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 2007-07-09

00:00:14saratoga2Frode_: since VS doesn't compile for ARM or Coldfire, thats not going to work
00:00:19stripwaxFrode_ there's a lot of information in the wiki about how to setup the gcc tools on your pc. I build on a pc too.
00:00:53Frode_saratoga2, ah, thats the reason ofcourse
00:01:23Frode_stripwax, i did give cygwin a try earlier, without success.. so i kinda gave it all up there, hehe
00:01:36stripwaxFrode_ - it's easiest way, i found anyway.
00:01:49 Join Rincewind [0] (
00:01:49[Ray]frode_: you can always fetch that preconfigured vmware image and work from that
00:01:52stripwaxDavide-NYC : - strange, rebuilt and it's there now. was in Plugins before, now it's in Open With too.
00:01:55*stripwax runs the thing
00:02:23Frode_[Ray]: oh.. how is that setup? for cygwin or?
00:02:47Davide-NYCgood luck, I just had to reboot. I think it just doen't "release" codecs politely. (this is as precise as I'm ever going to get on this one)
00:03:54 Quit Siltaar (Remote closed the connection)
00:04:20 Join LocoTe [0] (
00:04:27 Quit dj``bergamote (Read error: 110 (Connection timed out))
00:05:24 Quit belze_ ("Ich wei deine Monster sind genau wie meine, und mit denen bleibt man besser nicht alleine.")
00:05:49[Ray]Frode_: native linux distribution, see the wiki for details
00:06:06Frode_[Ray]: ok thx for the tip
00:06:52 Join Wiwie [0] (
00:08:36stripwaxinteresting - ipod 320kbps mp3 is 178% realtime and 128kbps mp3 is only 185% realtime. highbirate decode more efficient (per bit)? those extra bits just easier to decode somehow or overhead? but 96kbps is 241% realtime. seems inconsistent
00:10:10stripwaxdoes test_codec keep the cpu boosted or not? seems that a 192kbps took longer to decode than 320kbps, does not seem right
00:10:43amiconnDo you use the same track just encoded at different bitrates?
00:10:53stripwax(320kbps just happened to be the first track)
00:10:55 Quit Frode_ (Read error: 104 (Connection reset by peer))
00:11:04stripwaxamiconn - yes. using the CodePerformanceComparison sampleset
00:11:16stripwaxwhat Davide-NYC was discussing
00:11:38stripwaxI'll post all results when I'm done..
00:12:06 Join hannesd_ [0] (
00:12:23Davide-NYClet me just say that there may be a better sample track out there. Something a bit less noisy with more dynamics and more fidelity.
00:12:37Davide-NYCbut this one'll do for now.
00:13:44*stripwax just hit the codec-loading bug in test_codec, grah
00:14:14saratoga2stripwax: test_codec boosts before decoding
00:14:31 Quit davina ("xchat on Ubuntu 7.04")
00:14:41stripwaxsaratoga2 - cool, so should be unaffected by backlight fading in/out , accidentally brushing scrollwheel, etc
00:16:42 Join BigMac [0] (
00:16:44stripwaxwonder how 320 decode was faster than 192
00:18:27 Quit thegeek (Read error: 110 (Connection timed out))
00:19:57 Quit hannesd (Connection timed out)
00:20:10Davide-NYCstripwax: that's for mp3s? I just posted some results
00:20:25 Join Llorean [0] (
00:20:34 Quit ender` (" It's all fun and games until someone loses an eye. Then it's fun and games you can't see.")
00:20:40stripwaxyep. vorbis decode seems more consistent
00:20:57 Nick hannesd_ is now known as hannesd (
00:21:10*stripwax looks
00:21:16Davide-NYCgeneral question: at what point do we deem a codec optimized?
00:21:17stripwaxDavide-NYC : - hm, the wma is a longer file duration ?
00:21:41Davide-NYCmaybe the encoder chucks silence?
00:22:00*Davide-NYC remembers that there is no silence in the sample file
00:22:16LloreanDavide-NYC: What do you mean by "at what point"?
00:22:19linuxstbstripwax: No, the asf parser doesn't seem to give an accurate file duration
00:22:31Davide-NYCat what percentage of realtime decoding
00:22:39LloreanThere is no specific percentage.
00:22:42LloreanThere can't be
00:22:42stripwax> 100% for all useful bitrates?
00:22:52linuxstbI would say closer to 200%
00:22:56Davide-NYCit would have to be well over %100
00:22:57LloreanYou might need 130% or higher on the iPod video for it to play without skips, while 103% might do on the H120
00:23:17Davide-NYCso ballpark %200
00:23:33LloreanWhy pick a percentage, it's misleading.
00:23:53linuxstbWhy is a percentage misleading? How else can you judge it?
00:23:54LloreanThere is no point where optimization is 'done' anyway.
00:24:05Lloreanlinuxstb: It's misleading if you're using a percentage to say a codec is "optimized"
00:24:14linuxstbNo, but there's a point where a codec is useful...
00:24:23stripwaxLlorean - is there a better metric to say how efficient the codec is?
00:24:24LloreanBut that's a different question from what he asked.
00:24:29Davide-NYCI'm trying to save everyone time. For "mature" codecs it seems pointless to do extensive testing
00:24:47Davide-NYCif it runs at %400 it's not a priority
00:25:08LloreanThere are basically three states for a codec: Plays realtime without DSP effects, plays realtime with all DSP effects enabled, and doesn't boost (I suppose you could add doesn't boost with DSP effects enabled)
00:25:26saratoga2i'd consider a well optimized codec one that runs with minimum boosting reasonibly possible
00:25:38 Join My_Sic [0] (
00:25:42LloreanI'd consider "mature" one that doesn't skip with DSP effects on.
00:25:58LloreanBasically, one that allows the full use of Rockbox functionality.
00:26:03Davide-NYCtrue true and true.
00:26:08linuxstbThe idea in the past was, for each combination of codec and target, to benchmark a "typical" bitrate, and to benchmark the highest bitrate (or whatever is slowest to decode for a particular format - e.g. for APE, lower bitrates are harder to decode).
00:26:18amiconnNo boosting would be a nice goal
00:26:27amiconn(under normal circumstances)
00:26:28LloreanNo boosting would be a great goal.
00:26:43amiconnWell, mp3 achieves this goal on coldfire
00:27:18LloreanDavide-NYC: What you could do is figure out what % realtime is necessary on a target to play without skipping. A "calibration" as it were. Then figure out what percent is necessary with DSP settings on.
00:27:32stripwaxfair enough. lets say we have a codec X that runs at Y% realtime as a benchmark and is fast enough to run realtime with all DSP effects. Is %age realtime not a good benchmark for comparing other codecs to this one (even to see roughly how fast they are)?
00:27:40stripwaxheh. what Llorean said.
00:27:45Davide-NYCLlorean: isn't this also WPS dependat?
00:28:02LloreanDepends on the target.
00:28:17LloreanBut for this purpose, we could say "Rockbox default with removed peakmeters"?
00:28:25LloreanAt least for the base realtime score.
00:28:28Davide-NYCI say leave the peakmeters
00:28:45LloreanActually, use iCatcher.
00:29:20Davide-NYCright the test_codec plugin doesn't have peakmeters or any WPS for that matter so I guess it's moot?
00:29:22LloreanCloser to a realistic WPS, and either going to be our default, or likely to have similar performance characteristics to whatever will become the default.
00:29:58stripwaxa dummy codec that just uses up CPU and emits a signal that enables us to detect if it's skipping or not?
00:29:59Davide-NYCstripwax: is it possible you had a codec crash during that 320kbps decode?
00:30:03stripwaxit didn't
00:30:08stripwaxwas the first one
00:30:20pixelmabefore saratoga2's latest changes wma ran at 105% realtime on my M5 which still caused skipping now and then - even with a "calm" wps. Now at 122% it doesn't anymore - including a test on a wps with peakmeters (but then scrolling lines were slow)
00:30:51pixelmaa 128kbps track
00:32:03LloreanSo, what does 128kbps MP3 decode at on coldfire?
00:32:13Davide-NYCover %400
00:33:18amiconnEven 192kbps decodes >500% realtime on cf
00:33:54amiconnpreglow did a great job on libmad
00:34:10Davide-NYChear hear!
00:34:23LloreanAnd WMA is approximately realtime on CF now?
00:34:52amiconnIt's already usable (didn't try a colour cf target with peakmeters yet)
00:35:04pixelmaas I just said ;)
00:36:23 Quit sargun (Remote closed the connection)
00:36:40 Quit ]RowaN[ (Read error: 110 (Connection timed out))
00:36:49Davide-NYCThe test_codec plugin crashes a lot making testing a bit tedious. I'm constantly rebooting my h140
00:36:53LloreanDavide-NYC: Why is there a set of tools to use to create the sample audio files, rather than just a list of sample audio files?
00:37:15LloreanI think the test_codec plugin gets stack overflows right at the end, or at least, I know it did on my Sansa at one point.
00:37:32amiconntest_codec messes up codec swapping
00:37:33Davide-NYCa lot less to download
00:37:46amiconnWith voice enabled, it always crashes on exit
00:38:06Davide-NYCwithout voice enabled it crashes quite a bit.
00:38:09amiconnAnd I think there are also problems with voice disabled
00:38:30LloreanDavide-NYC: Yes, but also useless for anyone not running a compatible OS.
00:38:44stripwaxshall I upload the resulting files somewhere?
00:38:51LloreanOur only option is to use our own encoders, which can result in differing files.
00:39:26 Quit LocoTe ()
00:39:26Davide-NYCOK, but before we do all of that maybe we should see if we can find a better sample file.
00:39:27*Llorean goes back to getting caught up.
00:39:43DerPapstGood night everyone ;)
00:39:50 Quit DerPapst ("So Long And Thanks For All The Fish!")
00:40:12scorcheLlorean: i assume you would be willing to be one of those with access to
00:40:20Lloreanscorche: Aye
00:41:01amiconnOnce we settled for the definitive test track, I can provide mp2 and mp1 encodings
00:41:28stripwaxDavide-NYC : - hm, mine didn't actually generate any wma files ..
00:41:50Davide-NYCdid you install the WME9 package?
00:42:20stripwaxjust tried reruning encode.bat, seems an error message speeds past the screen way too fast to read.
00:43:27Davide-NYCI may have botched the link
00:43:31Davide-NYCtry this
00:43:31stripwaxWScript.CreateObject: Could not locate automation class named "WMEndEng.WMEncoder"
00:43:58*stripwax tries..
00:44:05stripwax^WMEncEng not WMEndEng in error, sorry
00:45:00stripwaxshould i uninstall the other package then .. ? :-)
00:45:44 Join thegeek [0] (
00:48:19stripwaxok, wma works now ;-)
00:51:48Davide-NYCsweet. I fixed the link in the twiki
00:54:24 Join Guile` [0] (n=Guile@
00:57:04 Quit linuxstb ("CGI:IRC")
00:58:27 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
00:59:25 Quit lee-qid ("aufwiederbyebientotsayonara")
00:59:40 Quit Wiwie ("Miranda IM! Smaller, Faster, Easier.")
01:01:27***Saving seen data "./dancer.seen"
01:01:46 Quit obo ("bye")
01:02:28saratoga2does anyone have a good resource on how the MDCT works ?
01:02:39saratoga2i'm trying to figure out why the pre and post roation code works the way it does
01:02:53 Quit hannesd (Read error: 110 (Connection timed out))
01:03:00stripwaxhm, how can I change/check my twiki password?
01:03:40stripwaxnemmind. will reset password.
