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 2005-05-11

00:00:03amiconn33 KB
00:01:31amiconnXShocK: The replaced functions seem to be not completely useless, at least with the 1.65 firmware... (thinking about the .ogg playback problem)
00:01:54XShocKamiconn: i might be a little bit outdated. :)
00:02:57rasherindeed, something fishy's happening on 1.65
00:04:30amiconnrasher: Would be interesting to compile the codec plugins with sectioned compilation + linker garbage collection. There should be quite a number of unused functions...
00:04:38 Nick tvelocity[away] is now known as tvelocity (
00:05:02rashersounds like a bit over my head.. at least while I'm still in bed
00:05:10rashernursing my poor throat
00:05:21rasherand general being-sick-ness
00:05:40*HCl bites i ninja
00:05:46HClgod that game is way too hard >/
00:06:45XShocKamiconn: final hex becomes around 90kb bigger
00:07:03XShocKthan the original
00:08:00XShocKto be exact 111,834kb
00:10:44XShocKi don't really understand why we need original firmware anyway, better solution would be just to have a failsafe mode + loading orig. frimware from disk.
00:12:18XShocKcompressed version of original version is 800kb
00:13:34 Join arfo_ [0] (
00:14:23amiconnQuite big compared to the archos firmware... but then it's multiple codecs, all in software
00:14:49 Part arfo_
00:15:01amiconnArchos firmware is 60...150 KB uncompressed depending on model
00:15:34 Join arfo [0] (
00:16:03rasherand lots of bitmaps :D
00:19:03 Part arfo
00:23:25austriancodersome c++ guy here?
00:24:44 Quit Sucka ("a bird in the bush is worth two in your house")
00:25:32austriancodercompile problems:
00:27:04 Quit ashridah ("Leaving")
00:29:52XShocKaren't all template stuff should be in h?
00:31:08XShocKi can't understand how it would figure out right values when it compile "cpp:"
00:31:18***Saving seen data "./dancer.seen"
00:31:31XShocKsince they are known only when compiling "other cpp:"
00:32:04XShocKplease don't kick me in a **** if i am wrong. :)
00:33:06amiconnIt seems my RLD problem analysis was indeed correct. :)))
00:33:11austriancoderstack.cpp should be ok...
00:33:31Bagderamiconn: absolutaly great amiconn!
00:33:59amiconnI tried playing some music on my Studio, constantly shaking it meanwhile
00:34:30preglowamiconn: what?
00:34:31amiconnThe reload stalled as long as I kept shaking, for >15 sec. When I stopped shaing, it finished the reload
00:34:31 Quit cYmen (Read error: 113 (No route to host))
00:34:34preglowi was watching a movie, heh
00:34:51amiconnNot even a gap in playback :))))
00:35:22amiconnpreglow: Question solved already
00:35:43rasherso RLD Is No More
00:35:55rasherwell, won't have long to live anyway
00:36:31amiconnrasher: RLD is our Most Famous Bug
00:36:36preglowred led death?
00:36:37HClis or was?
00:36:48amiconnWas, now ;)
00:36:52HClgood :P
00:36:53Bagderit'll remain the most famous even after fixed ;-)
00:37:03preglowi haven't heard much about it
00:37:06preglowwhat is/was it?
00:37:07Bagderwe'll talk about it around the camp fires
00:37:19amiconnrasher: It's been there for years now. See the very first entry...
00:37:21HCl"back when i was young, we used to have the rld bug!"
00:37:22rasherpreglow: subscribe to the damn mailinglist already
00:37:30preglowi've been on it for ages
00:37:46rasheroh ah
00:37:50preglowi don't read archos related posts
00:37:54amiconnHowever, the reason for the bug changed
00:38:17rasherI find myself reading most of the posts
00:38:31amiconnFirst, it was that the hitachi disk simply won't answer if they time out. They need a (soft) reset, which Linus added....
00:39:08amiconn...but accidentally used a mutexed function for that, calling it from within a mutexed function...
00:39:52rasherhilarity ensued
00:40:07amiconnFix committed.
00:40:32rasherthis mandates a digest entry
00:41:43Bagderpossibly await testing results from some of the frequent victims of this bug
00:42:09amiconnI know a quite reliable way, just it will take a while
00:42:42amiconnI need to run my player in the car, and run out of batteries while driving. This _always_ ended in an rld before...
00:43:50rashertalk to me about running out of batteries taking a while
00:43:57amiconnI knew there was a reason why I kept the 10 GB hitachi hd in the player and didn't swap it for the 20 GB IBM...
00:44:02rasherthose iriver battery tests were the height of boredom
00:44:48preglownow let me tell you of a little thing called imdct
00:45:06amiconnDid you find something about the slowdown?
00:45:46rasherthat is a horrible tale of misfortune :<
00:45:49preglowi could try to work on it some more now, but i'm a couple of notches too tired
00:47:10preglowi suspected the code was misaligned, but no
00:47:49preglowlinus said he'd shove it through the bdm one day, perhaps that'll clear some things
00:51:49HClsounds nice
00:51:55 Join lostlogic [0] (
00:51:58HClan gdb stub would be good too
00:52:14preglowmyes, i'd actually consider finding a fancy way to make use of that serial port
00:52:25preglowsince i more or less always have to test my code on target :/
00:52:35HClwe already have a gdb stub for archos
00:52:38HClno idea how portable that is
00:55:17*preglow grumbles and makes a cup of coffee
00:57:19HClpoor preglow :/
00:57:27 Quit Aison ("( :: NoNameScript 3.72 :: )")
00:57:29*amiconn prepares to run a really extensive set for hunting the recording problem on archos
00:58:31amiconnThis will keep my recorder busy for weeks...
00:59:13HClwhat recording problem?
00:59:16preglowthe rigours of development
00:59:36MoosCamaroIt's a pity amiconn don't work in iriver :(
01:00:37HClonly because he doesn't have one
01:00:55*HCl does an cvs update and finds something that he thought he committed days ago o.o;
01:00:58*HCl commits.
01:01:42amiconnHCl: There is a problem with recording on archos. With high quality settings and high sample rate, the recording 'breaks down' at some time within the recording, producing only corrupt frames after that point
01:02:01preglowand it's not a mas problem?
01:02:07amiconnIt might be either a problem of our transfer routines, or a MAS problem
01:02:26preglowlet me tell you how eternally gratefull i am that i will never write a codec for that chip
01:02:32HCl :)
01:02:36amiconnI have a suspicion, but to verify if, I need to do test recordings with all possible combinations...
01:02:51HClwhats your suspicion?
01:02:51amiconnpreglow: This is harvard....
01:03:05rasheramiconn: better write up a matrix sooner rather than later :(
01:03:08amiconn....and the test recordings need to be longer ones
01:03:13preglowamiconn: all dsps are harvard
01:03:16preglowamiconn: all i know of, at least
01:03:54amiconnpreglow: There are some docs... from looking at them, writing code is really an ugly piece of ****
01:04:13preglowamiconn: yes, my statement is based on having read those docs
01:04:22amiconnIt's harvard, has only little ram, and you need to synchronise the ops manually...
01:05:35amiconnHCl: My suspicion is that the MAS cannot handle m/s stereo properly at higher quality settings
01:06:12amiconnArchos firmware recording doesn't have this problem, and it uses simple stereo, not m/s....
01:06:30preglowwell, why don't you try feeding it difficult frames?
01:06:47amiconnTalking encoding here...
01:06:59amiconnBut then archos firmware does have another problem with long recordings... it sometimes simply freezes
01:07:07 Join |it|ChasKi` [0] (
01:07:09preglowyes, i know we're talking encoding
01:07:11 Nick tvelocity is now known as tvelocity[away] (
01:07:27amiconnSo how do you mean, feed it difficult frames?
