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 2004-03-18

00:04:55 Join AciD` [0] (~acid@
00:05:21 Quit Bagder ("")
00:13:34 Join midknight2k3 [0] (
00:13:41quelsarukhi midknight2k3
00:13:54midknight2k3heu quel!
00:13:56midknight2k3hey zagor as well
00:13:58midknight2k3and tops
00:18:28midknight2k3Zagor: i mentioned a bit ago about updating the AB Repeat patch and you considering it for inclusion?
00:19:01Zagori consider all good code
00:19:12midknight2k3it's very out of date
00:19:16midknight2k3but i offered to update it
00:23:24quelsarukZagor, now that i'm downloading the t-shirt contest... what happened with that?
00:23:33*Zagor didn't hear that
00:23:52quelsarukyou're getting old
00:25:39 Join oxygen77 [0] (
00:26:53quelsarukhm.. i just had a strange issue in the WPS
00:27:02quelsarukthe display was blank
00:27:08quelsarukbut... i can't reproduce the bug
00:28:51Zagorand we're supposed to believe THAT? ;)
00:30:11quelsarukthe song ended and with the new song, the screen was updated
00:30:22quelsarukreturned to the previous song and worked
00:30:31quelsarukused the hammer and got nothing..
00:30:36Zagoryeah. i've actually seen that one myself.
00:30:44Zagornever tried the hammer though...
00:30:58 Join wake [0] (~wake@
00:31:03quelsarukas i'm with the computer i don't really check the box, just listen music
00:31:14quelsarukyou should, is a great experience ;)
00:33:06quelsarukthe website checkout finished!!
00:33:13quelsaruki will update then tomorrow
00:33:39quelsarukgood night!
00:34:10 Quit quelsaruk ("may the wind be with you")
00:36:13***Saving seen data "./dancer.seen"
00:44:27 Quit AciD` (Read error: 104 (Connection reset by peer))
00:47:41midknight2k3upcoming patch attempts: (1) "get file properties" (2) tetris update (3) clock update (4) a-b-repeat update
00:47:44midknight2k3yea woo
00:57:51midknight2k3DIE DIE DIE
01:10:47 Quit Zagor ("Client exiting")
01:26:37 Part oxygen77 ("Leaving")
01:26:50 Quit mattzz ("Client exiting")
01:56:31midknight2k3ttyl all
01:56:33 Quit midknight2k3 ()
02:07:21 Join track [0] (
02:07:25 Quit track (Client Quit)
02:13:35 Quit joshn__ (Read error: 110 (Connection timed out))
02:13:40 Quit joshn (Read error: 110 (Connection timed out))
02:14:00 Quit joshn_ (Read error: 110 (Connection timed out))
02:14:28 Quit top_bloke (Read error: 113 (No route to host))
02:28:32 Part amiconn
02:36:14***Saving seen data "./dancer.seen"
02:45:18 Quit Nibbler (Read error: 54 (Connection reset by peer))
02:46:48 Quit scott666 (Read error: 60 (Operation timed out))
02:49:26 Join scott666 [0] (
03:02:22 Quit adi|home (Client Quit)
03:19:03 Join adi|home [0] (~adi|
03:48:17 Quit hardeep ("[BX] Beefcake! BEEFCAKE!")
04:00:56 Quit wake (Remote closed the connection)
04:29:30 Join Nibbler [0] (
04:36:15***Saving seen data "./dancer.seen"
05:08:35 Join Guest [0] (
05:12:31 Join Vandal-AFK [0] (
05:12:35 Quit Guest (Client Quit)
05:12:36 Nick Vandal-AFK is now known as Vandal (
05:15:10 Join jfab199 [0] (
05:15:40 Quit jfab199 (Client Quit)
05:21:31 Quit scott666 ("i'll be back...eventually...")
05:24:48 Join joshn [0] (~joshn@
05:53:35 Quit Nibbler (Read error: 104 (Connection reset by peer))
06:12:11 Quit Vandal (Read error: 110 (Connection timed out))
06:16:25 Nick dwihno_ is now known as dwihno (~dw@
06:36:17***Saving seen data "./dancer.seen"
07:33:06 Join Nibbler [0] (
08:36:20***No seen item changed, no save performed.