01:04:09 Join safetydan [0] (i=cbca159f@rockbox/developer/safetydan)
01:06:21 Quit entheh (Remote closed the connection)
01:06:23merbanansaratoga2: I'll mail you some papers, but basicly the cos function is expanded to e^'s and then some trig identities split them to a form that maps to a complex fft
01:06:45merbananwith the addition of the pre and post rotation steps
01:07:01stripwaxDavide-NYC : - nice stats! will have mine up shortly
01:07:14 Join entheh [0] (
01:08:10Davide-NYCoh man, I just realized they are all out of order.
01:08:38stripwaxi reordered for you :-)
01:08:40Davide-NYCI must have had shuffle or random on or something.
01:09:14stripwaxnah, i have the exact same order for mine too. perhaps the order the files were copied to device?
01:09:49saratoga2merbanan: can I ask you about the roation steps?
01:09:53 Nick Ryuu is now known as HellDragon (n=JD@unaffiliated/helldragon)
01:09:57saratoga2the ffmpeg code doesn't something very odd with them
01:09:58linuxstbstripwax: Yes, test_codec just processes the files in the order they appear in the directory (i.e. the order returned by readdir()) - it doesn't sort them.
01:10:39 Quit miepchen^schlaf ("Verlassend")
01:10:47saratoga2as they step through the nagles of sin and cos, they add 0.125/block_length to the phase of each angle
01:10:49merbanansaratoga2: sure but I reserve the right to not know the answer :), what do you need to know ?
01:11:05saratoga2that makes a really tiny difference
01:11:31saratoga2any idea why they do it?
01:11:34 Part dan_a
01:11:48saratoga2if i could toss that factor out, I could make things quite a lot faster
01:12:24safetydanAny reason not all codecs are listed on the CodecPerformanceComparison page? Seems we're missing ALAC and Speex at least.
01:12:31amiconnHmm, does wma need an mdct that's replaced with an fft or the other way round?
01:12:31merbanansaratoga2: what is the name of the function ?
01:12:59 Quit Guile`` (Read error: 110 (Connection timed out))
01:13:07pixelmasafetydan: and ape
01:13:22Davide-NYCsafetydan: never heard of these, wil gladly nclude them
01:13:28saratoga2the value of alpha is what i'm wondering about
01:13:29Davide-NYCurghh APE
01:13:35saratoga2that 1/8 correction seems very odd
01:13:48saratoga2amiconn: what do you mean?
01:13:53 Quit BHSPitLappy (Read error: 104 (Connection reset by peer))
01:14:06linuxstbstripwax, Davide-NYC: Any reason why the table contains the column headings in the cells (e.g. Decode time, File duration, realtime)?
01:14:38Davide-NYCOh, I was just trying to clean things up and left it. It's unnecessary
01:15:01pixelmamay I ask what "wassat" is?
01:15:05Davide-NYCthe plugin output is player friendly but wiki unfriendly
01:15:14linuxstbFeel free to patch...
01:15:21Davide-NYCI have no idea what the second column data is exactly
01:15:25 Quit Robin0800 (Read error: 104 (Connection reset by peer))
01:15:25safetydanDavide-NYC: There's also MPC, ADX, Shorten, and AIFF/WAV though obviously the last one isn't needed. See
01:15:29linuxstbYou could even make it output wiki markup...
01:15:55Davide-NYCanyone care to supply me with an itunes-less ALAC encoder
01:16:27amiconnsafetydan: WAV is actually multiple codecs, some of which might be worth testing
01:16:28Davide-NYClinuxstb: contributing to RB has a bi-annual even for me :-(
01:16:28linuxstbI think dbpoweramp have their own ALAC encoder
01:17:49 Part chrisjs169_ ("Konversation terminated!")
01:18:59stripwaxdone for 5g 64mb. mpc also.
01:19:15Davide-NYCMPC was crashing on me. I have the results now
01:19:26merbanansaratoga2: ok, I don't know the exact reason why the 1/8 is there, but it is needed for the transform to be revertable
01:19:44stripwaxstrange, can't find my weird mp3 data. output logs make sense
01:19:51stripwaxmaybe the screen output was garbaged
01:20:40merbanansaratoga2: anyway you shouldn't runtime calculate constant tables on embedded targets
01:21:33Davide-NYC(sorry about the screming)
01:21:46merbanansaratoga2: just precalculate them, you should do the same for the vlc tables
01:21:47stripwaxI didn't do wavpack + flac, worth doing?
01:22:05saratoga2merbanan: they're computed when the codec is initialized
01:22:16Davide-NYCstripwax: flac no. maybe wavpack.
01:22:55saratoga2merbanan: removing that 1/8 means I could merge all 5 MDCT rotation windows into 1 window, saving a lot of memory and fitting it all into IRAM
01:22:57Davide-NYCactually, I don;t know. do them and then we'll have a better idea for the future.
01:23:01Davide-NYCgotta go.
01:23:09 Quit Davide-NYC ("ChatZilla [Firefox]")
01:23:20saratoga2i tested it out and the RMS error between the before and after files is -73dB
01:25:10 Join petur [0] (n=petur@rockbox/developer/petur)
01:25:36merbanansaratoga2: the tremor mdct uses linear interpolation for something, it might be those factors
01:28:17merbanananyway if -73dB is less then +- 1 bit then it might be worth the sacrifice
01:30:19merbananum, less then +-2 bits
01:31:19stripwaxok, flac and wavpack up for ipod too
01:31:23stripwaxgotta go now, gnight
01:35:27 Quit spiorf (Remote closed the connection)
01:38:12 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
01:38:32 Join stripwax [0] (
01:40:56 Part pixelma
01:44:03Soul-SlayerIs the ipod mini build currently stable?
01:46:09 Quit entheh ("^~")
01:49:09 Join Nico_P [0] (
01:49:36merbanansaratoga2: if you use this shortcut then there should be a value that would minimize the distortion of the error introduced should be 0<1/8
01:57:16 Quit My_Sic (Read error: 104 (Connection reset by peer))
01:59:21 Quit stripwax (Read error: 110 (Connection timed out))
02:01:21 Quit Rincewind ("Cya")
02:02:14saratoga2merbanan: yeah I figured, i'm just apprehsive doing that since I have no idea what that factor was even supposed to do
02:02:37saratoga2i want to figure that out before I commit any changes to it
02:05:55 Quit petur ("Zzzz")
02:07:32saratoga2merbanan: interestingly, faad uses the same 1/8 factor
02:08:37saratoga2heh, if I go for this trick, it should work on FAAD too
02:09:00 Join lymeca_ [0] (n=lymeca@unaffiliated/lymeca)
02:11:27linuxstbsaratoga2: I've got a couple of cosmetic comments about your last commits - 1) You're using TABs; 2) I think it's better to delete old code from the .c file (it's still visible in the SVN history), rather than commenting it out. In fact, wnadeci,c seems to contain a lot of commented out code, which it would be nice to remove now it's in SVN.
02:12:59saratoga2linuxstb: I'll start converting my tabs to spaces from now on
02:14:12linuxstbA code-cleanup commit (with only whitespace changes) would be nice, rather than mixing it with functional changes.
02:15:20 Join qwx [0] (
02:17:37saratoga2ok, i'll do that now
02:19:12 Quit JdGordon_ (Read error: 110 (Connection timed out))
02:24:22 Quit lymeca (Read error: 110 (Connection timed out))
02:26:34RoC_MasterMindWhy are tabs bad?
02:31:08safetydanRoC_MasterMind: it's easier if everyone uses the same indentation method. Mixing spaces and tabs leads to badness. Rockbox has settled on spaces only for indentation. Hence, tabs are bad.
02:31:38LloreanTabs can also vary in apparent size depending on the editor you're using, right?
02:31:44RoC_MasterMindHow many spaces is a tab?
02:31:48scorchewe use 4
02:31:55RoC_MasterMindis that pretty normal?
02:32:03scorcheLlorean: it is able to be specified in many
02:32:06linuxstbRoC_MasterMind: That's the problem (or one of them) - there's no standard "X spaces is a tab"
02:32:08scorcheRoC_MasterMind: yes
02:32:17scorche4 is the "typical" though
02:33:01 Quit qwm (Read error: 110 (Connection timed out))
02:34:15 Join fiveofoh [0] (
02:34:36fiveofohHey, I'm looking for a wiki confirmation, of whatever you call it :)
02:34:43fiveofohJoelBradshaw being my username
02:35:09fiveofoh"Wiki write permission" being the technical term I suppose
02:35:46 Quit desowin ("use linux")
02:35:52linuxstbsaratoga2: Thanks - that's looking much nicer now ;)
02:35:52zeyou'd think one might like to take advantage of that "problem" with tabs...
02:36:12scorchewhat "advantage"?
02:36:15zei.e. people like different tab widths... tab shows up as different widths depending on their editor configuration
02:36:20zeput the 2 problems together and you've got a solution :p
02:36:28 Join Entasis [0] (
02:36:36scorchespaces are more compatible than spaces are, in general
02:36:58scorcheerr....tabs on the second one
02:37:48zeyeah tabs can be weird... if i try to paste them i'll usually end up with an inverse I instead, at least on my screen
02:37:55zein irssi
02:40:19fiveofohCan someone give me write access? I'm working on starting a wiki page for the Sansa m200
02:41:30Battousaiis m better than e?
02:42:13fiveofohHaha yeah but I own one :P
02:42:19saratoga2i have a struct entry declared as "fixed32 *coefs;" but later when I try to set that equal to an array declared as "fixed32 coefsarray[MAX_CHANNELS][BLOCK_MAX_SIZE] IBSS_ATTR;" it fails with a pointer mismatch error
02:42:21fiveofohJust a sec, I need to scan my board :D
02:42:35saratoga2what am i forgetting about how c works
02:43:08Battousaiyou'd need another pointer or *coefs[]
02:43:56safetydanze, the problem comes when you mix tabs and spaces. Sometimes you don't want to indent a full tab stop so you use spaces, which then breaks for everyone else who has their tabs set to a different number of spaces.
02:43:58Battousaiboth of which count as another pointer i suppose
02:44:38fiveofohBattousai: Yeah, scan my pcb
02:44:49fiveofohHmm, what dpi should I be scanning this at?
02:44:56Battousai1200x1200 baby
02:44:59Battousaiwe need full disclosure
02:45:05fiveofohI suppose as high as I can...which is like 4800
02:45:10Battousaithat works too
02:45:11fiveofohI can always scale down :)
02:45:20fiveofohScaling up isn't so easy
02:45:23zesafetydan: well yeah mixing would be bad
02:45:42*scorche sighs
02:45:46zesafetydan: if you're gonna use tabs, then a single tab should be the shortest you use
02:45:55scorchefiveofoh: as long as we can read the chips, it is fine
02:45:59saratoga2Battousai: what would I need?
02:47:01scorchefiveofoh: you have batteries ran out of battery, so i couldnt before =/
02:47:04Battousaiyou would have to declare coefsarray as **coefsarray instead of *coefsarray, if i understand your problem correctly
02:48:02Battousaii'm at the end of a long day so i could very well be misreading you
02:48:14fiveofohscorche: Haha thanks :)
02:49:01saratoga2Battousai: oh yeah, thanks
02:49:33fiveofohOh, btw, I was looking around and found this post
02:49:43fiveofohCan someone please find him and slap him for me?
02:50:20Lloreanfiveofoh: That has nothing to do with Rockbox. Please try to keep the channel on-topic
02:50:53fiveofohOh okay
02:51:13fiveofoh(Well, it sort of's in a thread about porting the sansa m240 to rockbox)
02:51:15fiveofohBut yeah
02:51:24saratoga2i changed the type to "fixed32 **coefs" but I still get an incomptable pointer warning
02:51:34fiveofohPoint taken, I'm used to IRC being 90% off-topic :P
02:53:22 Quit Entasis_ (Read error: 110 (Connection timed out))
03:00:03 Quit Arathis (Read error: 110 (Connection timed out))
03:00:46saratoga2can anyone help me with my pointer question?