01:07:37amiconnFeeding white noise, or what?
01:07:40preglowdata that's hard to encode
01:07:51preglowdepends on what you mean by 'higher quality settings'
01:08:03preglowwhite noise isn't necessarily that hard to encode
01:08:43amiconnThe MAS has only a rough method to set the desired average bitrate, a quality value, ranging 0..7
01:09:23amiconnMy idea is to test all possible combos of quality setting, sample frequency, mono/stereo
01:09:44amiconnAll this with and without m/s stereo enabled
01:10:18amiconn...doing at least 5 recordings of 2 h each, the check the mp3 files with mp3utility or similar
01:10:33amiconnI can probably cut short at some points
01:10:34HClsounds like a hell of a task, heh
01:10:35HCl :x
01:11:10amiconnLike, when going down with the quality value, and q=3 doesn't produce any errors, I can skip doing 2..0
01:11:40preglowwhy don't you use ordinary stereo at higher quality settings anyway? i believe that's the usual practice
01:11:41rasherbinary bugtesting
01:12:11amiconnpreglow: Even the highest quality setting isn't really high quality mp3 wise
01:12:19preglowdoesn't surprise me
01:12:28preglowdecent encoding is very resource intensive
01:12:37amiconnThe mas won't produce any frame with bitrate >192 kbps, and the average is ~175 kbps
01:12:53preglowit probably doesn't have enough ram for such frames, heh
01:13:06amiconnIt does for decoding...
01:13:34preglowencoding is quite a bit worse than decoding
01:13:41amiconnIt even handles the highest mp2 bitrate properly, which is 384 kbps
01:14:01preglowyeah, the frames themselves aren't very large
01:14:03amiconnpreglow: Yes, but the ram is the same
01:14:20amiconnThe codec itself doesn't reside in ram, but in rom
01:16:23amiconnMy guess for the frame size limit on encoding would be the limited processing power
01:16:49preglowmight be
01:17:47amiconnThe mas is running at 18.4 MHz
01:18:03preglowit's a frigging miracle it gets anything done at all
01:18:06amiconnTry to beat that with libmad ;)
01:18:57preglowif i double the efficiency it'll be realtime at 30mhz or something
01:19:06preglowand i'll be very, very hard pressed to do that
01:19:17 Quit crashd (Read error: 101 (Network is unreachable))
01:21:27 Join crashd [0] (
01:22:02 Quit MoosCamaro ()
01:22:58*HCl didn't understand the last sentence of preglow :/
01:23:02HClhard pressed? what does that mean :X
01:24:27preglow'hard pressed' means pretty much that i'll have a hard time in doing something
01:28:17HClgee, that was easy, says man, and procedes to prove that black is white and gets himself killed at the next zebra crossing...
01:28:26HCl*is adding startswith and endswith to the searchengine*
01:39:35HClstartswith works :)
01:41:09Aditya_I got cygwin working..
01:41:17Aditya_but the install for source talks about archos
01:43:03HClendswith works too
01:43:33*HCl commits
01:46:04 Quit _aLF ("Leaving")
01:46:46 Quit XShocK (" HydraIRC -> <- The future of IRC")
01:48:09 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
01:48:39 Join asdsd____ [0] (
01:53:31DBUGEnqueued KICK asdsd____
01:53:31asdsd____,00hey guys
01:54:37asdsd____,00whats the status on the firmware?
01:54:41***Alert Mode level 1
01:54:41asdsd____,00u guys working hard on it?
01:57:59rashereurgh.. could you please shut off the colour?
01:58:08preglowstill using that piece of shit client are you? :P
01:58:34Aditya_I am using gaim
01:58:43Aditya_filters out formatting/color..
01:58:49Aditya_so nyeah
01:59:14***Alert Mode level 2
01:59:14asdsd____,00lol hey preglow u remember me lol
01:59:33preglowyes, somehow i do remember you, heh
01:59:42preglowi've learnt how to strip the colours in my irc client since then
01:59:49***Alert Mode level 3
01:59:49asdsd____,00great hahaha
01:59:58***Alert Mode level 4
01:59:58asdsd____,00hey i went to compusa lately
02:00:04***Alert Mode level 5
02:00:04asdsd____,00and i saw an mp3 player called rockbox
02:00:09tvelocity[away]asdsd____, /remote off, please:P
02:00:23***Alert Mode level 6
02:00:23asdsd____,00i was like no way they ported their firmware to their own mp3 player!
02:00:28***Alert Mode level 7
02:00:28asdsd____,00remote off?
02:01:07tvelocity[away]it buuuuuurns
02:01:42***Alert Mode level 8
02:01:42asdsd____,00hahaha ok ill log in using mIRC, this is efnet right?
02:02:11***Alert Mode level 9
02:02:11asdsd____,00ah ok, brb
02:02:13 Part asdsd____
02:02:51 Join RichP [0] (
02:02:51RichPk so anyhow
02:03:02RichPits amazing how they totally ripped off ur name
02:03:16RichPi figured rockbox was copyrighted,
02:03:19rasherit's not such an original name
02:03:27RichPbut i guess those mp3 players makers can sue u guys now
02:03:45rasherthe wikipage even has a list of other products named rockbox
02:03:47rasherthey can't
02:04:11rasherthey'd have no legal leg to stand on at all
02:04:16RichPah okay
02:04:18RichPnever mind then
02:04:26RichPso whats the status on the iriver?
02:04:41RichPu guys busting ur ass off?
02:05:10RichPoh gawd they play this song so much on the radio; the 'ipod shuffle' song
02:05:47preglowi'm self administering some good ambient at the moment
02:06:17RichPhow so?
02:06:23rasherSuch a lawsuit would warrant this response:
02:06:25tvelocity[away]ambient... w000t :)
02:06:36 Part Aditya_
02:06:39 Join Aditya_ [0] (
02:06:56preglowtime to bed soon
02:07:45RichPhey did u guys see the gboy that was ported to the psp?
02:08:40RichPtheirs is in color ;-)
02:10:14HClwhats that?
02:10:24RichPlol omg!!!!!!
02:10:27tvelocity[away]playstation portable?
02:10:29RichPwhat rock have you been living under?
02:10:30Aditya_wow Hcl
02:10:36Aditya_just.. wow
02:10:38RichP*moment of the week*
02:10:41HClone in which i don't really care for popular shit :P
02:10:52HCli don't care for game consoles unless they're hackable
02:10:54HCllike my xbox
02:11:00RichPits the Sony Playstation Portable handeld gaming system with ultra super cool graphics
02:11:01tvelocity[away]yeah, PSP is also known as "a piece of crap":P
02:11:10RichPoh thats the beauty , i totally agree with u HCl
02:11:11HClbut ok, that thing
02:11:22RichPand as far as i can tell the PSP is gonna be so pirate/hacker friendly its not even funny
02:11:58RichPin fact im leaning towards the psp instead of the nds simply because it has usb/infra red and they just cracked the umd format it reads the disks on
02:12:12***Alert Mode OFF
02:12:16tvelocity[away]hm... i would buy an DS...
02:12:22RichP+ it uses the iriver's 5v charger, but i sitll wish i had one :-(
02:12:49HClautorating added to the searchengine
02:13:00HCli pondered buying an gamecube and hack into it over network
02:13:05HClbut its cumbersome
02:13:27HCljust like how the ps2 is a total bitch to hack
02:13:33RichPi totally love how any xbox can be modded in under 5 mins now lol
02:13:35CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
02:13:35*HCl patpats his xbox.
02:13:37HClyup :p
02:13:47HClwell, i didn't softmod mine
02:13:51RichPneither did i
02:14:02*HCl has a nice limited crystal version case thing..