08:37:41 Join matsl [0] (
09:00:06 Join LinusN [200] (
09:19:12 Join AciD` [0] (
09:33:22 Join c0utta [0] (
09:36:16 Nick c0utta is now known as c0utta{busy} (
10:10:40 Quit joshn (
10:10:40 Quit elinenbe (
10:11:40NJoinjoshn [0] (~joshn@
10:11:40NJoinelinenbe [0] (
10:31:57 Quit c0utta{busy} (Read error: 110 (Connection timed out))
10:36:24***Saving seen data "./dancer.seen"
10:46:14 Join c0utta [0] (
10:54:22LinusNc0utta: i, daniel and björn discussed your work over lunch
11:07:15c0uttaa curry ?
11:12:03LinusNwe want a configurable f2 button only (at least to begin with)
11:12:23LinusNthe default f2 config will be in a file, not in the ajz
11:13:31LinusNconfiguration should be done by editing the file, no on-the-fly menu creation
11:16:16dwihnoAnd if the file is not load:able, there should be some kind of nice error message :D
11:16:25LinusNof course
11:16:26dwihnoOr should there be a default built-in config?
11:16:32dwihnoI love error messages
11:16:35dwihnoThose are good for you
11:16:45dwihnoAs long as the menus won't reload everytime you press F2 :)
11:16:52LinusNthey won't
11:19:47 Quit c0utta (Read error: 60 (Operation timed out))
11:35:57 Join c0utta [0] (
11:36:26c0uttalinus - i checked the irc log
11:40:39 Nick c0utta is now known as c0utta{afk} (
12:04:34 Nick c0utta{afk} is now known as c0utta (
12:05:25c0uttadid your discussion with daniel and björn include the contents of the F2 menu, linus ?
12:09:30LinusNnot really
12:10:04LinusNwe don't have any strong opinions about that
12:10:15LinusNif we're talking about the default menu that is
12:10:35c0uttadefinition of the default menu is the next step - should we have a vote on the mailing list ?
12:11:10LinusNnah, let's create one with the options we like and then people will enter the discussion
12:11:28LinusNthen we'll add actions as we go
12:11:29c0uttabtw - menus now load once only. statics are a wonderful thing
12:11:45LinusNgreat, care to show me the code?
12:13:04c0uttayou've seen most of the code - it's in action.c
12:13:10c0uttado you want to see it again ?
12:13:29LinusNyes, please
12:14:21c0uttamail or dcc ?
12:15:47LinusNmail, please, firewall from hell
12:28:02*LinusN ducks
12:32:26LinusNgot it
12:32:50 Join track [0] (
12:33:00trackmorning even
12:34:50LinusNc0utta: you need to have one menu_entry[] array per context, don't you?
12:35:57c0uttayes, but f2 isn't context sensitive. i've taken the easy option at this stage
12:36:12LinusNmaybe it should be context sensitive?
12:36:25***Saving seen data "./dancer.seen"
12:36:35 Quit silencer (Read error: 60 (Operation timed out))
12:37:05c0uttait actually already is. if you choose to put other contexts into the config file then it could display different options
12:37:43LinusNbut only one menu is stored
12:38:07c0uttathat's right. since f2 wasn't going to be context sensitive i didn't bother at this stage
12:38:34c0uttai am happy to do it though
12:38:36LinusNi think it should be
12:38:47c0uttaok. i just did a "quick and dirty"
12:39:00LinusNi understand
12:39:23c0uttai would need to store and array of menus
12:40:12 Quit AciD` (Read error: 104 (Connection reset by peer))
12:40:18c0uttaactually, an array of structs with context, key and menu
12:41:14c0uttawhy have you decided that f2 should be context sensitive ? isn't that the role of f3 ?
12:41:44LinusNyes, but we imagined that people would want different options on the f2 menu in different screens
12:41:59 Join silencer [0] (
12:42:01c0uttabut isn't that just f3 ?
12:42:01LinusNelse the f2 menu would soon be pretty long
12:42:10LinusNnot necessarily
12:42:28c0uttaif both are context sensitive then you really only need one!
12:42:56c0uttabecause they'll have the same capability
12:43:07LinusNbut f3 isn't configurablre
12:43:37c0uttathat's right - you guys have stopped me from being creative
12:44:12c0uttaare you still saying that f3 should _not_ come from the file ?