03:01:28***Saving seen data "./dancer.seen"
03:05:33jhMikeSfixed32 (*coefs)[MAX_CHANNELS][BLOCK_MAX_SIZE]; coef = (*coefs)[a][b] <== I think
03:07:27saratoga2wouldn't that first part be a pointer to the 2D array?
03:07:54jhMikeSyes that's what I think was needed (I've no context though)
03:08:16 Quit PaulJam (".")
03:08:19saratoga2i'm trying to declare a 2D array of fix32, and then point to it from a struct
03:08:56jhMikeSthen the pointer to array is what's needed
03:08:57saratoga2basically, the struct had the 2D array inside it, I wanted to mvoe it to IRAM, so i changed it to a point and am trying to assign that pointer to the area in IRAM
03:09:45jhMikeSthe second part accesses the element [a][b]
03:10:15saratoga2yeah I know, but how does that help?
03:11:40jhMikeSThe pointers should be compatible then
03:13:18 Quit Thundercloud (Remote closed the connection)
03:14:11saratoga2to be clear, you're saying I should decleare fixed32 coefsarray[MAX_CHANNELS][BLOCK_MAX_SIZE]; as the array, then fixed32 (*coefs)[MAX_CHANNELS][BLOCK_MAX_SIZE]; in the struct and finally s->coef = (*coefs)[a][b] to assign it?
03:15:32 Join Thundercloud [0] (
03:18:01 Quit Soul-Slayer (Read error: 104 (Connection reset by peer))
03:18:09 Quit Xerion (" ")
03:18:11jhMikeSyes, iiuc what's being done
03:18:39jhMikeSwait, to assign the pointer or get the element?
03:18:58saratoga2i dont' care about getting the element right now
03:19:11saratoga2i just want to assign the array
03:19:26jhMikeSs->coef = coefs
03:19:33 Join andrew__4 [0] (
03:19:37saratoga2i've done this for a variety of structs and 1D arrays, I have no idea why this blows up for 2D
03:19:49Battousaiyou'd have to assign it by reference
03:20:00saratoga2jhMikeS: yes i've tried that but it doesn't compile
03:20:01Battousaii guess c can do that on its own
03:21:08jhMikeSBattousai: yeah, perhaps as &coefs even a type cast as (fixed32 (*)[MAX_CHANNELS][BLOCK_MAX_SIZE])
03:21:26Battousaithe compiler should usually take care of that though, right?
03:21:52Battousailike when passing arrays as arguments
03:23:00jhMikeSbut fixed32 ** coefs won't know about the dimensions. 2d array as an arg knows the dimensions.
03:24:53 Join TiMiD[FD] [0] (n=TiMiD[
03:29:15Nico_PLlorean: hi
03:29:49fiveofohHoly there a way to do a thumbnail on the wiki page?
03:30:32saratoga2hmm, so trying to access the array now gives me trouble: *(s->coefs)[0][i] = a + b;
03:30:48saratoga2"incompatable types"
03:33:22 Quit andrew__4 ("Lost terminal")
03:35:11 Quit dj`bergamote ("It's a me, Mario !!")
03:40:29 Quit atsea- (Remote closed the connection)
03:41:16 Quit BigMac (Read error: 110 (Connection timed out))
03:44:26*fiveofoh makes his own small versions
03:49:18saratoga2all that for a 5% speed up on ARM
03:50:53 Join billytwowilly [0] (
03:51:45fiveofohThere we are...anything that I should add at this point?
03:54:40 Join lymeca| [0] (
03:56:31jhMikeSsaratoga2: perhaps try creating a local copy of the array pointer so it can be sure to have the address in a register. I really don't trust gcc regarding those matters.
03:59:32 Quit Nico_P (Remote closed the connection)
03:59:57saratoga2just commited
04:00:00saratoga2will fix that
04:04:56*jhMikeS has to actually try this wma thingy he's been hearing about :p
04:06:48 Quit Thundercloud (Remote closed the connection)
04:08:10 Quit billytwowilly (Remote closed the connection)
04:08:43saratoga2jhMikeS: oddly enough, creating a temp variable made that slightly slower
04:09:02 Join billytwowilly [0] (
04:09:14saratoga2182% realtime verses 185%
04:09:18jhMikeSfor arm or coldfire?
04:09:47 Quit lymeca_ (Read error: 110 (Connection timed out))
04:10:15jhMikeSthat stuff usually takes a lot experimenting with "what if"...some works some fails miserably
04:12:09 Quit safetydan ("CGI:IRC (Ping timeout)")
04:12:28saratoga2theres only one loop anyway that accesses that variable, the rest its always passed to funtions that work on it
04:12:32saratoga2or rather work on one channel's worth of it at a time
04:13:22saratoga2anyway, if someone gets a chance, please try out the changes I made on coldfire
04:13:28saratoga2i'll check the logs in the morning
04:13:52jhMikeStrying now
04:16:18 Quit billytwowilly (Remote closed the connection)
04:16:51 Quit Soap_ ()
04:17:08saratoga2hmm i just noticed there doesn't appear to be a need to actually have three seperate buffers in the ff_imdct_calc function
04:17:33saratoga2i could use the output as a scratch buffer, and then the input and finally the output again
04:17:45saratoga2problem is the buffers have different types
04:18:07saratoga2is there someway I can make the same block of memory be both a struct and an array?
04:18:34 Quit hcs ("Leaving.")
04:18:44 Join billytwowilly [0] (
04:21:11Battousaibrick mania rocks my socks off
04:21:21jhMikeSabout 81% boost on my x5 (192kbit)
04:21:34jhMikeSthat codec doesn't yield very much does it?
04:21:53saratoga2i think its just very slow on coldfire
04:21:59saratoga2you didn't try test codec?
04:22:08saratoga2it used to be around 125%
04:22:10jhMikeSno, I'm just getting a feel for it
04:22:42jhMikeS81% boost shouldn't make the system so choppy. spc doesn't do that even at 90% boost
04:24:10 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
04:27:45jhMikeSthat outbuf is huge 262144 bytes :D
04:30:43TiMiD[FD]are there other people getting a stack overflow with the mpeg plugin on iriver h1x0 ?
04:31:23TiMiD[FD]since some days it crashes instantaneously
04:31:36 Quit saratoga2 ("CGI:IRC (Ping timeout)")
04:38:55 Quit hcs ("Leaving.")
04:41:44 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
04:43:56 Quit saratoga (Client Quit)
04:46:32 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
04:50:40 Quit saratoga (Client Quit)
04:53:53 Join saratoga2 [0] (
04:54:11saratoga2my crazy idea for in place MDCT actually worked
04:54:24saratoga2should do a lot for coldfire I think
04:54:47saratoga2just got to figure out how to compact a few more tables and I'll have essentially the entire MDCT/FFT/Windowing process out of IRAM
04:58:06 Part EbErT
04:59:42 Quit Soap ()
05:01:29***Saving seen data "./dancer.seen"
05:02:41 Quit saratoga2 ("Leaving")
05:10:32 Join jhulst [0] (
05:22:44alienbiker99site is down again?
05:23:47scorchei can ping it
05:24:13alienbiker99oh nevermind just slow
05:32:22 Join lymeca_ [0] (n=lymeca@unaffiliated/lymeca)
05:48:04 Quit lymeca| (Read error: 110 (Connection timed out))
06:17:43 Quit Nibbier (Read error: 110 (Connection timed out))
06:18:24 Join Nibbier [0] (
06:19:42 Join sharap [0] (
06:20:41sharapdoes the 80G video ipod use the ipodvideo build or the ipodvideo64mb build?
06:42:08 Join EMonk [0] (
06:43:49 Nick EMonk is now known as ElectricMonk (
06:45:02 Part ElectricMonk
06:45:11 Join ElectricMonk [0] (
06:56:52 Join Gibbed [0] (
06:56:54 Quit Rick (Nick collision from services.)
07:00:23 Quit billytwowilly (
07:00:23 Quit krazykit (
07:00:23 Quit seclein2 (
07:00:23 Quit midkay (
07:00:23 Quit ansivirus (
07:00:23 Quit _pill (
07:00:23 Quit SirFunk (
07:00:23 Quit Lynx_ (
07:00:23 Quit GodEater (
07:01:30***Saving seen data "./dancer.seen"
07:03:53 Join Lynx- [0] (
07:03:53NJoinbillytwowilly [0] (
07:03:53NJoinkrazykit [0] (
07:03:53NJoinseclein2 [0] (n=scl@
07:03:53NJoinmidkay [0] (n=midkay@rockbox/developer/midkay)
07:03:53NJoinansivirus [0] (
07:03:53NJoin_pill [0] (
07:03:53NJoinSirFunk [0] (
07:03:53NJoinLynx_ [0] (
07:03:53NJoinGodEater [0] (
07:07:30 Part Llorean
07:10:00 Quit jhMikeS (Nick collision from services.)
07:10:06 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
07:13:40 Quit Lynx_ (Connection timed out)
07:13:40 Nick Lynx- is now known as Lynx_ (
07:14:59 Join JdGordon [0] (
07:25:35 Join BHSPitLappy [0] (
07:31:32 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
07:33:57 Join Triple-z [0] (i=Mp3@
07:37:27 Join mbr [0] (
07:37:44ptw419finally have a viable printf function for the gigabeat S serial port
07:39:52scorcheJdGordon: i found you a new job
07:40:27 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
07:43:23Triple-zanyone know if there is a specific rockbox chans for each player?
07:45:47 Quit linuxstb (Read error: 110 (Connection timed out))
07:46:32scorcheTriple-z: there isnt
07:47:02scorchewhy would we?...rockbox's aim is to be as constant as possible through all targets
07:49:33 Join petur [0] (n=petur@rockbox/developer/petur)
07:55:00 Join lostnihilist [0] (
08:00:20 Quit Echelon (Remote closed the connection)
08:00:58 Join Echelon [0] (
08:05:55 Join bdgraue [0] (
08:11:03 Join linuxstb__ [0] (
08:12:32*amiconn had an interesting idea how to save battery power
08:12:50amiconnWill most certainly work on PP and probably work on cf
08:13:31amiconnIf we set the PLL to bypass while the CPU is sleeping (both cores on PP), it should draw a little less power
08:13:59amiconnAnd switching PLL bypass doesn't need an expensive relock wait
08:14:34amiconnThe problem on cf might be timer precision, as the prescaler(s) also need to be switched
08:15:35amiconnIf this works, we should be able to choose the unboosted cpu clock somewhat higher on colour targets in order for better ui performance
08:21:57 Quit petur ("home->work")
08:23:11 Quit linuxstb_ (Read error: 113 (No route to host))
08:24:00JdGordonamiconn: wouldnt setting a higher unboosted speed cancel out the savings with the switching?