02:14:04RichPi remember when u had to telnet into it and do all this weird stuff to it
02:14:19RichPah u punk, well i painted mines silver :-P
02:14:30HCli just soldered the writing pins to the tsop, and reflashed it
02:14:31HClworks great.
02:14:41HCltotal obedience :p
02:14:46RichPto the umbrella thing right?
02:14:47HClgot any good games i don't have?
02:14:52HClumbrella thing?
02:15:11RichPi got 64+ games for it
02:15:34 Quit Strath (Read error: 110 (Connection timed out))
02:15:35RichPi can send u a list, brb
02:15:36HCli'm still searching forrr.... dead or alive 3.. and..
02:15:46RichPi got doa:u
02:15:55HClyea i got that one
02:15:58HClbut its not the same as 3
02:16:04RichPMadden NFL 2005
02:16:04RichPNFL Street 2
02:16:04DBUGEnqueued KICK RichP
02:16:04RichPNBA Live 2005
02:16:04RichPNBA Street Vol. 2
02:16:05***Alert Mode level 1
02:16:05RichPNBA Street Vol. 3
02:16:07RichPFIFA Soccer 2005
02:16:09RichPFIFA Street
02:16:11RichPTony Hawks Underground 2
02:16:12HClnot in #rockbox
02:16:13HCl :X
02:16:15RichPDoA Beach Volleyball
02:16:17RichPSSX Tricky 3
02:16:19RichPMLB Slugfest Loaded
02:16:23RichPmy bad
02:18:58amiconnHCl: Did you really use the same token id twice? (# 12)
02:20:01HClat least its correct in the code :p
02:20:04HCllet me fix the wiki
02:22:47RichPhave u guys gotten greyscale to the gboy yet?
02:23:33preglowit's not in cvs
02:24:31HClgrayscale in rockboy has been working for ages
02:24:33HClits just not in cvs
02:24:45tvelocity[away]why not?
02:24:53HClcause the grayscale patch isn't finished.
02:25:22tvelocity[away]ah i see
02:25:28RichPwhat's left?
02:25:53HCllots of stuff...
02:26:05HClany more requests for search engine capabilities?
02:26:06***Alert Mode OFF
02:26:12HClaside from the ones planned already
02:28:08RichPweeeee, my BT client works again
02:28:12RichPi can download forza now
02:29:28CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
02:29:28*HCl found the rename option on the wiki :d
02:29:49HClits about time that the description of the new database gets a proper wiki name xD
02:31:19***Saving seen data "./dancer.seen"
02:36:14*HCl goes to sleep, night.
02:37:43*rasher tries to put his mind into functional mode
02:37:50rasheras in, functional programming
02:38:03rasherI know ><
02:38:08*austriancoder is happy
02:38:14*HCl 's never done that, but knows its scary without variables and stuff.
02:38:19*austriancoder has finished coding for uni
02:38:25HCl :3
02:38:33rasherthere are variables, but
02:38:49austriancoderwill go to bed tooo
02:38:52austriancodergood night
02:39:09 Part austriancoder
02:45:31 Quit preglow ("fjone........")
03:29:34 Join |it|ChasKi` [0] (
04:05:25 Join QT_ [0] (as@area51.users.madwifi)
04:09:22 Quit QT (Read error: 60 (Operation timed out))
04:11:54 Quit RichP ("I-n-v-i-s-i-o-n 2.0 Build 3515")
04:13:37 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
04:31:20***Saving seen data "./dancer.seen"
05:31:23 Join |it|ChasKi` [0] (
06:03:18 Quit Aditya_ ("Download Gaim:")
06:11:45 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
06:31:24***Saving seen data "./dancer.seen"
06:38:42 Quit AEnertia (Read error: 104 (Connection reset by peer))
07:06:11 Join LinusN [0] (
07:18:22*LinusN reads the irc log and feels stupid for using ata_soft_reset() instead of perform_soft_reset()
07:19:03Bagderfix of the month, really
07:27:11LinusNhow silly
07:27:14LinusNof me
07:33:14 Join |it|ChasKi` [0] (
07:42:34LinusNrasher: (from your question in the irc log) the short-press bootloader bug should be pretty easy to fix
07:55:30LinusNmorning amiconn
08:06:03 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
08:07:56 Join Theking^ [0] (
08:09:26 Quit Theking^ (Client Quit)
08:11:05LinusNelvis has left the building
08:14:13 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
08:31:26***Saving seen data "./dancer.seen"
09:07:57 Join asgsa [0] (~asgas@2f8bef865366ff44.node.tor)
09:12:04LinusNamiconn: did we come to a conclusion about the C version of the recording transfer loop?
09:26:48*t0mas had a 30 minutes meeting...
09:27:03t0masstarting with the question to everybody to turn their phone off...
09:27:59t0masand in the middle of the meeting the phone of some girl from Microsoft Netherlands rang...
09:28:06 Join ashridah [0] (
09:28:08t0masand se was like "Eh... I turned it off..."
09:28:16t0mas5 minutes later... same phone again...
09:28:51t0masand again 5 minutes later... guess what... het phone again... then some guy laughed for about 2 minutes...
09:29:14t0masand then he was able to say what he wanted to say... "Eh... [name] you ever heard of the snoze function?"
09:29:25t0masturned out to be her alarm clock...\
09:29:55t0mashow stupid...
09:43:04 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
09:45:13 Join Harpy [0] (
09:53:01 Join Lost-ash [0] (
09:56:37 Join |it|ChasKi` [0] (
10:01:41rasherLinusN: Sounds great, because it's not only a problem with the boot-iriver option, but also would be when dealing with a locked remote of course.. hurray
10:03:05 Join Aison [0] (
10:03:55rasherI thought it was a problem with how fast you could read the buttons, but when iRiver can - so can we (you).. at at least, it's possible :)
10:04:23LinusNthe thing is that the bootloader reads the keys after the entire initialization is done
10:04:31LinusNwhich can take quite some time
10:05:11LinusNthe bootloader isn't exactly top prio for me right now
10:05:26LinusNbut it needs an update
10:05:35rasheryeah, it's not like it's a huge problem
10:05:47rasherthe 1.65 thing is a bugger though
10:06:19 Quit ashridah (Nick collision from services.)
10:06:21 Nick Lost-ash is now known as ashridah (
10:06:26ashridahwhat's the prob with 1.65?
10:06:57rasherSome things in the iriver firmware stops working
10:07:01rasherogg playback for one
10:07:04LinusNit fails to play Ogg, for example
10:07:14*ashridah takes mental note: no upgrading for him
10:07:34ashridah3/4ths of my music'd be useless
10:08:52 Join sox [0] (
10:10:00soxok, time to ask the classic question
10:10:08LinusN"when is it ready?"
10:10:36rasher"to be or not to be?"
10:10:39soxyep, sort of, or at least a functioning libmad in cvs.....
10:10:43Bagdernoooo, "when will you guys go get more coffee?"
10:10:49LinusNlibmad works perfectly
10:10:51Bagderand the answer is... "now!"
10:11:11ashridahheh, "how drunk should i get"
10:11:33soxso when is it usable
10:11:36LinusNit's the music playback code that isn't done
10:12:03LinusNi have been busy with real-life stuff lately
10:12:14soxhow fast is realtime playback nowadays of mp3s?
10:12:26LinusNabout 200% methinks
10:12:48LinusNsort of, but i'd like it to be faster
10:13:24 Join merbanan [0] (
10:13:25soxis it only you who can do the playback api
10:13:36rasherpoor preglow :(
10:13:49soxpoor LinusN...
10:13:51t0masthere are others sox... but he is the only one doing it...
10:14:04t0masand he's the best at it ;)
10:14:36soxif he's the only one, i agree his the best!
