crwlmisticriver is definitely one of the worst i know :|
18:59:20crwlmaybe the worst one that still has some valuable information in it
19:01:57Slasheripreglow: you better not to use the seeking at all while testing, it's very buggy at the moment :D
19:04:01 Quit Bger (Read error: 113 (No route to host))
19:04:15preglowSlasheri: i've noticed
19:04:16 Join Bger [0] (~Bager@
19:04:37preglowthe bloody track start clicking annoys the hell out of me
19:04:49preglowdid you try clearing the dac when the user changes track?
19:05:18Slasherinot yet
19:05:38Slasherii will do that
19:05:59preglowthere's still a click
19:06:03preglowwhen changing tracks
19:06:48 Quit silencer (Nick collision from services.)
19:06:49 Join silencer [0] (
19:06:53 Quit silencer (Nick collision from services.)
19:08:01preglowthink i'll just make me a couple of test files
19:08:08preglowi can't stand waiting a minute to see if my code works
19:08:41markunpreglow: you get a click even after putting silence through the dac?
19:10:12preglowno, not that
19:10:17preglowi'm trying to code proper lame gapless
19:10:24markunah, ok
19:11:49 Join silencer_ [0] (
19:11:54markunpreglow: any idea how the dac and headphone amp can be powered on quietly?
19:13:39preglowi can't hear a gap in my test files :/
19:13:48preglowmarkun: no idea at all
19:13:57preglowmarkun: not even the iriver firmware manages that completely
19:16:04markunpreglow: no, but a lot quieter than we do
19:21:13markunpreglow: I think I will just commit the center-scrolling patch. I will change the font-changing bug later (or someone else can have a go at it)
19:21:40 Join postglow [0] (
19:21:49preglowthere he is again!
19:21:56markunthe evil twin!
19:23:19preglowI00 at 40102700
19:26:56preglowgood god
19:27:00preglowdon't let the remote touch the main unit
19:27:04preglowi nearly burst my ears
19:28:15stripwax_seems we init the uda1380 chip with everything turned on at once. might be better to turn on e.g. dac and bias before headphone amp
19:29:29preglowoh yes
19:31:49 Quit postglow ("CGI:IRC")
19:31:51stripwax_in fact, turn the chip on and wait a bit before powering up the individual circuits (and yeah, probably leave the amp til llast)
19:33:46preglowanyone know of anything that read .pcm in linux?
19:34:04HCl /dev/dsp
19:35:03preglowahh, i need to see the data
19:35:08preglownevermind, xxd will do
19:35:41Slasheripreglow: oh, it still crashed? :(
19:36:05Slasheriwhat did you do?
19:36:52***Saving seen data "./dancer.seen"
19:37:13preglowturned on DEBUG_GAPLESS
19:37:17preglowi'm missing _ONE_ samples
19:37:19preglowit seems
19:37:33Slasheriah, so that caused it :)
19:37:48markunok, center-scolling commited
19:38:55CoCoLUS<fn~preglow> i wonder if the hardware keeps us from draining the battery too far
19:39:21CoCoLUSif that would be the case, why would the original firmware include such a do-not-start-when-battery-is-low feature?
19:40:07amiconnstripwax_: Section 8.12 in the uda1380 datasheet might be helpful
19:40:12preglowCoCoLUS: exactly
19:40:54CoCoLUSi think that topic is quite important... imagine the misticriver posts: "do not use rockbox!!! it kills your battery!!!" ;)
19:45:37stripwax_amiconn - oh, good find
19:58:10]RowaN["do-not-start-when-battery-is-low" .. at times for me, it seems like the battery is too low for the original firmware to start (hangs on "starting original firmware screen") yet i can boot into rockbox with no probs
19:58:50]RowaN[oh, but you knew that
19:59:33linuxstbIt seems that Ogg doesn't actually store the total length of the track anywhere - you need to seek around the file and calculate it...
20:00:32linuxstbI thought Ogg was supposed to be an improvement on MP3.
20:01:32nobbehits not probably designed primarily for streaming
20:02:12preglowshould i comment away padding and delay reading for archoses?