08:28:04 Join ender` [0] (
08:29:56 Join DarthShrine [0] (n=Angus@pdpc/supporter/student/DarthShrine)
08:43:22 Join atsea- [0] (i=ariel@gateway/tor/x-bd1593c8e15ce28d)
08:46:16 Quit atsea- (Remote closed the connection)
08:48:21 Join petur [0] (n=petur@rockbox/developer/petur)
08:51:46 Quit Triple-z ()
08:52:55 Join Rob2222 [0] (
08:52:57 Join atsea- [0] (i=ariel@gateway/tor/x-ec355590b7451c5e)
08:56:07 Join Zagor [0] (n=bjorn@rockbox/developer/Zagor)
09:01:33***Saving seen data "./dancer.seen"
09:01:48 Quit perplexity (Read error: 110 (Connection timed out))
09:04:42TiMiD[FD]oauis a/c
09:05:02TiMiD[FD]an unfortunate paste
09:05:45*scorche has seen worse
09:05:57 Quit Gibbed (
09:05:57 Quit FOAD (
09:05:57 Quit gtkspert (
09:05:57 Quit bospaadje (
09:05:57 Quit amigan (
09:06:21amiconnJdGordon: No, since if the cpu has almost nothing to do, it will be sleeping most of the time, running at idle speed
09:06:55amiconnBut you won't notice the idle speed in the form of sluggish ui response, because it can be switched to the standard clock in an instant
09:07:59 Join stripwax [0] (
09:09:08amiconnIn theory (if there were no overhead at all), we ould even get rid of the usual boosting business
09:09:12 Quit Siku ()
09:09:22 Join davina [0] (
09:11:30 Quit Rob222241 (Read error: 110 (Connection timed out))
09:12:54 Join kaaloo [0] (
09:13:05 Part kaaloo
09:17:38 Quit ze (Read error: 110 (Connection timed out))
09:19:16stripwax mpa_removed_iram seems to give 20% speedup on 5g 64mb according to test_codec. will post results this evening
09:20:50amiconnThere's the other libmad speedup patch (by tomal)
09:21:39amiconnI would like to commit it, but there is a problem with too much iram usage in mpegplayer
09:22:34 Part fiveofoh
09:23:35 Join Llorean [0] (
09:24:21 Join perplexity [0] (
09:32:27 Join kubiix [0] (
09:35:50*petur saw in action yesterday evening :)
09:37:05Lloreanpetur: Planning a Rockbox port?
09:37:59 Quit sharap ("I-n-v-i-s-i-o-n 2.0 Build 3515")
09:38:25stripwaxamiconn - tracker # ?
09:38:26 Join miepchen^schlaf [0] (
09:39:11stripwaxamiconn - wait, isn't this the patch you're talking about anyway?
09:39:12 Join pondlife [0] (
09:41:39 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
09:41:56amiconnAh, yes
09:41:59 Join stripwax [0] (
09:42:12amiconnDidn't know that there was an update that solves the iram clash
09:42:44 Quit Neovanglist (Read error: 104 (Connection reset by peer))
09:44:20 Join nardul [0] (
09:45:00nardulMorning, just a quick question, what's the standard mAH on a 30GB ipod video?
09:45:17nardulI'm asking because the battery life estimator on my ipod is wildly inaccurate
09:46:22 Join spiorf [0] (
09:48:18peturnardul: don't use that, it is not reliable at all
09:48:25peturuse the % indication
09:49:38amiconnstripwax: The no_iram variant should be limited to libmad for mpegplayer
09:49:48amiconn(dunno if the patch already does that)
09:50:53nardulpetur, I knwo it's not reliable, when fully charged it says 2.5 hours, and it lasts around 9 :) But i've seen there's an option to change the mAH on the battery in rockbox, that won't help a bit?
09:50:58amiconnOr maybe iram usage should be removed for pp5020+, but not for pp5002 and pnx0101
09:53:06LloreanDo the fixes for CPU scaling, etc, resolve some of the problems that were happening with the current dual core implementation?
09:55:03 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
09:59:58JdGordonamiconn: ah ok, good luck with it then :)
10:00:01peturnardul: battery time estimation for ipod isn't calibrated at all so while changing the mAh may help a bit it still won't be accurate. As I said, don't use it...
10:00:22nardulokay, thanks
10:00:26 Join Paprica [0] (n=Paprica@
10:01:35 Quit stripwax (Read error: 110 (Connection timed out))
10:06:54 Quit linuxstb__ (Read error: 113 (No route to host))
10:17:28 Quit ^BeN^ (Read error: 110 (Connection timed out))
10:19:42 Join bluebrother [0] (i=GToREavO@rockbox/staff/bluebrother)
10:21:21 Join ^BeN^ [0] (n=Paprica@
10:21:46 Quit RoC_MasterMind (Remote closed the connection)
10:22:54 Join safetydan [0] (n=dan@rockbox/developer/safetydan)
10:25:26 Join obo [0] (
10:25:27 Quit Llorean (Remote closed the connection)
10:25:51 Join Llorean [0] (
10:31:50 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
10:31:51 Join kretender [0] (
10:32:08 Join miepchen^schlaf [0] (
10:37:17 Join Neovanglist [0] (
10:39:01 Join ze [0] (
10:40:14 Quit ElectricMonk ("Chatzilla [SeaMonkey 1.1/2007011111]")
10:41:13 Quit Paprica (Read error: 110 (Connection timed out))
10:44:46 Join pixelma [0] (i=pixelma@
10:45:00 Quit darkless ("Leaving")
10:55:40 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
10:55:46 Join miepchen^schlaf [0] (
10:58:38 Join lee-qid [0] (
11:01:38***Saving seen data "./dancer.seen"
11:03:11 Quit ze (Read error: 104 (Connection reset by peer))
11:03:14 Join ze [0] (
11:05:03 Quit jhMikeS (Nick collision from services.)
11:05:08 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
11:05:41scorchelinuxstb_: around?
11:08:18 Join aliask [0] (
11:11:56linuxstb_scorche: Yes.
11:12:01 Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
11:12:49scorcheis there any other reason than the play button depressed that would cause the bootloader to say the can't find linux.bin error?
11:12:50 Quit kretender (Read error: 104 (Connection reset by peer))
11:12:55scorchesuch as formatting?
11:13:35linuxstbI wouldn't have thought so. Let me check the source...
11:14:35linuxstbNo, that code is within a "if (btn==BUTTON_PLAY)" test, so it should only be run if play is held.
11:14:59linuxstbIs someone experiencing that?
11:15:30scorcheyes, but let me get a bit more info...
11:17:09 Quit nardul ("Leaving")
11:17:24linuxstbamiconn: Regarding Tomal's libmad patch - yes, I think we should disable IRAM usage for all PP502x targets, and disable it in mpegplayer for the PP5002. Using #if(n)def MPEGPLAYER should work now. The patch on the tracker disables it unconditionally for PP IIRC.
11:18:00linuxstbamiconn: Also, I think preglow wanted to test the accuracy of the output before committing, but I don't think he's had chance yet.
11:32:45 Quit perplexity (Read error: 110 (Connection timed out))
11:35:10 Join perplexity [0] (
11:36:29linuxstbsaratoga: (for the logs) I don't know what WMA file people have been testing on Coldfire, but the Bangles track (80kbps) from mplayer's test directory is currently 151% realtime on my h140.
11:37:13bluebrotherI just got 138% on h120 with a 192 kbps track
11:39:35pixelma~142% with a 128kbps wma
11:42:07*amiconn got 138% with 128kbps wma
11:42:35pixelmahmm... but it seems I have a broken wma which locks up the player with the "write wav" option in test_codec - the speed test runs ok but it's a tad bit slower than the other tracks (with the same bitrate) that I got (1 track in 5)
11:43:02 Join webguest91 [0] (i=c023110a@gateway/web/cgi-irc/
11:43:13 Nick lymeca_ is now known as lymeca (n=lymeca@unaffiliated/lymeca)
11:45:17pixelmathat track is also played in "normal" playback but you can hear a "pfft" (sorry can't explain better) in maybe 3 places. Before last night's changes it wasn't that audible but IIRC there were very short dropouts (more like "was there something or was it just my imagination?")
11:47:44 Join desowin [0] (
11:51:51 Join darkless [0] (
11:56:34 Join Wiwie [0] (
12:00:40markunI wonder why saratoga is using C++ style comments to remove some code instead of just deleting it
12:00:49 Part maffe
12:00:51markunthe original code will always be in svn if he needs it
12:01:42scorchethat was brought up before
12:02:01markunwhat did he say?
12:02:14scorche"ok" pretty much
12:02:22scorchehe will start doing so
12:03:31linuxstbmarkun: Yes, I've mentioned that.
12:06:12 Quit desowin (Read error: 104 (Connection reset by peer))
12:08:20 Join desowin [0] (
12:10:35amiconnmarkun: Did you see my blind attemts to enable misalignment abort for gigabeat in the logs?
12:10:52markunno, I didn't
12:11:06amiconnI made a patch and Nico_P tried it, but it doesn't seem to work
12:11:27amiconnFirst I pointed him to the mmu setup, but that's only run in the bootloader
12:12:14amiconnMy second attempt adds it to arm/crt0.S, but for some reason it still doesn't work...
12:12:28markunthere is a wakeup from sleep patch in the tracker which also needs the mmu code to be called in rockbox
12:12:30 Join Nico_P [0] (
12:13:10amiconnEnabling alignment check simply means setting CP15 register 0 bit 1...
12:13:41amiconnI don't know whether this would be a good thing to do in the bootlader, maybe the OF doesn't like it?
12:14:28markunthe OF doesn't get loaded from the rockbox bootloader
12:39:20 Quit TiMiD[FD] ("leaving")
12:47:06amiconnSo, no dualboot on gigabeat?
12:50:36 Join scorche` [0] (i=HydraIRC@rockbox/administrator/scorche)
12:50:40 Join kubiixaka [0] (
12:51:02linuxstbamiconn: No. The Toshiba firmware is so bad, no-one has been motivated to do it...
12:51:43 Quit Wiwie (
12:51:43 Quit lee-qid (
12:51:43 Quit obo (
12:51:43 Quit Nibbier (
12:51:43 Quit pabs (
12:51:43 Quit crashmatrix (
12:51:43 Quit andrew__2 (
12:51:43 Quit tumu (
12:51:43 Quit Farp (
12:51:43 Quit h4ngedm4n (
12:51:43 Quit godzirra (
12:51:43 Quit mtoups (
12:52:27NJoinWiwie [0] (
12:52:27NJoinlee-qid [0] (
12:52:27NJoinobo [0] (
12:52:27NJoinNibbier [0] (
12:52:27NJointumu [0] (
12:52:27NJoinpabs [0] (
12:52:27NJoincrashmatrix [0] (
12:52:27NJoinandrew__2 [0] (
12:52:27NJoinFarp [0] (i=Farp@
12:52:27NJoinmtoups [0] (n=mtoups@HENSON.ISR.CS.CMU.EDU)
12:52:27NJoingodzirra [0] (
12:52:27NJoinh4ngedm4n [0] (
12:52:30 Join petur2 [0] (
12:56:00 Quit desowin (Remote closed the connection)
12:58:12 Join desowin [0] (
13:01:39***Saving seen data "./dancer.seen"
13:06:39 Quit scorche (Read error: 110 (Connection timed out))
13:06:39 Nick scorche` is now known as scorche (i=HydraIRC@rockbox/administrator/scorche)
13:07:32 Quit Wiwie ("Miranda IM! Smaller, Faster, Easier.")
13:08:40 Quit kubiix (Read error: 110 (Connection timed out))
13:08:42 Quit petur (Read error: 110 (Connection timed out))
13:10:40safetydanAnyone think it's worth committing some changes to speex that make it about 10% faster on coldfire? It's just the lazy shotgun approach of ICONST_ATTR any table.
13:11:01 Join Crash91 [0] (n=crash@
13:11:17LloreanIs there any reason not to?
13:12:00safetydanLlorean, well... it's not much of an improvement for a fairly large change.
13:12:18safetydanI'm sure there's far more significant gains to be made elsewhere, but my very limited asm skills aren't up to it.