10:15:01BagderI guess I and Linus should have another dev night again ;-)
10:15:03t0masBagder is working on it too... iirc
10:15:18soxBagder: I agree, you should!
10:15:33t0masBagder: if I were you I would do that in the weekend ;)
10:15:34soxLot's of stuff happened last time you did...
10:15:46t0masas working a day after a dev night is a bad idea
10:15:57Bagderit has worked before ;-)
10:16:09t0maswhat's your job?
10:16:12t0masbed testing? ;)
10:16:20soxa working day after a dev night is like cleaning up after a party...
10:16:24Bagderthat hurts
10:16:46sox"who forgot this piece of code here"
10:16:51LinusNBagder, LinusN and Zagor works at the same company
10:16:52t0masBagder: you can dev all night, and then just continue devving at work?
10:17:00LinusNt0mas: of course!
10:17:22Bagderthe dev nights are usually not the whole night, just a late night
10:17:30t0masok, then it's possible...
10:18:00t0masI was thinking of working till 8:00 on rockbox, then eat something and start working again on something else..
10:18:13t0maslike: "wtf... how does he do that.."
10:18:24LinusNdrugs, lots of them :-)
10:18:34t0mastime to move to Holland then :P
10:18:51soxthanks for answers, guys, see you around
10:19:13 Quit sox ("Snak 4.13 IRC For Mac -")
10:20:39*HCl rubs his eyes and listens to his beeping alarm clock
10:21:04t0masdamn.. you get up now?
10:21:32*t0mas already had a meeting... and wrote a tiny piece of his SSH articel
10:31:29***Saving seen data "./dancer.seen"
10:37:32amiconnLinusN: Which C recording loop do you mean? The original one before my asm stuff, or my alternate one with the timeout?
10:37:44LinusNthe alternate
10:38:20LinusNthere is so much talk about corrupt recordings nowadays
10:40:15amiconnI know.
10:40:26amiconnI guess you read about my intention
10:40:46amiconnThe alternate routine seems to work a little better... sometimes
10:41:06amiconnBtw, good news from the rld front:
10:41:43HClt0mas: yes :P
10:41:47amiconnI used my Studio in the car this morning... and it ran out of batteries
10:41:54HCland no rld?
10:42:16BagderI read longhorn is introducing a red screen of death. That just must be a secret tribute to our rld! ;-)
10:42:19amiconnRed led went on, no music.... but, after ~30sec it played a tiny snippet of my music again... led still on, then it said 'end of playlist', red led still on...
10:43:19amiconn...after another 30 sec, the browser appeared, and the red led went finally off
10:43:32amiconnOnly problem is that the browser showed garbage
10:45:56HClwho wants longhorn anyways, it has tcpa
10:45:57 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
10:45:57 Join bobTHC [0] (
10:46:08bobTHChi folks !
10:46:13HCl*sits next to markun in compiler creation class*
10:47:24BagderHCl: people will upgrade when MS drops support of their older win versions, I'm sure
10:49:31ashridahwhat are you talking about? people still use NT4 and win95, and neither of those have had support for years
10:49:50Bagdersome do, yes
10:49:55Bagderbut many follow
10:50:01LinusNashridah: of course, but the vast majority will upgrade
10:50:39LinusNdo not underestimate the power of the dark side
10:50:46ashridahi'm not sure it's as true. it becomes more true over time, as older pcs are discarded
10:50:55ashridahbut not because ms stops support
10:51:09LinusNashridah: true
10:51:16BagderI'd say it is both
10:51:34LinusNthe corporations need the support
10:51:36Bagderand that all new PCs will have the new MS os installed
10:51:40LinusNhome users don't care
10:51:48ashridahi know people who STILL keep installing nt4 on stuff because it's what they know. and lots of existing sites don't rip out stuff that works unless they can possibly help it
10:52:38Bagderthere are people using C64s too
10:52:51Bagderthe vast majority don't however
10:53:18BagderI'm just saying tcpa or not makes very little difference
10:54:04ashridahtcpa might actually be useful, just not the way ms wants it to be.
10:54:12Lynx_just wait until dell and co stop delivering drivers for old windows versions. ist really hard getting a new dell laptop to run win2000 already.
10:54:44ashridahLynx_: yes, which is why i said discarding old pcs tends to be the driving force behind upgrades.
10:55:34ashridahon the desktop side of things, that usually happens before the service period ends, but certainly not in the server world.
10:56:29Lynx_hmm, i'm not sure the server world will have too much an impact on the success of longhorn?
10:57:54LinusNcall me a cynic, but longhorn isn't about success or not, the consumers don't have a choice
10:58:44LinusNtheir choice is to run another os
10:58:52ashridahnope. there'll be a crossover period where one can get XP or longhorn on a new laptop, but that'll shift pretty quick.
10:59:08LinusNif they want windows, longhorn it will be
10:59:17LinusNsooner or later
10:59:27ashridahbut it'll take the expiry of existing hardware (either at the end of a service contract with a pc supplier, or age in general) to force the mass move to longhorn
10:59:41LinusNamiconn: yes?
10:59:50t0masI guess... with some right marketing linux can make use of tcpa
11:00:20ashridahthe big problem is that tcpa keeps getting associated with DRM
11:00:21t0masjust some commercials telling people what tcpa can do... and how it works... and that linux doesn't have it...
11:01:02ashridaht0mas: there's nothing intrinsically wrong with tcpa as a general idea, the problem is with whom the ultimate control lies.
11:01:25t0masyes, and that should be explained to some people
11:01:36 Join amiconn_ [0] (
11:01:44t0masmaybe linspire or some other payed linux distro should start such a campaign...
11:01:53t0masas they have money
11:01:57ashridahbut linux will get support for tcpa, but is more likely to allow the person operating the system to set it up, as an extra defense against penetration
11:02:59t0masand it's your own choise if you want it or not...
11:03:10t0masand with MS it's obligated afaik?
11:07:03 Quit amiconn (Nick collision from services.)
11:07:04 Nick amiconn_ is now known as amiconn (
11:11:35 Quit bobTHC (Read error: 110 (Connection timed out))
11:11:35t0masnice one
11:11:37*t0mas prints
11:14:57LinusNit's easy to come up with doomsday arguments against tcpa
11:15:19LinusNnobody knows how true they will be
11:16:02LinusNthe problem is that tcpa certainly seems to pave the way
11:16:21HClashridah: woah, let me get one thing straight, tcpa *does not* protect your computer.
11:16:29HClit only limits the software you're allowed to run
11:16:47HClin order to make people pay for software more
11:16:51HClrather than copying it
11:17:26HClalso, linux will be possible, but it will effectively kill oss since you'd have to buy an entirely new certificate if you'd even make a tiny change to the kernel
11:17:38HClnot to mention the entire kernel has to be screened before it'll be accepted
11:17:47ashridahHCl: not if you're the one doing the signing.
11:18:05HCli highly doubt that any end user will be able to sign, lol.
11:18:13HClthat would completely destroy the entire purpose of tcpa ;p
11:19:26ashridahthat depends on which way you use it.
11:19:40HClthat depends on which way microsoft uses it, heh
11:19:43ashridahyou can use it to protect software you distribute to other people, thus, windows can use it to enforce drm.
11:20:07ashridahbut if you implemented in your own linux kernel, signed your own software, and then took that stuff off the net, it becomes a far more advanced tripwire
11:20:21HClas far as i know, it will not allow people to "sign" software.
11:20:22ashridahsince the pc literally refuses to run signed stuff, and you've signed your own gear.
11:20:51ashridahnow, if pc manufacturers are the ones holding the signing process, that clearly fucks things up.
11:21:01HClyes, thats the entire idea, they will :p
11:21:19t0masofcourse they will...