20:02:35preglowand don't expect they'll ever need it
20:10:59 Part stripwax_
20:11:07 Join yngwi [0] (
20:24:52 Join Nuxator [0] (
20:25:17Nuxatori found a new bug
20:25:30Nuxatorif you click one song to play it it plays
20:26:09Nuxatorthen click again to choose another file :buffering start but play stop and then player return to file list
20:26:40Nuxatorcan someone test to see if it's confirmed
20:26:51linuxstbAre you using the very latest bleeding edge build?
20:27:26Nuxator19h41 version
20:27:36SlasheriThat problem is fixed in the bleeding edge build
20:27:55preglowif you're going to report a bug on audio playback, always download the latest build first
20:28:07preglowthings are moving fast
20:28:12Nuxatorbut i dl it 30 min ago
20:28:39Slasherihmm, bleeding edge?
20:29:08Nuxatorjust a minute i'll test the lates one
20:29:08Slasheriinteresting.. it should not have that bug anymore
20:29:12linuxstbI can't see a 19h41 version on the website
20:29:18preglowSlasheri: it doesn't matte rhow big the buffers i pass to audiobuf_insert are, no?
20:29:45Slasheripreglow: no, just try to keep them below 32 kB
20:29:51preglowyeah, sure
20:29:59preglowi've almost got gapless playback here
20:30:06preglowbut something is just a wee bit off
20:30:06Slasherioh, great :)
20:30:53preglowi wish audacity could read raw pcm
20:31:05Slasheritry hexedit :)
20:31:06CoCoLUSwhich codec? ac3?
20:31:37Nuxatorok bug was fixed
20:31:40linuxstbpreglow: Just use "sox"
20:32:17preglowi also wish audacity was actually usable
20:32:19preglowgod, how i hate it
20:32:21preglowi loathe it
20:33:13linuxstbDoes anyone know if any of the Windows audio editors work with Wine? I tried installing Adobe Audition, but it wouldn't work.
20:33:37linuxstbI couldn't get cdwave to work either.
20:34:30preglowaudition works poorly in wine
20:34:35preglowoh, how i misses it
20:35:42Nuxatorok the bug is still here
20:35:46Nuxatormore tricky
20:36:09Nuxatoryou have to change song before hd spin down from first 2mo buffering
20:38:12linuxstbWhat kind of files are you using? I can't get it to crash with FLAC or Vorbis
20:38:29Nuxatori'll try with ogg
20:38:37Nuxatorbut it doesn't occur all time
20:38:48Nuxatorhave to click on several track
20:39:06Nuxatorbut when it occurs it's for each track until reboot
20:39:36preglowhmm, the start of the waveform wobbles like hell
20:39:44preglowthat's to be expected, i suppose
20:40:30Nuxatorgot it too with oggs
20:41:18Nuxatormaybe initial bffering when interupted doen't free mem
20:41:41Nuxatorand if you switch several tracks it shows this bug
20:41:58preglowit actually looks like i have managed gapless playback
20:44:07linuxstbpreglow: Congratulations.
20:44:26preglowa bit too early
20:44:37NuxatorSO if you change tracks lie a crazy (by clicking on files in file tree) you'll finish to see the bug
20:45:05 Quit lodesi (Read error: 110 (Connection timed out))
20:45:47preglowthere's still a little bloody click here
20:46:14preglowbut i can't imagine why
20:46:31 Join lodesi [0] (
20:48:39 Join Stryke` [0] (
20:49:15preglowbut i really need to stop coding rockbox now
20:49:20preglowso it'll have to lie
20:51:05preglowSlasheri: when i pasted first.pcm and second.pcm together in audacity, it sounded a bit smoother than the track change did on rockbox
20:52:47Slasherimaybe audacity did some filtering
20:52:52preglowsincerely doubt it
20:53:00preglowthere's still a glitch
20:53:07preglowbut i believe that's from the distortion in the waveform
20:53:40preglowbut ogg vorbis transitions are completely smooth
20:53:45preglowso can't be rockbox code
20:53:47Slasheripreglow: You should also try not to reinitialize the codec between track-change
20:54:04Slasheri(put next_track: label after all initialization stuff)
20:54:22preglowdoes this matter mpeg audio?