13:12:27 Part Crash91
13:12:30 Quit HellDragon (Connection timed out)
13:14:06 Join Crash91_ [0] (n=BotsUser@
13:14:23LloreanI'd say do it though. They can have the attribute removed later if it turns out it's better applied somewhere else.
13:14:24linuxstbsafetydan: The only question is if the IRAM is better used elsewhere. If I was you, I would commit it, but with the disclaimer that you don't know if this is the best use of IRAM (assuming that's the case). That way, people looking at speex in the future won't assume your changes are optimal.
13:17:24safetydanlinuxstb, the IRAM is probably better used elsewhere, but what the heck it's not used now so I might as well use it.
13:17:47linuxstbsafetydan: I agree - you should commit.
13:19:02linuxstbjhMikeS: Did you benchmark your changes to the asm lcd_bitmap_transparent_part() function? I'm curious if it's actually better than the C version, in which case it could be used on all colour ARM targets (IIUC).
13:24:47 Join Thundercloud [0] (
13:24:48jhMikeSlinuxstb: no. I never did. I just managed to get an instuction between the load/store and one out of the inner loop
13:26:24jhMikeSI was thinking too that all of lcd-as-*.S is identical between gigabeat and sansa now. They really could share the same file.
13:26:52jhMikeSI think you wanted to do that before. :P
13:30:52linuxstbYes, I already did some cleaning of the gigabeat lcd driver (removing unnecessary gigabeat-specific code), but didn't get as far as writing a benchmarking plugin for that function.
13:31:56safetydanIs the codec stack in iram?
13:32:07jhMikeSthe asm output of the c version is one instruction more than the asm but almost identical. it doesn't turn it into a decrement loop.
13:33:02 Join Xerion [0] (
13:37:40jhMikeSmmm...actually in total, the asm from c is 15 instructions but the asm is 11 taking the outer loop into account
13:37:51 Join spiorf_ [0] (
13:41:27 Join DerPapst [0] (
13:41:45DerPapstHi :)
13:41:58 Join midkay_ [0] (
13:43:01 Quit midkay (Nick collision from services.)
13:43:06 Nick midkay_ is now known as midkay (
13:43:52petur2DerPapst: goodmorning :p
13:44:18 Quit spiorf (Read error: 110 (Connection timed out))
13:44:28DerPapsthey petur2 :D
13:44:38 Nick petur2 is now known as petur (n=petur@rockbox/developer/petur)
13:47:52 Quit JdGordon ("Konversation terminated!")
13:48:30 Quit relaxed ("")
13:52:20 Quit safetydan ("Ex-Chat")
14:10:01 Quit webguest91 ("CGI:IRC")
14:10:37 Join maffe [0] (
14:21:38 Join PaulJam [0] (
14:28:27*petur is listening to a very nice live recording of Björk and hugs the rockbox recording code :D
14:31:11Nico_Ppetur: do you have a special mic for your recordings ?
14:32:08peturyes, I bought a pair of cardoids from Chris Church and also a mic-amp
14:32:50Nico_Pnice... how much did it cost ?
14:34:27peturI paid around 200 euro for preamp and mics, but I got his top preamp model ;)
14:34:37peturand taperssection members get reduction
14:34:58peturand I bought other mics before that so I got even a better price
14:37:06DerPapstGood bye.
14:37:12 Quit DerPapst ("So Long And Thanks For All The Fish!")
14:37:29linuxstbpetur: Do you have any recordings from "taper-friendly" bands? i.e. ones that explicitly allow recordings from their shows to be shared? I'm wondering if you have a recording we could use as a test track to benchmark the codecs...
14:38:38peturdefine taperfriendly... my fav band allows taping but to put the song online... I have to ask...
14:38:45 Join GodEater__ [0] (
14:39:42linuxstbYes, I mean bands that have published statements to say they are happy for recordings to be shared online. I guess that probably means most bands represented on
14:39:52peturlinuxstb: also, they told me that they have nothing against taping, but they can't put it on paper because the record company wouldn't like it
14:40:15linuxstbYes, I think most bands are like that - quietly tolerant.
14:40:43linuxstbI guess we just need to keep searching
14:40:45peturso that probably doesn't make them taper-friendly in the true sense
14:41:17peturbut I can still ask them... it is however a 100% Flemish band so on english lyrics ;)
14:41:21 Join keksrocker [0] (
14:41:23 Quit lee-qid ("aufwiederbyebientotsayonara")
14:49:39 Quit Crash91_ (Read error: 110 (Connection timed out))
14:53:06 Part keksrocker
14:53:44 Quit SirFunk (Read error: 110 (Connection timed out))
14:59:23 Join kaaloo [0] (
14:59:28 Part kaaloo
15:01:39 Join otih_ [0] (
15:01:43***Saving seen data "./dancer.seen"
15:02:56 Join austriancoder [0] (n=austrian@
15:03:05 Quit Nibbier ("bei uns hat noch niemals nicht koana koan durscht ned ghabt")
15:03:20 Join Nibbier [0] (
15:04:04 Quit perplexity (Read error: 110 (Connection timed out))
15:05:03austriancoderpetur: can we meet us later the day?
15:05:44peturyes, after 18:30 if possible and probably better after 20:00 ;)
15:05:53mentit's there any faq how to fdisk/format/upload firmware to CF card to make it ipod-recognizable?
15:07:42peturaustriancoder: how's the stack working btw?
15:07:54austriancoderpetur: okay.. sounds fine..
15:08:44mentpetur: thanks
15:09:10 Join webguest60 [0] (i=c023110a@gateway/web/cgi-irc/
15:10:08 Quit jhMikeS (Nick collision from services.)
15:10:13 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
15:11:04austriancoderpetur: I started with a test gadget driver - serial driver - and got the basics.. execpt... I dont know why my packets dont get send. It would be fine if you have later this doc - chapter 32.14.5.x read... I have done exactly the same.. everything is in IRAM and I dont get a "packet was send". Also we need to talk about the memory stuff.. at the moment the basic buffer for all dtd's is very very p
15:16:32peturok, we'll talk tonight...
15:16:57pondlifelinuxstb: I can donate a test track from my old band if you want... MP3s are at"> but I have the WAV multitracks (24/44.1 mainly).
15:16:59austriancoderpetur: fine...
15:17:19*austriancoder time for a longer break
15:17:26 Nick austriancoder is now known as ac_away (n=austrian@
15:17:52 Join Wiwie [0] (
15:23:21pondlifelinuxstb: Although I'm sure there are other users who could do the same and might be more pleasant to listen to :)
15:24:58 Join SirFunk [0] (
15:32:06 Join lee-qid [0] (
15:32:31linuxstbpondlife: I think the criteria for a test track is that it should be relatively hard to encode - e.g. a test would be to see what bitrate lame -V2 generates. IIRC (amiconn suggested this test), around 240kbps would be considered "hard".
15:33:40pondlifelinuxstb: Don't we need to start with a standard WAV... maybe artificially generated to include 0dB bits and slow fades?
15:33:54pondlifeWith a wide frequency range..
15:34:23 Join saratoga [0] (i=980398fe@gateway/web/cgi-irc/
15:34:33pondlifeIs this for *encoder* profiling, or decoder performance?
15:34:50saratogathats a bit less of an improvement on Coldfire then I was expecting
15:34:52saratogastill nice though
15:35:42pondlifesaratoga: What sort of % is 128kpbs up to now on Coldfire?
15:36:16linuxstbpondlife: It's for decoder performance - both to test optimisations to a decoder, and to try and compare the relative efficiency of different decoders.
15:37:32saratogapixelma said 142%
15:39:26pondlifeWooh, not bad!
15:39:46linuxstbsaratoga: I've been playing a little with the wma_decode_superframe() function, and splitting it into "init" and "decode" functions - so each call to the decode function decodes one frame (2048 samples) instead of the whole superframe.
15:39:53 Quit Entasis ("Leaving")
15:40:54 Quit Wiwie ("Miranda IM! Smaller, Faster, Easier.")
15:42:53 Join webguest09 [0] (i=4cbecdfb@gateway/web/cgi-irc/
15:44:23 Quit webguest09 (Client Quit)
15:46:23 Join amigan [0] (i=dcp1990@unaffiliated/amigan)
15:46:27 Join FOAD [0] (
15:46:29 Join gtkspert [0] (
15:46:29 Join bospaadje [0] (
15:47:21 Join Gibbed [0] (
15:48:23mentchmm, contact apple support... i've must forgotten something
15:49:24linuxstbThat normally means your ipod's firmware partition is corrupted. What did you do?
15:49:34linuxstb(a restore with itunes will fix it)
15:50:51mentdd mbr, fdisk & resize second partition, dd firmware to first partition, mfs.vfat -F 32 second partition, ipodpatcher, copy rockbox
15:51:56 Join Arathis [0] (
15:52:06menti'll try itunes
15:52:50mentmaybe it's because the cf card it's unusually small (128mb)
15:54:33linuxstbWhich MBR did you copy?
15:55:00linuxstbAh, I see you used fdisk to fix it... But maybe you would be better to just create a new partition table from scratch.
15:55:49 Nick Gibbed is now known as Rick2 (
15:56:01mentoriginal mbr from the microdrive
15:56:07 Nick Rick2 is now known as Rick (
15:56:30linuxstbAlso, I'm not sure if all CF cards work in a mini - did you see the MiniCF wiki page?
15:57:43 Quit GodEater_ ("Leaving")
16:00:25Arathispixelma: did I (need to) mention that my h10 freezes about 200% more often since the time you asked me after amiconn's (?) commit?
16:00:26 Quit GodEater (Read error: 104 (Connection reset by peer))
16:00:49 Nick otih_ is now known as otih (
16:05:41 Join GodEater [0] (
16:06:52mentlinuxstb: that page explains many things
16:07:13mentbut now after manual fdisking i got 'dead ipod' instead of 'support folder'
16:08:47markunlinuxstb: could be useful:
16:08:58linuxstbment: Can you paste your partition table layout to somewhere like ?
16:11:03amiconnlinuxstb: Will your change fix the bad yield behaviour of the wma codec?
16:11:36linuxstbThey'll probably be about 10 times as many yields, maybe more...
16:12:58linuxstbamiconn: It also gets rid of that massive 256KB buffer (replacing it with a 8KB one).
16:13:35 Quit Zagor ("Client exiting")
16:17:59linuxstbment: It looks fine to me. What do you mean by "dead ipod" ?
16:18:18mentx x
16:19:29 Quit midkay (Connection timed out)
16:19:41linuxstbWhat are you writing to the first partition? A backup you took from your mini?
16:20:33mentnow i see that i uploaded a different firmware (and also different mbr)
16:20:47linuxstbDoes ipodpatcher recognise your ipod?
16:20:52mentit does
16:21:13mentbut on the old microdrive the firt partition has size 41199 instead of 35309
16:21:30linuxstbAbout 30MB of that should be unused.
16:21:59linuxstbA better firmware to write would be one downloaded via the link on the IpodManualRestore wiki page.
16:22:12linuxstbThat's a copy of the firmware partition, but just the used part.
16:22:45 Join Wiwie [0] (
16:22:48 Quit Wiwie (Read error: 104 (Connection reset by peer))
16:23:27 Join Wiwie [0] (
16:23:32 Quit webguest60 ("CGI:IRC")
16:23:33 Quit Wiwie (Read error: 104 (Connection reset by peer))
16:23:54saratogalinuxstb: thanks for looking into those changes
16:23:59saratogai was going to start on them today
16:29:11mentlinuxstb: i made an exact copy of the original MBR together with original first partition
16:29:47mentlinuxstb: ./ipodpatcher -r / -w, mkfs.vfat and it works (the original firmware booted up)
16:30:37mentbefore that, i was trying to restore from original firmware (from IpodManualRestore) and i got the 'support folder' screen
16:33:25linuxstbment: ipodpatcher does a lot of checks of your ipod's firmware partition and partition table - if it recognises your ipod, then I'm not sure what could be wrong...