11:21:22Rickimagine a h120 with tcpa in it, ebil >:o
11:21:28HClusers will just have access to signing documents and stuff like that
11:21:33Rickno rockbox ;(
11:21:37HClbut nothing that can potentially breach security, like software
11:21:41t0masHCl: no, that will be done automaticly ;)
11:21:50t0masword signes all doc −−> openoffice can't read them
11:21:58t0masnoobs don't care...
11:21:58HClwell, yea, but the users can specify who's allowed to read them
11:22:19*t0mas is afk
11:24:04 Join preglow [0] (
11:24:12ashridahHCl: the documents i've read seem to suggest one can ask the tcpa chip to generate a machine specific private key.
11:24:26ashridah(and that it doesn't take control of the boot process)
11:25:46preglowdamn, sounds like the cpu fan is trying to gnaw its way out if its own casing
11:25:51 Join amiconn_ [0] (
11:28:09rob-As far as I understand, tcpa will be able to be disabled by the user, in which case it will act exactly like a pc without it. Of cource any software that relies on it for drm won't work.
11:28:44preglowahh, that's a fine initiative
11:29:04ashridahrob-: which won't matter one iota to someone running linux in the first place
11:30:33ashridahthe only people who have anything to fear from DRM are people who use proprietary software. same goes for TCPA.
11:30:52ashridahand to a lesser extent, since TCPA can simply be disabled.
11:30:53rob-and it can actually help the security of linux by allowing you to verify the authenticity of all programs running.
11:31:06*ashridah nods.
11:31:10ashridahcan't modify it without resigning it
11:31:24ashridahthus it becomes all good until someone invents a constant-time factorisation algorithm :)
11:31:33rob-great for servers:-)
11:31:53*ashridah could seriously see tcpa as a souped up tripwire in the right setting
11:32:10ashridahirritating for a developer's platform tho
11:32:11LinusNtcpa will probably be used to sign documents etc
11:32:38LinusNmeaning you won't be able to read some documents without the tcpa hardware
11:32:59 Quit amiconn (Read error: 60 (Operation timed out))
11:32:59 Nick amiconn_ is now known as amiconn (
11:33:13LinusNeven worse, not being able to read Office docs at all with any other applications
11:33:17rob-just like encrypted pdfs today.
11:33:25ashridahLinusN: yeah, but if the signature is done in tcpa, wouldn't that imply that any software can unencrypt the data, provided it has the keys? and wouldn't the key need to be distributed to the people whom are supposed to be reading it?
11:33:43ashridahor do you encrypt it with THEIR key to start with, again, what's the problem, if it's tcpa's job to decode it?
11:34:18ashridahit looks more like 'i'm me, i'm the recipient, give me the raw data' wihch you then feed to openoffice.
11:35:07HCli think it'll mostly annihilate copying software (illegally)
11:35:11ashridahat least, from the looks of the tcpa pdf you mentioned before (which looks more plausible than most other doomsday scenarios, (like windows media player actually NEEDS tcpa to drm at all)
11:35:24*LinusN goes to lunch
11:35:38ashridahHCl: that's not necessarily a bad thing. it could be really really frustrating, particularly if a vendor dies
11:36:22ashridahif a company dies however, legally, they're obligated to give you everything you need to maintain the support implied in anything you signed. (of course, for shrink wrap software, that support isn't much)
11:36:52amiconnLinusN: I mean, any comments on the recording loop, planned recording tests and/or the rld report?
11:37:06rob-That will be an advantage for opensource, until the vendors realize that drm hurts their business
11:41:02ashridahthe part that annoys me the most is that any stupid law like CBDTPA that'd really fuck over non-drm'ed machines would automatically get accepted by our country, just because america likes shooting itself in the foot.
11:42:46HClaren't they already? with their dmca..
11:43:25ashridahnot to the same extent
11:43:34HClheh, yea, tcpa is worse.
11:43:43ashridahthe only companies that a law like CBDTPA would help would be hardware manufacturers and microsoft.
11:43:50rob-that's what happens when the law makers are controlled by corperations
11:43:54ashridah(in the short to medium term)
11:46:21preglowwouldn't be the first time a law happends to cater to big companies, no
11:47:17ashridahyeah. those guns those nuts claim are to protect them from the government or the kind of england (long live the king) clearly need some dusting off :)
11:48:51ashridahhm. perhaps in order to protest it, we should suggest to the protestors that they dress the law up like the king of england (long live the king)
11:49:01ashridahnevermind that the king is actually a queen atm, but yeah: )
12:00:07 Join NibbIer [0] (
12:08:14 Join MoosCamaro [0] (
12:08:24MoosCamaroHello all
12:22:52 Join |it|ChasKi` [0] (
12:24:22 Join cYmen [0] (
12:31:31***Saving seen data "./dancer.seen"
12:43:16 Join asdsd____ [0] (
12:43:19 Part asdsd____
12:50:55LinusNamiconn: are there any drawbacks with the alternate C version?
12:52:11amiconnI did not observe any... It only is a little larger than the asm version
12:52:17 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
12:52:21amiconn(because of the timeout handling I think)
12:53:03amiconnI have the alternate routine enabled in all my personal builds for a while, both for my rec v1 and my sis' Ondio FM
12:53:28amiconnThen I don't record often
12:53:41LinusNlet's commit it
12:54:38preglowdo you have any idea why the bootloader destroys vorbis playback?
12:55:02preglowsounds incredibly unlikely that modifying your firmware would have such a subtle effect
12:55:38LinusNbut i guess they assume that the onboard peripherals are uninitialized, and the boot loader changes that
12:55:59preglowprobably, but vorbis playback, of all things?
12:56:07LinusNweird indeed
12:57:12LinusNhmmm, i have an idea for a test
12:58:28LinusNanyone here up for a test with the 1.65 firmware?
12:59:22preglowdon't know how keen i am to possibly destroy my vorbis playback, since most of my files are vorbis...
12:59:40dwihnoI see Jens' did some kind of ATA thing... Supposedly it helps RLD. Any confirmations on this one?
13:00:06preglowwell, he thinks he's got it fixed
13:00:33dwihno*checks out*
13:01:00LinusNdwihno: it is definitely good rld medicine
13:02:13dwihnoRLD medicine for all \o/
13:02:29dwihnoI thought the Hitachi disks were to blame
13:03:02LinusNthey are
13:03:19LinusNif it weren't for them, we wouldn't need the medicine
13:03:21 Join einhirn [0] (
13:03:26preglowdidn't hitachi buy the old ibm disk fabs?
13:04:16LinusNpreglow: any ideas about the imdct_l issue?
13:04:18dwihnoMy disk has started to "click" occasionally (it reads/writes lot of data, clicks and then resumes without any error)
13:04:35dwihnoSupposedly I should check the backup's ok
13:05:20preglowLinusN: none, and no time, i have to code other things these days
13:05:49LinusNhow often is imdct_l called?
13:05:52preglowit's not cache issue, and it's not misalinged, that's more or less what i know
13:05:54preglowvery often
13:05:57preglowonce for every subband
13:06:03preglowwhich there are 32 of, i think
13:06:19preglowtimes two for stereo
13:06:39LinusNok, and you tried to exit before the loop, and it was still slower than the original?
13:06:40preglowand a frame isn't very long, so it's called pretty often
13:06:53LinusNfishy indeed
13:07:04*preglow concurs
13:08:40preglowit's very memory intensive in places
13:08:47preglowthat might be it, but i doubt it, since all data is in iram
13:09:40preglowother places again don't touch memory at all
13:10:35preglowwhen i put the loop in iram, performance went up to old imdct_l standards
13:10:51preglowbut i was thinking more in the lines of better performance than old imdct_l ;)
13:16:56amiconnLinusN: Do you have a clue what might cause the garbled data in case where there would have been an rld before?