20:54:24preglowok, i'll try
20:54:31Slasherii don't know, might be
20:56:23 Quit nobbeh (Read error: 54 (Connection reset by peer))
20:56:40preglowstill a tiny glitch
20:59:14preglowsounds like there's a tiny impulse or something
21:02:43]RowaN[someone please tell me that url that contains wps files, i want to have a go at making one
21:04:57 Quit Nuxator (Read error: 145 (Connection timed out))
21:08:00]RowaN[not possible to centre lines in the wps?
21:08:30yngwiSorry, I don't know
21:09:41Stryke`not possible, yet
21:10:30bill2or3wps ="whats playing scren" ?
21:10:34bill2or3err, screen
21:11:00markunwhen a scrolling line is bigger than 'scroll limit' it keeps on scrolling. Shouldn't it wait 'scroll start delay' ms when it's back at the beginning of the line?
21:13:06yngwiwps= while playing screen
21:15:00bill2or3sweet, down to 9
21:15:55CoCoLUShm why wont next artist/title work for me when it obviously worked for those in the gallery? :)
21:16:11 Join LinusN [0] (
21:16:24LinusNhere we go again...sigh:
21:17:46LinusNit's funny how clueless people can be
21:17:56preglowLinusN: i've hacked mp3data.c to read the lame delay and padding fields, should i define that out for MAS players?
21:18:19LinusNnah, don't bother
21:18:37preglowi've _ALMOST_ got gapless mp3
21:18:56preglowthere some sample lying around unaccounted for
21:19:15LinusNmarkun: the scroll delay is only for the initial display of the text
21:19:22preglowbut no i'm shit tired of watching this screen, so going for a walk
21:19:26preglownow, even
21:19:56yngwicocolus, i also have sometimes problems with next artist /title, but it seems its rather a problem with buffering or so (?) because if I wait long enough (somtimes more than a half song) the next playing usually shows up
21:21:08preglowagain, listening to music using rockbox
21:21:25preglowLinusN: if i ever received that email, i'd frame it
21:21:31yngwiPreglow: have fun and enjoy your success
21:21:45preglowi'll be enjoying slasheri's success ;)
21:21:48markunLinusN: When in bidirectional mode it waits scroll_delay * 2 after reaching the end or the beginning
21:22:15Slasheripreglow: btw, i just probably found some pcm buffer problem. Fixing that now :)
21:22:44CoCoLUSslasheri, you stopped the buffering of the next track from happening until 20 seconds of the first file are played, right?
21:22:46Slasherihmm, maybe not..
21:22:53preglowSlasheri: anything that might benefit gapless? :>
21:23:00SlasheriCoCoLUS: no, after 2 MiB are buffered
21:23:07Slasheripreglow: i doubt so :)
21:23:13preglowoh well
21:23:17preglowi'm gone for a while
21:23:29CoCoLUSmaybe that's the reason for the next artist/track info not being available?
21:24:26SlasheriCoCoLUS: yes, that's the reason
21:24:39Slasheriwill be fixed later :)
21:24:42CoCoLUSi see
21:25:22CoCoLUShm how do you think you'll fix it? either the next file is loaded or it's not, so how would it be possible to retrieve the next-info...
21:26:51Slasherimetadata of next track can be loaded without buffering the track itself
21:27:01linuxstbCoCoLUS: The meta-data can be read from the file directly from disk, without it being buffered first. I don't know if that's how we'll do it though.
21:27:28linuxstbWe are thinking alike then...
21:27:51linuxstbI thought that was what was happening now.
21:29:06linuxstbOgg files are a pain to parse though. I may need to borrow some code from Tremor to do it.
21:30:01linuxstbAll they needed to do was to add a few bytes in the header to store the number of samples...
21:36:54***Saving seen data "./dancer.seen"
21:37:56]RowaN[how do i get back to the wps from browsing files?