16:35:09 Quit markun (Read error: 104 (Connection reset by peer))
16:35:14 Join markun [0] (n=markun@rockbox/developer/markun)
16:36:50mentha, even rockbox is working
16:45:33 Quit daurn ("Cyas later...")
16:45:56 Join perplexity [0] (
16:52:56 Join lazka [0] (
16:53:28 Quit Thundercloud (Remote closed the connection)
16:54:22linuxstbsaratoga: Do you understand the memmove at the end of wma_decode_frame() ? I can't understand why it's needed, but the output is corrupted if I remove it.
16:54:38 Join Soul-Slayer [0] (n=Administ@
16:56:43 Join Thundercloud [0] (
16:57:20linuxstbsaratoga: Ah, I see now - the wma_window function appears to write data into the future...
16:57:37linuxstb(i.e., the next frame)
17:01:46***Saving seen data "./dancer.seen"
17:07:42 Join Psilonaut [0] (
17:23:23 Quit Arathis ("Bye, bye")
17:25:08 Join Arathis [0] (
17:26:32 Join toffe82 [0] (
17:27:02 Join gregj_ [0] (
17:28:34 Quit gregj (Read error: 113 (No route to host))
17:29:42 Quit petur ("reboot")
17:31:48Soul-SlayerI'm looking for a replacement battery for the Gigabeat F20 (I fully charged mine last night, used it today, admittedly using plugins and pitch controls and stuff quite extensively, and it's now empty, after using it for about 3-4 hours max)and the one I have found has a little note under it saying 'Please note that this battery is slightly thicker than the standard battery'... From what I understand, the battery sits under the control pad... Which s
17:31:53 Join ackbahr [0] (
17:32:16markunSoul-Slayer: yes, the thickness might be a problem
17:32:19Soul-SlayerAlso, the hard drive, when it spins up, spins up really quickly and vibrates quite a bit, I'm not sure if that's common to all gigabeats or not, I only got mine recently off ebay
17:32:23 Join rogelio [0] (n=rogelio@
17:32:37dionoeaSoul-Slayer: you sentence was cut after "Which s"
17:32:48Soul-SlayerWhich sounds like it will be an issue, although it does say it's compatible?
17:33:40 Join petur [0] (n=petur@rockbox/developer/petur)
17:33:48 Quit aliask ("ChatZilla [Firefox]")
17:33:50Soul-SlayerDoesn't say it there, but on their ebay shop it says it's slightly thicker
17:34:15 Quit rogelio (Connection reset by peer)
17:35:19Soul-SlayerAnd this is the only place I can find any replacement battery for the Gigabeat
17:35:50markunSoul-Slayer: is the battery really empty or does it just say it's empty?
17:36:03Soul-SlayerI turned it on and it switched back off again
17:36:10Soul-SlayerI was running on 0 for quite a while
17:36:12markunok, sounds pretty empty :)
17:36:30Soul-SlayerLike I said, I bought it on ebay so there is no telling how old it is or how much use it had
17:37:30Soul-SlayerDo you know if this erratic drive spin up is different? It vibrates more than other MP3 players when I access the disk
17:38:01Soul-SlayerDisk itself has no problems reading/writing, so I don't think there is anything wrong with it
17:38:54markunI think all HDDs make more noise when they get older
17:39:44 Join rogelio [0] (n=rogelio@
17:40:30Soul-SlayerFair enough :p
17:40:34 Quit otih (Remote closed the connection)
17:41:06Soul-SlayerKinda worried about this battery life though... Shall run a battery bench on it once it's fully charged again... Although I may need a hand interpreting it :$
17:44:23 Quit GodEater__ ("User pushed the X - because it's Xtra, baby")
17:47:25 Join RaRe` [0] (
17:51:41 Nick gregj_ is now known as gregj (
17:52:04 Quit Arathis ("Bye, bye")
17:56:56 Join ackbahr_ [0] (
17:57:06 Join Arathis [0] (
17:58:05 Join webguest88 [0] (i=4cbecdfb@gateway/web/cgi-irc/
17:59:31webguest88I have just added rockbox to my toshiba gigabeat f40 and it will not allow me to play any songs, help any one?
18:04:06linuxstbAre you trying to play the songs you had on there before Rockbox?
18:04:39linuxstbThat won't work - the Toshiba firmware encrypts all your music files.
18:05:00webguest88oh so i have to load them all again
18:05:13linuxstbYes, without using gigabeat room - just drag and drop them.
18:05:27webguest88k, thanks
18:06:06 Quit pondlife (Read error: 110 (Connection timed out))
18:07:34ackbahr_Hi there! I'm using a H320 Iriver, and I was wondering if anyone was working on the implementation of the microphone function of the line-in connector? And maybe the host USB? Thanks!
18:07:54 Quit RaRe (Read error: 110 (Connection timed out))
18:08:38peturackbahr: usbhost is not working yet, the rest should be working fine
18:08:52peturwhat's wrong with the microphone?
18:10:12peturbeware that the original firmware lets you select line-in or ext-mic but that just changes the gain range. Rockbox gives you one big gain range
18:10:20peturand mic power is always available
18:12:06peturackbahr ?
18:14:56 Quit ackbahr (Read error: 110 (Connection timed out))
18:15:10peturah, that's why...
18:23:20 Quit webguest88 ("CGI:IRC")
18:25:32 Quit PaulJam (Read error: 110 (Connection timed out))
18:27:19 Quit Nico_P (Remote closed the connection)
18:31:10 Join Siku [0] (
18:31:10 Quit petur ("work->home")
18:31:48 Join jgarvey [0] (
18:32:23 Join ackbahr__ [0] (
18:32:29 Nick ackbahr__ is now known as ackbahr (
18:34:24 Join My_Sic [0] (
18:35:56amiconnArathis: Hmm, that's not good :/
18:36:07 Quit obo ("KVIrc 3.2.6 Anomalies")
18:36:16ArathisI'm sure it's not ;)
18:36:24*amiconn would like to know what defaults the H10 loader sets for the appropriate registers
18:37:00amiconnCould you perhaps load a specially prepared build and then tell me some values from the debug menu?
18:38:18 Join chrisjs169 [0] (
18:39:07chrisjs169is it possible to charge the Sansa while still listening to music (by default, the usb logo image appears when a USB cable is inserted, stopping the music)
18:40:12linuxstbTry holding MENU (or whatever button brings up the main menu) whilst inserting the USB cable.
18:41:30 Quit ackbahr ("ChatZilla [Firefox]")
18:47:32 Join Nico_P [0] (
18:49:55 Quit ackbahr_ (Read error: 110 (Connection timed out))
18:50:38Arathisamiconn: I could
18:51:20amiconnAre you able to compile yourself, or should I provide a build?
18:51:24Arathiswould you prefer a clean installation or my default (my config etc. provided by a script)
18:52:03amiconnYou will only need a modified rockbox main binary. Configuration doesn't matter
18:52:09ArathisI'd need to run vmware for that cause I haven't managed to set up a cross environment in ubuntu
18:52:49amiconnOnce you have that build, you will need to replace your usual binary with the special one, shut down (important!), reboot, and then go to the debug menu and read some values
18:53:06amiconnThe modified build will most likely not be able to play music, but that doesn't matter
18:53:24amiconnAfterwards, just put your usual binary back on
18:53:54amiconnJust tell me what H10 build you need (6GB or 20GB)
18:54:00Arathisshould I check the debug values with my normal binary too for comparison?
18:54:13 Join petur [0] (n=petur@rockbox/developer/petur)
18:54:17 Join HellDragon [0] (i=JD@unaffiliated/helldragon)
18:56:48 Quit ac_away ("Kopete 0.12.4 :")
18:57:26 Join lymeca_ [0] (n=lymeca@unaffiliated/lymeca)
18:59:04amiconnArathis: No. The special build will show the values set by the H10 loader (which is what I am interested in). Normal builds touch these registers, so you just see what rockbox writes there
18:59:13amiconnAnd I know what it writes...
19:01:31amiconnThe loader will most likely set the CPU to 24MHz, that's why this build won't play music. Too slow hence a lot of skipping
19:01:47***Saving seen data "./dancer.seen"
19:02:07amiconnI want to know (1) if it really sets 24MHz (2) What other special bits are set
19:03:03 Join dan_a [0] (n=dan_a@
19:05:10amiconnThis *only* contains rockbox.mi4
19:08:31 Join moos [0] (
19:10:41Arathisamiconn: so after I reset the player cause it's freezed again what should I look for exactly?
19:11:10*bluebrother wonders if we should reject FS #7417
19:11:23bluebrotherotoh, it's some kind of "favorites" menu
19:12:45 Quit lymeca (Read error: 110 (Connection timed out))
19:13:10 Join obo [0] (
19:13:24amiconnArathis: Debug->View I/O Ports
19:13:37amiconnI want to know the 6 values below the GPIO values
19:14:17amiconnAfter that, go to 'View HF Info' and check the 'Est. Clock (kHz)' value
19:14:50amiconn..and the exact CPU model
19:15:37 Part godzirra
19:16:44chrisjs169As of a few hours ago, I started noticing a divide by zero error in a custom build - it also happens in the official build (i completely removed the .rockbox dir) - can someone confirm?
19:17:31Arathisamiconn: CLOCK_SRC: 10137647
19:17:31ArathisCLOCCK_0x2C: 00000000
19:17:31ArathisCLOCK_0xA0: 00000000
19:17:31DBUGEnqueued KICK Arathis
19:17:31ArathisPLL_CONTROL: AA020A03
19:17:31ArathisPLL_STATUS: 00000A03
19:17:32***Alert Mode level 1
19:17:32ArathisDEV_0x34: 00001515
19:19:11ArathisEst. clocck (kHz): 08967
19:19:32amiconnNot 80...
19:19:59Arathisoh, sry
19:20:05Arathis80970 atm
19:20:20Arathiswhere can I find the exact CPU model?
19:20:23amiconnInterestingly enough, the H10 loader sets 80MHz at boot
19:20:42amiconnIn the same screen
19:20:59 Quit Nico_P (Remote closed the connection)
19:21:33amiconnPP version:
19:21:45obochrisjs169: checking for divide by zero has only just been enabled for ARM targets
19:22:25amiconnArathis: Thanks.
19:22:38obochrisjs169: do you have a recipe to reproduce it?
19:23:00Arathisif it helps getting rid of these freezes :)
19:23:05amiconnSo this build could even play music, but battery life would be poor (worse than old rockbox builds without cpu scaling)
19:23:06chrisjs169obo: I just removed the entire .rockbox dir and reinstalled rockbox
19:25:23linuxstbWhen does the error occur? Immediately when you boot?
19:25:37obolinuxstb: yes, for me - one of the ata functions
19:26:45chrisjs169linuxstb: about .1 seconds after the Rockbox logo appears
19:26:51 Join Wiwie [0] (
19:26:52oboooo, no I wasn't reading the map properly - fat_mount I think
19:27:06linuxstbThis is just a Sansa issue?
19:27:12oboipod 5g
19:27:33***Alert Mode OFF
19:27:45chrisjs169I don't have any other DAP to test it on
19:28:24linuxstbOK, I can try an ipod build as well now.
19:29:18 Quit My_Sic (Read error: 110 (Connection timed out))
19:29:22linuxstbchrisjs169: You're testing on a Sansa?
19:29:38chrisjs169linuxstb: yes
19:30:21linuxstbMy ipod Color works fine...