13:17:11amiconnShouldn't ata_read_sectors() return an error?
13:17:48HClgarbled data?
13:20:21LinusNamiconn: it probably returns an error, but i guess it isn't handled correctly in all places
13:20:42LinusNi suspect the sector cache in read()
13:25:42LinusNpreglow: isn't the old imdct_l implementation inlined?
13:28:23amiconnLinusN: Iiuc, there should be room for improvement of the retry.
13:28:47LinusNamiconn: probably
13:29:06amiconnAfaics, the retry currently retries the whole transfer, ignoring any sectors that might already be transferred correctly
13:30:12preglowLinusN: yeah
13:30:27*LinusN has a theory
13:30:40preglowi'd sincerely like to hear it if it involves inlining :P
13:31:05LinusNthe cache is one-way only
13:31:57LinusNthe imdct_l function may be located such that it occupies the same cache lines as the III_decode loop
13:32:18LinusNso it has to refill the cache for every call to imdct_l
13:32:26LinusNand when it returns
13:32:52LinusNthat wouldn't be a problem with a two-way cache
13:33:04LinusNbut hey, this is coldfire...
13:33:56preglowoccupies the same cache lines? what? i thought physical addresses were pretty irrelevant for caches
13:34:21LinusNon the contrary
13:34:29amiconnThe coldfire cache is direct-mapped, not associative
13:34:34LinusNit's a set-associative cache
13:34:54amiconnLinusN: It's set-associative? Maybe I overlooked sth
13:35:03preglowno wonder i get huge friggin boosts from stuffing the loop in sram, then
13:35:18preglowwait, that i still don't understand :V
13:35:18rasherDo I sense a breakthrough?
13:36:20preglownot really
13:36:35preglowthere's not much i think i can do to remedy this anyway
13:36:44LinusNoh yes
13:37:23amiconnMCF5249 manual: 5.1 INSTRUCTION CACHE FEATURES: 8KByte Direct-Mapped Cache
13:37:34LinusNdirect mapped
13:37:55preglowapart from stuffing everything in iram or somehow remapping function locations
13:39:17LinusNamiconn: a direct mapped cache is the same thing as a 1-way set-associative cache
13:39:41LinusNwhich is pretty lame
13:40:16LinusNpreglow: yes, we can make sure that the imdct_l function is right next to the III_decode function in memory
13:40:17amiconnnitpicking? ;)
13:40:24LinusNamiconn: :-)
13:41:36LinusNnot only does the coldfire have an instruction-only cache, it has a *lame* instruction-only cache :-)
13:41:47preglowno surprises
13:41:59Bagderit goes fine with the rest of it ;-)
13:42:14LinusNconsistency is a good thing, isn't it? :-)
13:42:21preglowmy new screen is due in a couple of hours, but i want it now!
13:42:29LinusNpreglow: i know the feeling
13:42:37preglowthis flickering piece of shit is mocking me
13:43:09LinusNi remember having to wait three days for the wiggler to clear customs in sweden
13:43:25preglowgod, how i love customs
13:43:25LinusNit took one day to ship it to sweden
13:43:41rasherBut you're in euroland
13:43:41preglowthey usually spend about three days in just looking at stuff before they actually process it here
13:43:53rasherso no customs :)
13:44:10LinusNrasher: hehe, except when it comes from usa
13:45:02rasherTrying to avoid that, myself
13:45:35amiconnHmm, I hope I won't get those problems...
13:45:38preglowmy god, direct mapped caches suck
13:45:51LinusNpreglow: they sure do
13:45:53preglowamiconn: bought a h140? :P
13:46:06amiconnTrevor is sending me an 0308 Ondio to hunt the multivolume/hotswap problems
13:46:20rasherHaha, that Rocbox player sure looks.. dull
13:46:30preglowrasher: url?
13:47:32Lynx_gah, i need a firefox plugin that automatically clicks on 'skip' links on webpages with "intro"
13:47:58preglowyou aren't hardcore when you endorse mp3 players in your free time, no matter how mean you look
13:48:00rasherFlashblock is pretty good
13:48:10rasherit doesn't do that, but it's fairly nice
13:48:48rasherWhy oh why did they chose to advertise it with a girl who's roughly 50% larger than the DAP
13:48:48preglowLinusN: are fully associative caches common?
13:48:57rasheraren't you supposed to make it look SMALL
13:50:16preglow2 way set-associative looks like a nice trade-off
13:50:48amiconnrasher: I don't see anything on that page.. apart from the top banner and a black window...
13:51:16amiconn...ah, and a faint "roc" background to the left & right
13:51:36rasherWay to have flash not installed/disabled
13:52:40amiconnMeh, web "designers" :/
13:52:50Lynx_rasher: yes, i love flashblock
13:52:51LinusNpreglow: fully-associative caches are extremely uncommon
13:53:47LinusNthey use up huge amounts of die space, and can actually make the system slower
13:54:03LinusNsince the cache search can take time
13:54:13LinusNto find out if it is a hit or not
13:54:37LinusN2-way set would be nice
13:54:50preglowLinusN: but anyway, finding out it your suspicion is correct would just entail having a look at where stuff is mapped after loading mpa2wav, yes?
13:55:07preglowand all plugins are mapped to the same address?
13:55:17preglowthen this shouldn't be too hard
13:55:33*preglow gets MCF5249UM.pdf
13:57:18*HCl goes to work on homework...
14:01:33 Join kurzhaarrocker [0] (
14:06:12preglowso bits 12 through 4 needs to be the same over most of the function address space for this to be a problem
14:06:38preglowin the same range, at least
14:09:35amiconnpreglow: Should be easy to find out from the .map file
14:09:44kurzhaarrockerDo mp3 files have ape tags?
14:09:49amiconn(temporarily making the functions non-static as needed
14:09:53LinusNfrom the .map file, i see that they collide
14:10:07LinusNkurzhaarrocker: yes, if you have run mp3gain on them
14:11:52preglowLinusN: so, how to shuffle stuff around? just move the function definitions around? :P
14:12:05kurzhaarrockerBut why? At first glance ape tags don't seem to contain any volume related info. Do I miss something?
14:12:46LinusNafaik, it's a special ape tag
14:13:00preglowape tags har used for other files as well
14:13:03preglowmusepack especially
14:13:04LinusNcontaining all info needed to reconstruct the original mp3
14:13:13preglowhar = are, yes
14:13:14Bagderafaik, the store undo info in the ape tag they add
14:13:21Bagderthey store
14:13:32Bagdermp3gain that is
14:13:56kurzhaarrockerAh, thats what they call "lossless" :)
14:15:29amiconnkurzhaarrocker: It is lossless as the mp3 compressed data isn't recompressed
14:16:28kurzhaarrockerYou're right: I mustn't mix up "lossless" and "non destructive".
14:17:47amiconnsplitedit.rock is also lossless in this sense
14:18:58kurzhaarrockerHm. Some day I want spitedit to be destructive - maybe next millenium....
14:20:46preglowto split in the middle of frames?
14:22:28kurzhaarrockerNo. Now splitedit writes completely new files. That's not necessary. It could alter the fat, and prepend a sector to the tail file.
14:23:29preglowplugins that modify the file system = the shiznit
14:24:29kurzhaarrockerIt's a question of the api. We could add a function that prepends a disc sector to a file.
14:25:18LinusNkurzhaarrocker: a disk cluster
14:25:25amiconnkurzhaarrocker: Writing new files is a goodie imho
14:25:31 Join |it|ChasKi` [0] (
14:26:11kurzhaarrockerI think copying vast amounts of data on a mobile battery driven device is a waste. Relinking the files would be more appropriate.
14:26:54amiconnI certainly don't want to destroy my source file...