21:38:32markunlinuxstb: I found this ogg parser writen in lisp:, maybe find-playback-length is of some help.
21:38:43 Quit Harpy (Read error: 60 (Operation timed out))
21:38:44Slasheri]RowaN[: press play
21:40:48linuxstbmarkun: Thanks. I'll have a look.
21:41:49linuxstbBut it's been about 13 years since I looked at a lisp program...
21:48:37 Join thegeek_ [0] (
21:48:37 Quit thegeek (Read error: 54 (Connection reset by peer))
22:01:13 Join Meaningful [0] (
22:06:48preglowa great flaming pink thank you goes out to everyone involved with rockbox
22:06:59preglowmy iriver's finally serving me the way i want it to
22:07:39markunpreglow: gapless mp3?
22:08:30preglowno, i had a walk in marvelous weather, all the while listening to gapless oggs queued using a nice on-the-fly playlist :P
22:08:58linuxstbDid you have my latest Ogg changes (elapsed time, and estimated duration)?
22:09:19preglowhaven't transfered a new rockbox yet
22:09:28 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
22:09:31preglowon a related note, i had to reset my player with a piece of wood i found in a hedge :PP
22:09:32 Join bipak_ [0] (
22:09:40linuxstb:-) Very resourceful.
22:09:41preglownow there's technology and nature hand in hand for you
22:09:46 Quit xen` (Read error: 60 (Operation timed out))
22:10:03 Join Stryke` [0] (
22:10:12linuxstbWell, that's what you told the policeman when he asked what you were doing in the hedge...
22:11:29linuxstbAnyone fancy converting the parse-ogg.lisp function that markun found into C? I think I've had enough of ogg.
22:13:45markunWhere is the spdif encoding done in the iriver?
22:14:52markunI wonder if we can send a 'new track' signal so minidisc will split the tracks during recording
22:15:05preglow'course we can
22:15:13preglowjust find out how
22:16:12 Quit bipak (Read error: 60 (Operation timed out))
22:16:39markunI thought the optical diode would be connected to the uda1380, but I can't find anything about spdif in the datasheet.
22:20:46[solid]hmmm weird... seems something happened with rockbox while i was out
22:21:04[solid]it's showing only 45 minutes of runtime and the battery is quite fool... weird
22:21:11[solid]and it turned off
22:21:36[solid]recharging and repeating the vorbis test, with wps info this time :)
22:22:51SlasheriI located one new crashing issue. It's caused by pcm_play_pause -function
22:23:00HClthey should learn how to spell on misticriver
22:23:05SlasheriThat gives the I04: xxx
22:23:32preglowHCl: plzz, it's called kernal if it's on a c64 :P
22:26:37LinusNmarkun: the s/pdif is connected directly to the cpu
22:27:19markunLinusN: thanks, then I will look in the coldfire manual again
22:28:35*LinusN should complete the schematics
22:28:59[solid]"First version of Musepack playback (not working)" - what does 'not working' mean here?:)
22:29:44[solid]not realtime or disabled on purpose?
22:30:20preglow[solid]: it's just barely realtime
22:30:27preglow[solid]: and the codec plugin isn't working
22:30:42preglowkudos to the musepack people for making my work in optimising it hard
22:32:04preglowthe ac3 codec was a nice break from that ;)
22:32:34[solid]so it basically means it won't play any .mpc at the moment, realtime or not, right?
22:34:40LinusNgotta go, cu later
22:34:43 Part LinusN
22:36:02 Part Meaningful
22:45:52 Join asdsd____ [0] (
22:45:56 Part asdsd____
22:47:48yngwiwow, this iriver battery is awesome, playing for 12 hours now (with orig. fw) and still two bars left... when is it gonna end
22:48:52markunAh, I found it! If we change the tracknumber in the cd-subcode that can be send in the spdif packets the minidisk will insert a new track mark.
22:49:18markunNow I just have to borrow a minidisc player..
22:49:27SlasheriNow started new battery test (128k CBR MP3, 50% volume, no cross-fade, 12 tracks with suffle & repeat)
22:49:30Slasherinite :)
22:49:40markunSlasheri: good night
22:49:46bill2or3yngwi, which battery in which iRiver?