19:30:49chrisjs169I tried the Sansa build twice - same resul
19:32:34 Quit Xerion (" ")
19:35:53linuxstbobo: You're saying the problem is in fat_mount() ?
19:36:09 Join Xerion [0] (
19:36:10oboif I'm reading the map correctly, yes
19:36:56linuxstbMaybe you could debug it - i.e. put if statements around any divisions. I'm guessing it happens with unexpected values in the filesystem, so not everyone will experience it.
19:37:30oboyup, I'm about to give it a go...
19:37:44 Join boom82 [0] (
19:38:10 Join webguest82 [0] (i=507935d9@gateway/web/cgi-irc/
19:38:54Soul-SlayerOkay, my gigabeat is fully charged and I want to find out if I have a battery problem, what do you think I should do to test it out?
19:39:11 Join ender [0] (
19:39:11 Quit DarthShrine (Read error: 104 (Connection reset by peer))
19:39:28toffe82Soul-Slayer: battery bench with the default settings
19:39:39toffe82and compare to the result on the wiki
19:39:55Soul-SlayerRight, reset all settings and run an album on repeat?
19:39:59Soul-SlayerOr just one song on repeat?
19:40:11Soul-SlayerDoes volume default to something when I reset it?
19:40:12toffe82an album so there is acces to the hd
19:40:47toffe82I think the volume should go back to default too
19:41:26Soul-SlayerWhy does my gigabeat sometimes not turn on after a charge until I turn battery off and on again>
19:42:06toffe82you charge with th eusb or the charger
19:42:22 Quit tumu (Read error: 110 (Connection timed out))
19:42:32toffe82I don't see why ?
19:42:48Soul-SlayerI'm sure my battery is messed up... No idea how old it is =$
19:43:43 Part webguest82
19:44:46Soul-SlayerRight well, off it goes :p
19:45:15 Join AceNik_ [0] (n=AceNik@
19:45:58AceNik_changename AceNik
19:46:05saratogatwo days without a crash using frequency scaling on the sansa
19:46:16saratogaprobably safe to commit it
19:46:19AceNik_how do u change the name
19:46:23AceNik_on this irc
19:46:48saratogaslash then nick
19:47:02 Join entheh [0] (
19:47:36 Nick AceNik_ is now known as AceNik (n=AceNik@
19:47:43AceNikok thanks
19:47:45 Join austriancoder [0] (n=austrian@
19:47:58AceNikhey markun u thr?
19:49:15austriancoderpetur: ping
19:49:33AceNikcan anyone help me with a headphone problem?
19:49:48peturaustriancoder: just a sec
19:51:34 Quit petur ("switching...")
19:51:43 Join petur [0] (n=petur@rockbox/developer/petur)
19:51:52 Quit AceNik ("bye guys have fun , enjoy !!!!!")
19:55:55 Quit ender` (Read error: 110 (Connection timed out))
19:55:57obolinuxstb: "fat_bpb->dataclusters = datasec / fat_bpb->bpb_secperclus" in fat_mount
19:56:47 Join freqmod [0] (
19:58:05austriancoderpetur: okay
19:58:38peturthe second is over, but I'm getting your patch now
19:59:11 Join sarixe [0] (
19:59:36austriancoderlook at init_dtd(..) and the following two functions
19:59:42sarixei've been keeping an eye on the SVN changes, and i see a lot of references to "Coldfire". what's that?
19:59:48austriancoderin arcotg_udc.c
20:00:15sarixeis that the m68k processor used by some targets?
20:00:20obosarixe: yes
20:00:35peturaustriancoder: is your patch on your site the latest?
20:01:03amiconnfat_bpb->dataclusters = datasec / fat_bpb->bpb_secperclus can of course fail without actually causing a problem, especially on 5.5gen
20:01:07markunsarixe: iriver h1xx, h3xx and iaudio x5, m5
20:01:19sarixemarkun : thanks
20:01:27austriancoderpetur: the generation of the dtd will take place, in arcotg_ep_queue
20:01:52 Quit sarixe (Client Quit)
20:02:06amiconnThis div0 exception is not a good idea imo
20:02:38saratogaamiconn: two days of testing without a crash, so I just enabled your scaling improvements for the Sansa
20:02:57peturaustriancoder: can it be this doesn't apply cleanly?
20:03:53austriancoderpetur: maybe
20:03:58petursomething is wrong with that diff, it is 319KB
20:04:08austriancoderpetur: its normal
20:04:26austriancoderpetur: there is a some unneeded stuff in it
20:04:28peturit even patches the manual ;)
20:06:26austriancoderpetur: for the moment it should be enought to read the patch... I am not at home at the moment where my work is
20:06:47peturit's ok, I'm looking now
20:06:49 Join Domonoky_ [0] (
20:08:19austriancoderpetur: look at init_dtd(..) and two following functions
20:08:50linuxstbamiconn: Why is a div0 exception handler a bad thing?
20:10:19 Join kubiix [0] (
20:10:31austriancoderpetur: and have a look at the datasheet... can you find an error in my code?
20:11:12amiconnlinuxstb: It requires additional checks everywhere where the divisor might be 0. Even if that's perfectly ok (like in fat_mount)
20:11:38austriancoderpetur: the stuff is called in arcotg_ep_queue
20:11:57linuxstbHow can a divisor being zero be OK? What's the expected result?
20:12:18peturaustriancoder: give me some time to get familiar with the code
20:12:28austriancoderpetur: no problem
20:15:06austriancoderpetur: is it okay, when we meet us later or tomorrow.. my girlfriend wants to spend some time with me...
20:15:20peturheh, sure
20:15:39austriancoderin the meantime you can look at the source and tell me what you dont like, what should be improved...
20:15:48peturwillco ;)
20:16:06peturhave fun ;)
20:16:42lostlogiclatest build divide by zero on iPod video :-\
20:17:15austriancoderpetur: thanks... see ya
20:17:17 Part austriancoder ("Kopete 0.12.5 :")
20:17:38obolostlogic: scroll back 20 mins
20:17:45lostlogicobo: thanks
20:18:53 Part boom82 ("poof")
20:21:15amiconnlinuxstb: In this case, the result won't be used. fat_mount() probes partitions as to whether they'll mount as FAT(32). If they don't, disk.c just tries the next one
20:21:51amiconnSo having a div0 exception means there needs to be an extra check which isn't necessary otherwise
20:22:28amiconnMaybe it's not _that_ bad, depending on the number of places where such checks would be needed
20:23:01amiconnBut it also makes rockbox behave different on different targets.
20:23:02linuxstbSo is there no div0 exception on other targets?
20:23:15CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
20:23:15*linuxstb just assumed there was...
20:24:19amiconnThat is, on coldfire it there might be one if the coldfire 'div' instruction fires a div0 exception - not sure
20:25:06linuxstbMaybe it would be worthwhile to keep the check for now (add the check needed in fat_mount), and see how many other problems come up.
20:25:20 Quit kubiixaka (Read error: 110 (Connection timed out))
20:25:46*amiconn wonders how this div0 handler is called
20:26:39amiconnarm doesn't have a div instruction so gcc adds a library routine - how does the div0 handler hook there?
20:27:28 Join Buschel [0] (
20:33:44 Join Arwat [0] (i=4cb431de@gateway/web/cgi-irc/
20:35:09 Part Arwat
20:38:32 Join webguest22 [0] (i=4cb431de@gateway/web/cgi-irc/
20:38:57 Quit webguest22 (Client Quit)
20:41:46 Join Siltaar [0] (
20:42:35alienbiker99i can sing the alphabet backwards
20:43:22alienbiker99i cant sing though
20:43:53amiconnWhy does that remind me of midkay ;)
20:44:02chrisjs169aww...forgot a letter
20:44:04markunhe would sing it upside down
20:44:29alienbiker99its less impressive with text imo
20:46:12 Join hannesd [0] (
20:57:35 Join tumu [0] (
21:01:50***Saving seen data "./dancer.seen"
21:05:10 Join ompaul [0] (n=ompaul@freenode/staff/gnewsense.ompaul)
21:15:22 Quit seclein2 (Read error: 104 (Connection reset by peer))
21:15:57 Join seclein2 [0] (n=scl@
21:25:55 Quit Siku ()
21:32:25 Join kaaloo [0] (
21:32:30 Part kaaloo
21:32:51 Quit Domonoky_ (Read error: 104 (Connection reset by peer))
21:34:44oboamiconn: does fat_bpb->dataclusters need to be set to anything, or is it okay to skip it to prevent a div0?
21:36:40Buschelhi there, does anybody of you know on which cpu-frequency the iPODs run when doing a test-bench via test_codec?
21:37:11amiconnobo: If it wouldn't be needed, it wouldn't be there
21:37:44amiconnIn case of fat_mount, we need to check bpb_secperclus, and fail if it's zero
21:39:03obookay, I'm just wondering why mine booted and seemed to work...
21:41:14 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
21:41:33amiconnAlready checked the disk size reported in rockbox info?
21:41:45amiconnIt's pure luck that your ipod works
21:42:21amiconncluster numbers are range checked in the driver. Out of range accesses will fail
21:43:28 Join My_Sic [0] (
21:46:58obosize and free space are correct - cluster size is reported to be 16384 bytes
21:47:39 Join stripwax [0] (
21:48:07saratogaBuschel: I believe they max out the speed (currently to 78 MHz)
21:49:15 Join Stribbs [0] (i=dcf0795f@gateway/web/cgi-irc/
21:50:09saratogaBuschel: yes they do the entire test boosted
21:50:25StribbsI made sure the directory is in the right place too
21:51:17Buschelsaratoga: thanks, this explains some measurements results...
21:53:53amiconnobo: ANd you removed the calculation?
21:53:56 Quit linuxstb (Read error: 113 (No route to host))
21:54:39StribbsHey, I just tried installing Rockbox on my 5th Gen iPod 30GB and I get a "Divide by Zero" error at boot. (I can't see if this was sent last time)
21:55:20oboI put a if (fat_bpb->bpb_secperclus) above it
21:55:37Buschelstribbs: you've built by yourself?
21:55:51 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
21:56:04Stribbsbuschel: no I've downloaded the latest current release
21:56:16Stribbscurrent build, not release
21:56:39Buschelhmm, then you shoodl either try the build before or build the software by yourself
21:56:55Buschelstribbs: this bug was introduced with some change today...
21:57:14 Join linuxstb [0] (i=5343d4aa@rockbox/developer/linuxstb)
21:57:20oboamiconn: rockbox tries to mount all available partitions at boot?
21:57:21stripwaxtry yesterday's daily build instead
21:57:23StribbsI've never compiled anything before so I'll opt to try an older archive
21:58:43linuxstbStribbs: Today's "archived build" should work fine.
21:58:45amiconnobo: No, it tries to mount the first mountable partition
21:58:52Buschelstribbs: seems to be the best way. I guess the problem is solved soon
21:59:40Stribbslinuxstb: what size is the one you suggest, 1984 KB or 1985 KB?
21:59:43 Quit hannesd (Read error: 110 (Connection timed out))
21:59:58linuxstbStribbs: The "latest" will be fine.
22:00:50Stribbsit's a brand new iPod, I nearly shat myself when I saw "Divide by Zero"
22:01:03Stribbs(thought I just bricked it)
22:01:30linuxstbYou can't brick an ipod by following the Rockbox install instructions.
22:01:51Stribbsthank god
22:02:08 Join My_Sic [0] (
22:03:31Buscheldoes anybody know how the power consumption of an PP-processor scale with the CPU frequency?