14:27:27rasherMake it an option!
14:27:30preglowLinusN: did you see my last question?
14:27:30amiconnLinusN: There is a slight problem with file renaming... maybe you have an idea how to handle it... or tell Zagor
14:27:32rasher(lots of options!)
14:27:51amiconnLinusN: If you want to change filename case _only_, it fails
14:28:25LinusNamiconn: oh, badness
14:28:26rasherTaht fails even on Windows and linux :)
14:28:50LinusNpreglow: i'm not sure how to handle it
14:29:24LinusNsince libmad is a library
14:29:26amiconnrasher: No, it does work on windows
14:29:54LinusNit works on linux as well, provided you mounted with the correct flags
14:30:05rasheramiconn: On all versions?
14:31:26 Join ripnetuk [0] (
14:31:35***Saving seen data "./dancer.seen"
14:31:57amiconnrasher: Surely not on 3.x ... ;)
14:32:21rasherHeh, could've sworn it didn't work on '98
14:33:28Bagderprovided that the name is longer than 8
14:35:12LinusNwindows has a setting for that
14:37:26LinusNpreglow: you could try to prove the theory by moving around things so they don't overlap in the cache, just as a test
14:37:44dwihnoWhat is the SEARCH stuff I see in the HTTP logs?
14:37:56dwihnoSome MS-"extension"?
14:38:14ripnetukMay I ask a quick curl question (since I happen to know one of you guys wroite it ;) )
14:38:33ripnetukI want to download an entire dir off a ftp server, and cant seem to find the recursive option
14:38:37Bagderdwihno: I believe it is a WebDAV request
14:38:47ashridahripnetuk: mget?
14:39:05dwihnoBagder: Hm. WebDav. Read about it. Haven't had the time to fiddle with it.
14:39:25ripnetuku mean wget? i assumed curl did recursive...
14:39:40Bagderripnetuk: curl has no recursive download!
14:39:51ripnetukthat would be why i cant find it :)
14:40:17ripnetukw shall use wget then
14:40:19LinusNuse wget, or write a script
14:40:20Bagder <= one of the new "no"s
14:40:34ripnetukhows iRiver coming on btw?
14:40:53preglowLinusN: i'll try it in half an hours time
14:41:07ripnetuki noticed you have mastered the remove LDC - nice work :)
14:41:21ripnetukLCD (doh)
14:41:27preglowhow'd you see the III_decode address, btw? you made it non-static?
14:44:17ripnetukthanks guys... see ya soon (i keep up by reading CVS logs) cya
14:44:20 Quit ripnetuk ("Leaving")
14:49:49LinusNhmm, the question is how we deal with cache issues in general
14:50:39LinusNperhaps add some section/linker magic to align code with the cache
14:56:38HClscrew this, where's my tablet :/
14:56:39preglowwould of course be nice if possible
15:02:48LinusNpreglow: as a test: place III_decode and imdct_l in a separate section, .cachetest
15:03:28preglowyes, think i need to do that, relocating stuff inside libmad files was not very easy
15:03:46pregloweverything's static, so they never declare their functions
15:04:26LinusNi have a patch, but it includes the entire imdct_mcf5249.S diff
15:04:36LinusNwant it?
15:10:11 Quit kurzhaarrocker (Remote closed the connection)
15:13:30*rasher watches curiously
15:13:37preglowyou were most certainly in the right, linux
15:15:45preglowso, you think grouping all cache critical functions close is the way to go?
15:16:04 Quit lostlogic ("Going to the moon")
15:16:53LinusNit's a good start
15:17:08preglowbut yeah, i think it's on a level with the old imdct_l now
15:17:12LinusNkeep all cache critical stuff in the same 8k block
15:17:16pregloweven if i don't go above that, this is a good deal cleaner
15:18:08preglowit's 230% realtime at 320kbps
15:18:13LinusNman, this has puzzled me a lot the last couple of days
15:18:32preglowyes, you're not alone in having been puzzled, heh
15:18:33LinusNfeels good to know what it was
15:18:49preglowtoday i learned a good deal about caches ;)
15:19:00LinusN230% without dithering, i suppose
15:19:20preglowwithout dithering, yes, but i refuse to believe dithering should be a big showstopper
15:19:30HClgod i love my tablet
15:23:54LinusNok, so now we have III_decode sorted out? no more tricks in your sleeve?
15:25:06preglowdct32 needs optimizing, but i wont do it
15:25:12preglowi'm so incredibly slow at doing that kind of thing
15:25:20preglowmost kind of things, actually, but especially that kind of thing
15:25:48preglowi'll see if there's more to be optimized when i have time
15:25:50preglowwhich is not right now
15:27:02LinusNpreglow: would a debugger help you?
15:27:13preglowhmm, i don't think so
15:28:17preglowhow so, have you got one? ;)
15:29:00LinusNwe can either get you a wiggler, or up the prio on the gdb stub
15:29:59preglowbut no, i don't really think a debugger would help me much
15:30:11preglowit's fairly streamlined code
15:30:24preglowand i really shouldn't touch it further until summertime, heh
15:31:25LinusNok, so how about committing it then?
15:31:34preglowcan do, how about the cachetest stuff?
15:31:39LinusNwithout the cache test
15:31:56preglowok, i'll just ensure things are up to snuff
15:32:19*LinusN has to go
15:33:09preglowyup, later
15:33:11 Part LinusN
15:33:39HCl230% realtime at 120mhz, i assume?
15:33:43HClwhat was it before?
15:33:53preglowaround the same
15:33:57HCloh :/
15:36:45preglowoh, bleh,i can push it above 230% with linus' fixes
15:37:47webmindoverclocking ?
15:38:37 Nick tvelocity[away] is now known as tvelocity (
15:41:30preglowit's still a little bit slower
15:43:20HClwebmind: 120mhz is already the max
15:43:26HCl(though underclocked)
15:52:34 Join Lost-ash [0] (
15:52:36 Quit ashridah (Read error: 54 (Connection reset by peer))
15:54:30 Quit odd ("leaving")
15:54:50preglowwell, well
15:55:16preglowthe old imdct_l including windowing is about as fast as the new imdct excluding windowing
15:55:22preglowwhich is kind of irritating
16:08:30 Join Sucka [0] (
16:09:17 Quit Lost-ash ("Leaving")
16:16:58 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
16:31:39***Saving seen data "./dancer.seen"
17:34:26 Quit asgsa (Ping timeout: 14400 seconds)
17:47:45preglow2001 <-
17:52:35preglowany good ways to spot dead pixelsz?
17:53:45*HCl patpats antlr and runs his homework through the tool to get his answer :X
17:56:31 Join |it|ChasKi` [0] (
17:57:23rasherpreglow: try all-white, red, blue, green screens?
17:58:10rasherand black
18:00:05preglowdust makes that impossible
18:00:21preglowi tried on black now, and i saw dead pixels everywhere!
18:00:30 Join bobTHC [0] (
18:00:35preglowfuck, shit
18:00:37preglowi've got a dead pixel
18:00:56rasheroh dear
18:01:03rasherhow dead?
18:01:09preglowi've got two of them
18:01:13preglowright next to each other
18:01:16bobTHCpreglow >
18:02:11bobTHCworks fine
18:02:46rasheror just xsetroot
18:03:13preglowbut first i want to enable dvi
18:18:34preglowtwo always on green pixels
18:18:50preglowat least they're not THAT noticable
18:19:32Lynx_preglow: you won't just send it back and order a new one?
18:19:38rasherI have an always on blue one
18:19:43rasherit's not too bad
18:20:18preglowLynx_: that'll take bloody ages
18:20:27t0masisn't there an option in X11 to do something about it?