22:50:00yngwithis is going to be interesting, good night, slasheri
22:50:13yngwibill2or3: the standard one..
22:51:03bill2or3which iRiver?
22:51:32yngwih140, 11 days old :-)
22:51:54bill2or3ahh. I just ordered an H340
22:52:29preglowoooh, sufflé and repeat
22:52:46yngwii bought the h140 mainly because rockbox :-) (and 'couse i don't care for fancy colour lcd displays and stuff.. :-) )
22:53:17bill2or3yeah, I'm hoping rockbox ends up getting ported
22:53:39yngwii'm sure it will...
22:54:13bill2or3someone added them to the DeviceChart in the wiki, that's a good sign
22:54:23yngwibut i was very impatient
22:54:36yngwithere is even a forum for h3xx
22:55:05yngwiwhich is a good sign too
22:58:13preglowperhaps it's time to remove the wav player in the debug menu? :>
23:01:36 Quit DomZ ("CGI:IRC")
23:12:40preglowlike i suspected
23:13:13preglowmp3 might be gapless, but it will never be glitchless for all inputs
23:14:53HClwhy not?
23:17:22preglowbecause the waveforms rings like a bell in the crossover
23:19:10preglowgimme a sec
23:22:01 Join prethom [0] (
23:22:01 Quit preglow (Read error: 104 (Connection reset by peer))
23:24:00prethomi think you can see where theres a track change
23:24:36 Join Querty [0] (
23:27:30 Nick prethom is now known as preglow (
23:29:44markunDoes it sound very bad?
23:31:04preglowyou can hear it
23:31:10preglowbut it's not a very probable scenario
23:31:14preglowi very seldom listen to sines ;)
23:31:36markunsurfing on sine waves :)
23:31:50preglowthere's very few sine waves in that, actaully
23:31:51preglowbut nice album
23:32:53markunIt would be fun to insert the tracknames in the spdif stream, but I doubt there will be many people who use it..
23:33:05preglowhey, why not
23:33:09preglowif you find out how, do it
23:34:22markunYes, I found it more or less.
23:35:27markunBut I don't have a minidisc player. A friend has one, but I don't think it can display the cd-text.
23:35:34preglowmine can
23:36:57***Saving seen data "./dancer.seen"
23:37:36markunYou can insert cd-subcode in the spdif stream, MCF5249UM.pdf (page 300)
23:38:10markunDoes the minidisc save the original tracknumber of the track?
23:38:17 Quit niobos_study ("off to bed")
23:38:34preglowsincerely doubt it
23:39:33Stryke`preglow: how does it sound decoded by foobar?
23:39:33markunthat's good, because the mindisc starts a new track marker when tracknumber changes. We can flip it between 1 and 2 on track change or something..
23:39:55preglowStryke`: that _is_ foobars output :)
23:40:01Stryke`oh, heh
23:40:20Stryke`if rockbox's output == foobar's output, NO ONE will complain about gapless
23:41:15preglowthen you don't need to worry about that, 'cause foobar source code is how i found out how to do it
23:41:36thegeek_preglow: I doubt anyone would expect gapless to mean acutally transparently crossfading/syncing the sinewaves in the two tracks ;)
23:41:36Stryke`excellent =D
23:41:47preglowthegeek_: vorbis manages it :P
23:41:57thegeek_I'm impressed;)
23:42:02preglowthegeek_: but no, it _is_ unreasonable for a subband codec
23:42:48webguest65perfectionist indeed :)
23:43:15Stryke`i thought MPC was the only major subband codec and that MP3 and Vorbis were transform codecs
23:43:25preglowStryke`: mp3 is a hybrid
23:43:55Stryke`so decoded MPC should suffer from the same glitchiness?
23:44:04preglowi believe it will
23:44:07preglowbut hey
23:44:08preglowit's a sine
23:44:13preglowdon't worry :P
23:51:49preglowi do wonder why my gapless playback isn't perfect, though

Previous day | Next day