22:04:02 Join hannesd [0] (
22:04:21BuschelI've only read 30Mhz -> 75MHz is a difference of roughly 12mA
22:05:09 Join RaRe [0] (
22:15:26 Join DerPapst [0] (
22:16:20DerPapstgood afternoon ;)
22:16:43*petur slaps DerPapst with a large watch
22:17:26 Quit ompaul ("Leaving")
22:18:31bluebrotherpetur: did you recording field test work as expected?
22:18:48bluebrothernice. How was the concert?
22:19:10peturit was ok... not super but nice
22:19:24 Quit desowin ("use linux")
22:19:39bluebrotheryou said it was Björk? Never managed to go to a concert :(
22:19:41peturI was lucky to get through security - they did check
22:20:32*petur is listening to the recording right now ;)
22:20:56*bluebrother likes the old stuff much better
22:21:16bluebrotherthe older albums
22:21:23peturof bjork?
22:21:38peturwell... same here actually
22:22:15bluebrotherVespertine started getting a bit much ... experimental.
22:22:19petur3 of the 18 songs were from last album, the rest was old stuff :)
22:22:27bluebrothercool :)
22:23:55*stripwax saw bjork at glastonbury last month - using one of these
22:24:52DerPapstwoohhoo petur is one of the 6 out of 238 who made it pon the page ^^
22:25:44peturbeen there, taped it :)
22:27:08DerPapstahh.. you'Re on the page because you're a "Top Listener" ^^
22:27:15*bluebrother did the RTC mod yesterday
22:27:32DerPapstdoes everything still work?
22:28:08bluebrotherworks nice.
22:29:30 Join funky [0] (n=repulse@unaffiliated/funky)
22:30:04funkywill rockbox take advantage of the new kernel improvements regarding to power comsumption?
22:30:17saratogawhat new improvements?
22:30:20bluebrotherRockbox is not based on Linux.
22:30:48funky <- saratoga
22:31:06funkythere have been some since 2.6.21 and tickless feature was released
22:31:16bluebrotherfunky: Rockbox is not based on linux.
22:31:28funkybluebrother: I read you
22:31:43bluebrotherah. Thought you missed that.
22:31:49 Join webguest92 [0] (i=4157837d@gateway/web/cgi-irc/
22:31:50funkybluebrother: what is based on?
22:31:57bluebrotherit's written from scratch
22:32:14bluebrotherto quote the front page: Rockbox is a complete rewrite and uses no fragments of any original firmwares.
22:32:16funkyoh oki, I thought was based on ipodonlinux
22:32:35bluebrotherit borrows some code from IPL
22:32:52bluebrotherthe hardware low level stuff mostly.
22:33:07amiconnRockbox already existed back when there where no ipods, let alone ipodlinux
22:33:40amiconnHmm, actually the ipod 1st gen might already have existed...
22:34:17saratogai don't understand why people think rockbox has something to do with linux
22:34:29saratogawhy not assume its based on bsd or windows
22:34:38DerPapstwindows... hahahahaha
22:34:40bluebrotherwindows? Nah, no sources around ;-)
22:35:06stripwaxmaybe the people who find out about ipodlinux first and discover rockbox shares some code with ipl assume rockbox is based on linux..
22:35:06DerPapstwindoes CE.. to make it even more bloat.
22:35:14stripwaxprobably nobody else thinks rockbox is based on linux however
22:35:23webguest92hello, im havingf a bit of a problem with rockbox on my sansa. with a recent change, (the build from yesterday works fine) when it boots into rockbox the screen will turn black and the scroll wheel light stays on, and i have to hold the power button for 15 secs. anyone know what could have caused this?
22:35:37 Quit RaRe` (Connection timed out)
22:35:47 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
22:35:51funkysaratoga: it wouldnt be so extrange, lots of other projects are based on linux
22:35:52 Join webguest02 [0] (i=4a3361a2@gateway/web/cgi-irc/
22:35:54 Join RaRe` [0] (
22:36:49funkysorry for my english :P
22:36:51webguest02rockbox will not start backup after the latest svn updata. this was on the sansa e280
22:37:19stripwaxfunky - linux is not the ideal solution for an embedded device with realtime requirements and limited cpu power..
22:37:53 Join Domonoky_ [0] (
22:38:38funkystripwax: I dont doubt that, it wasnt created bearing in mind those concepts, although there is a RTC linux kernel
22:39:05funkybut I think is mostly used for audio composers
22:39:08 Part webguest02
22:39:11stripwaxfunky - really?
22:39:28funkystripwax: are you being sarcastic ?
22:40:47stripwaxfunky - :) no, (although it's off topic), expressing surprise. which apps?
22:40:48bluebrotherreal time linux has quite some usage fields. Most of them are not audio at all ...
22:40:52 Quit webguest92 ("CGI:IRC (Ping timeout)")
22:42:23 Quit rogelio ("Kopete 0.12.4 :")
22:43:30funkystripwax: <- projects
22:43:30 Quit RaRe (Read error: 110 (Connection timed out))
22:44:02funkyanyway I knew about it through an audio composing distro
22:44:26stripwaxfunky - ah, maybe I misunderstood. I thought you were saying realtime linux is the thing that is used most by audio composers (compared to the alternatives). agree that realtime linux has use in audio field
22:44:43peturthe title 'real-time' all depends on what latencies you can have... I've done a project once that was real-time and rt-linux was waaay too slow ;)
22:45:38funkyhehe, well gtg
22:45:45*Domonoky_ thinks "real-time" only means you can predict how long it can take worst time :-)
22:45:51*stripwax agrees
22:45:54funkya gentoo installcd and amd64 X2 are waiting for me
22:46:02stripwaxhave fun!
22:46:13funkythanks, see you later
22:51:24 Part Domonoky_
22:52:03 Quit amiconn (Read error: 110 (Connection timed out))
22:52:03 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
22:52:22 Join webguest72 [0] (i=4157837d@gateway/web/cgi-irc/
22:52:38 Join Janascho [0] (i=d958a88c@gateway/web/cgi-irc/
22:52:58 Quit SirFunk (Read error: 113 (No route to host))
22:55:44 Quit Buschel ()
22:55:45 Quit HellDragon (Nick collision from services.)
22:56:20 Quit saratoga ("CGI:IRC (Ping timeout)")
22:57:22 Quit webguest72 ("CGI:IRC (Ping timeout)")
22:58:54 Join HellDragon [0] (i=JD@unaffiliated/helldragon)
23:00:24 Join RaRe [0] (
23:00:41lazkawebguest02 is right, latest SVN build for sansa doesn't work
23:00:52lazkahangs at startup
23:00:52 Quit Janascho ("CGI:IRC (EOF)")
23:01:01 Join Robin0800 [0] (
23:01:46 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
23:01:52***Saving seen data "./dancer.seen"
23:01:55 Part Llorean
23:04:28lazkaamiconn: ping ;)
23:06:18 Part maffe
23:07:29 Nick lymeca_ is now known as lymeca (n=lymeca@unaffiliated/lymeca)
23:09:10Robin0800just updated to latest build now ipod won't start just sayes Divide by Zero at005438c can the divide by zero check be reverted?
23:10:55 Quit linuxstb ("CGI:IRC")
23:12:17 Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
23:12:40 Quit Bawitdaba (Read error: 111 (Connection refused))
23:12:49 Quit bagawk (Read error: 110 (Connection timed out))
23:12:58saratogalinuxstb: is there a fix in the works for the current sansa build issues?
23:13:12 Quit Psilonaut ("Gone trippin")
23:13:21linuxstbobo: Are you planning on committing a fix to fat_mount() ? Does everything work OK with the divide by zero check apart from fat_mount() ?
23:13:22 Join mirak [0] (
23:13:30linuxstbsaratoga: I only know what's been said in IRC.
23:13:34saratogai'm about to revert it on my own svn check out, and i'd like to know what I should be doing when I commit stuff tonight
23:14:10linuxstbHow do you normally commit? You can just type "svn commit" from within the apps/codecs/libwma directory if that's all you want to commit.
23:14:29saratogathanks, didn't realize that
23:15:16linuxstbYou can also just commit specific files - e.g. svn commit -m "message" apps/codecs/libwma/wmadeci.c, or (from within apps/codecs/libwma) svn commit -m "message" wmadeci.c
23:15:35linuxstb(you can specify multiple files as well)
23:15:51 Join maffe [0] (
23:16:16linuxstbRobin0800: Either the check will be reverted, or the divide by zeros will be fixed...
23:16:17 Join SirFunk [0] (
23:16:59obolinuxstb: I'm not exactly familiar with the fat code.. but saying that I haven't noticed any other issues
23:17:19 Quit saratoga ("CGI:IRC (EOF)")
23:18:44 Quit RaRe` (Read error: 110 (Connection timed out))
23:19:13 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
23:20:22linuxstbWhat did you do - return from the function if the value is zero, or just skip the division?
23:20:53amiconnIf you skip the division, fat_bpb->dataclusters will be wrong
23:21:03oboreturn if zero
23:21:06amiconnThis is the only statement that sets this struct member
23:21:09 Quit davina ("xchat on Ubuntu 7.04")
23:21:28amiconnReturn if zero is okay, as this can't be valid fat
23:21:58obookay, I'll commit that
23:21:59amiconnShould return with -2 as the other return codes are taken, and it's similar in meaning to the other -2 cases
23:22:34*amiconn still wants to know how this div0 exception handler is hooked
23:26:56amiconnAh, now I understand...
23:27:41*DerPapst wonders why amiconn still asks questions if he figures it out himself 10minutes later anyways...
23:28:38scorcheDerPapst: saying things outloud typically helps the thought process
23:29:21 Join FOAD_ [0] (
23:29:58pixelmascorche: so true...
23:30:06peturasking questions too
23:30:21scorcheeven if the other person doesnt respond
23:30:35 Quit freqmod (Read error: 104 (Connection reset by peer))
23:30:40peturif I would count the number of times I find something the moment I ask the question....
23:30:54 Quit amiconn (" gotta reboooot")
23:31:21*petur hopes he won't have a div-0 at boot
23:32:38 Quit moos ("/me says moo")
23:32:46*linuxstb notes the Sansas have FAT16 support enabled, and there are more divisions there...
23:34:32 Join amiconn [0] (
23:34:51*linuxstb also wonders why the Sansas had this problem - the first partition should be a valid FAT partition (it's the second partition on ipods)
23:35:20oboAlso at least one report just mentioned a freeze at boot, without a div0 error
23:36:02linuxstbWe'll see what happens...
23:36:03DerPapstChuck Norris can devide though 0
23:37:07oboI've just booted using a FAT16 enabled build without any problems...
23:38:12amiconnobo: A fat16 enabled build won't use its fat16 code paths unless the first partition found is a fat16 partiton
23:38:49amiconn(it's a bit different with multivolume + hotswap of course)
23:43:39scorchewow...i havent ever seen one of these on craigslist...
23:43:40 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
23:45:34 Quit FOAD (Read error: 110 (Connection timed out))
23:48:15Robin0800thanks all ipod working now
23:49:13scorche I am so tempted to put a link to the guidelines and say PLZ READ
23:49:50*scorche would if he wasnt already getting hatemail...
23:53:20*amiconn wonders whether scorche is doing a space bar stress test ;)
23:53:35scorcheperhaps... cut
23:53:45scorche perhaps...
23:53:48scorchebetter =)
23:54:10 Quit mirak (Read error: 110 (Connection timed out))
23:54:27petur heh
23:54:47scorche at least this is easier than typing backwards...
23:55:31petur not when you try to align it ;)
23:55:47*amiconn uses a proportional font
23:55:50scorche but now we have a precedent to copy
23:56:01*petur thought amiconn did :p
23:58:53 Join mirak [0] (

Previous day | Next day