18:20:36t0masit can be white/blue or black/blue
18:20:58Lynx_preglow: every time you will notice the dead pixels for the next few years they will irritate you ;)
18:22:24bobTHCin few years, he will notice some others dead pixels imho
18:23:14*HCl remembers sending his tft back 3 times till he finally got one without a dead pixel
18:23:30bobTHCbad luck
18:23:45HCli was mostly lucky they let me swap it
18:23:50HClcause technically, they weren't required to.
18:25:41bobTHCtft are too delicate to be shipped
18:26:02Lynx_bobTHC: dead pixels don't come from shipping
18:27:01 Quit |it|ChasKi` (Read error: 145 (Connection timed out))
18:27:11bobTHCnowdays, when they are out from factory tft are dead pixel free
18:27:42Lynx_bobTHC: are you sure? i highly doubt that.
18:28:04bobTHCfor samsung i'm barely sure
18:28:47Lynx_i have never heard of a pixel going dead after buying the tft
18:30:50preglowpeople can ship displays with dead pixels
18:31:02preglowand dell isn't required to take this display back
18:31:42***Saving seen data "./dancer.seen"
18:31:52Lynx_aren't they required to take anything back?
18:32:04preglowi think they've got a limit on five dead pixels
18:32:09preglowbefore they're letting you have a new one
18:32:32Lynx_it seems samsung is offering a zero pixel defect guarantee, but they just throw out all defect panels after fabrication.
18:33:15Lynx_preglow: i meant the law that lets you send any order back within a few weeks, not lcd specific
18:34:49preglowLynx_: ahh, i don't know if that applies after you've actually opened it
18:35:34Lynx_preglow: well, it wouldn't make much sense if you could not even look at it. i have sent stuff back to amazon that was opened.
18:36:59Lynx_you could also kill 5 more pixels ;)
18:38:10t0mashmm... anybody here knowledgable about SSH?
18:38:36preglowi could of course try massaging these dead pixels and send it back when i have a hundred more
18:38:49t0maspreglow: they'll suspect something then ;)
18:39:10Lynx_before you do that you punch a hole in the box and claim transport damage ;)
18:39:27Lynx_t0mas: i know the basics...
18:39:38bobTHCme too t0mas
18:39:38t0masthe protocol basics or the usage?
18:39:45Lynx_t0mas: usage
18:39:56t0masyes, I know that too
18:40:04Lynx_t0mas: i assumed that :)
18:40:15t0masI need some info on the AES encryption thing used...
18:41:06bobTHChe encryption used by ssh can be either AES-128, 3DES, Blowfish, cast-128, ...
18:42:10t0masbut I don't know what it uses AES for...
18:42:27t0masencryption of the whole session?
18:42:39t0mas(and AES is public-key right?)
18:42:40Lynx_t0mas: yes
18:42:44 Join webguest19 [0] (
18:42:49t0masok, not key exchange only?
18:43:08t0masas I thought it exchanged the keys for some normal non public key thing with AES
18:43:26bobTHCto remotely send and receive files under the strong encryption
18:43:29Lynx_t0mas: as far as i understand the whole session is aes
18:43:44t0masok, no switching...
18:43:49t0masor AES isn't public key?
18:46:19bobTHCAES is symetric
18:46:47t0masehrm... how does it send keys?
18:47:22Lynx_hmm, there is a transport layer and a user authentication layer...
18:47:58Lynx_ah, it uses rsa / dsa
18:50:10t0masjust wondering...
18:50:10t0masdebug1: SSH2_MSG_KEXINIT sent
18:50:10t0masdebug1: SSH2_MSG_KEXINIT received
18:50:10t0masdebug1: kex: server->client aes128-cbc hmac-md5 none
18:50:10t0masdebug1: kex: client->server aes128-cbc hmac-md5 none
18:50:28t0maskex = Key Exchange I guess... and it uses aes128-cbc for that..
18:50:36t0masso AES for keyexchange?
18:51:28*t0mas is away: eating
18:53:14bobTHCthere is no server key. The session key produced by Diffie-Hellman is used to encrypt the real session key. Once the session key has been established, the traffic is encrypted using one of the following symmetric algorithms: 128 bit AES,Blowfish,3DES,CAST128,Arcfour,192 bit AES, 256 bit AES
18:54:49 Quit bobTHC ("" smoke Weed Everyday" ;)")
18:55:27Lynx_and the rsa is for user authentication
18:59:51 Quit webguest19 ("CGI:IRC (EOF)")
19:08:50 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
19:14:43 Part MoosCamaro
19:28:14 Join TCK- [0] (
19:33:01preglowcleartype fonts looks nice
19:34:21 Quit thegeek (Read error: 131 (Connection reset by peer))
19:36:37 Quit TCK (Read error: 60 (Operation timed out))
19:40:41 Join Strath [0] (
20:01:44 Join thegeek [0] (
20:03:35 Quit merbanan ("Leaving")
20:06:04 Join arfo [0] (
20:08:01 Part arfo
20:24:55 Join LinusN [0] (
20:25:57LinusNpreglow: the cache trick didn't help?
20:30:59preglowLinusN: yes it did
20:31:05preglowLinusN: most definitely
20:31:42preglownow it's just a tiny bit slower than the old one
20:31:44***Saving seen data "./dancer.seen"
20:33:06LinusN15.55.16 # <preglow> the old imdct_l including windowing is about as fast as the new imdct excluding windowing
20:33:28preglowyes, exactly
20:34:00LinusNso we still have a mystery to solve :-)
20:34:17preglowwell, are you completely certain that i just isn't plain slower in some way?
20:34:24preglowthe other one promises to be very fast once it's in cache
20:34:27preglowsince all constants are code
20:34:38preglowand a immediate move takes one cycle if it's cached
20:34:40LinusNtrue, true
20:34:47preglowthen again, so should a move from iram...
20:34:52preglowand it's smaller in code to boot
20:34:53LinusNooops, gotta run
20:34:57LinusNcu later
20:35:05 Part LinusN
20:35:25*preglow does the dead pixel lament
20:36:42*preglow suddenly remember he hasn't watched any demos on the new screen yet!
20:53:39 Join soxx [0] (
20:53:40 Quit TCK- (Read error: 54 (Connection reset by peer))
20:54:46 Join TCK [0] (
20:55:44 Quit Sucka ("a bird in the bush is worth two in your house")
20:59:59 Join Sucka [0] (
21:08:59 Join webguest53 [0] (
21:12:46 Join lodesi [0] (
21:13:27preglowthe screen sometimes "overflows" in some resolutions
21:13:50*preglow summons 2001fp owners
21:31:42 Quit soxx ("Snak 4.13 IRC For Mac -")
21:49:05 Quit webguest53 ("CGI:IRC (EOF)")
22:01:36 Join TCK- [0] (
22:10:08 Quit TCK (Read error: 60 (Operation timed out))
22:11:08 Nick tvelocity is now known as tvelocity[away] (
22:17:15*HCl yawns
22:31:47***Saving seen data "./dancer.seen"
22:38:49*HCl prods people awake
22:39:23preglowi'm awake!!!!!1
22:40:08HClentertain me ;/
22:40:12HClsup preglow? :/
22:41:38 Join StrathAFK [0] (
22:43:19 Quit Strath (
22:53:39 Join TCK [0] (
22:53:39 Quit TCK- (Read error: 104 (Connection reset by peer))
22:59:09 Quit Aison ("( :: NoNameScript 3.72 :: )")
23:30:37 Quit Harpy (Read error: 145 (Connection timed out))
23:34:09 Join xam [0] (
23:41:05preglownothing up with me
23:41:10preglowstroking my new monitor
23:43:08 Join XShocK [0] (
23:43:10 Part XShocK
23:45:56 Join matsl [0] (

Previous day | Next day