12:44:28LinusNat this point, yes
12:44:52c0utta"at this point" - very political speak
12:45:19LinusNif it turns out that f2 isn't enough, we may conside rf3 as well
12:45:47c0uttai hate the idea of hard coding F3 :(
12:46:34LinusNi hate the idea of losing file delete options because of a badly configured f3
12:46:46c0uttayeah yeah yeah
12:47:21c0uttahey, _you_ wouldn't lose them. only some turkey who didn't understand basic file systems
12:48:23LinusNremember, there will be options in the f3 menu that won't exist in another menu
12:48:57LinusNimagine the guys who upgrade to the latest daily without updating the f3 keyfile
12:49:15c0uttayes, i understand. how about a compromise ??
12:49:22LinusNi'm all ears
12:49:43c0uttahave a standard hardcoded f3, but if there's any entries in the config file then add those to the list ?
12:50:24LinusNsounds ok to me
12:51:36c0uttain summary, i need to get the menus saved for each context and keypress
12:52:58c0uttacan you, daniel and björn discuss about standard entries for each context in F2 ?
12:53:17LinusNyou mean available actions?
12:53:37LinusNor default config?
12:54:10c0uttai think all actions will eventually be required, but just the default config at this stage
12:54:28LinusNi'm sure you can come up with a nice set of defaults
12:54:52c0uttayes. i currently have sound menu, volume (my personal favourite) and plugin
12:55:05 Join quelsaruk [0] (~v65wf@
12:55:09c0uttai have seen requests for backlight
12:55:13LinusNquelsaruk: hi
12:55:19c0uttahey quelsaruk - just the man i need to speak to
12:55:19LinusNc0utta: backlight options?
12:55:24quelsarukhi LinusN
12:55:26*quelsaruk hides
12:55:32LinusNquelsaruk: i never got anything from you
12:55:36c0uttalinus, yes backlight options
12:55:37quelsarukare you gonna kill me c0utta?
12:55:46quelsaruknor even this second time??
12:56:09c0uttano, quelsaruk. i saw that you were about to change menu initialisation
12:56:23LinusNquelsaruk: no email at all this time
12:56:49quelsarukc0utta, it's just an idea, not sure if i will be implemented
12:57:05quelsarukLinusN, ok, i'm going to try a 3rd time..
12:57:16LinusNit didn't bounce?
12:57:22quelsaruki don't know
12:57:27quelsaruklet me check the mail
12:57:42quelsarukyou know... spanish university mail servers are odd
12:58:10c0uttaquelsaruk, that's ok. just keep in touch when you're thinking about changing all the current menus
12:59:27quelsarukwell... i want to change the visualization, not the contents of any menu, are you thinking in changing the contents?
13:00:34c0uttai have add some new items to the menu_items struct (jörg has already added some for voice)
13:00:59quelsarukme too
13:01:01c0uttamenus can no longer be initialised upon declaration
13:01:33quelsaruki had to add 2 items right now
13:01:54c0uttathey will rely on an "action" code that will retrieve the lang (including voice), func pointer and another "setting" value
13:02:40quelsarukthat happens for not ending this last year! ;)
13:03:38LinusNc0utta: meybe i've misunderstood, but the constant menus won't have to use the action code, would they?
13:03:55c0uttalinus, yes they will
13:04:00c0uttawe can add menus to f2
13:04:29quelsarukthen c0utta, seems that my code must change a bit to fit the new changes :)
13:04:34LinusNyes, but that doesn't change anything, does it?
13:05:19quelsarukLinusN, message sent, please check if you get the patch file
13:05:25c0uttai see your point linus. are u concerned that someone might duplicate the f1 menu on f2 ?
13:05:44LinusNi mean, the recording settings menu would still work the same, even if called from the f2 menu?
13:06:17LinusNmy concern is that i don't want to change the menu creation on the existing menus
13:07:19c0uttai personally don't see a problem with it
13:07:28LinusNmore code
13:07:35LinusNlarger footprint
13:08:13LinusNand possibly having to define unnecessary actions
13:08:38c0uttahmm. i will have to review what i've done. i had always planned that existing menu creation would need to be changed
13:09:04LinusNfor instance, i don't want the debug menu to be action-based
13:10:10quelsarukc0utta, just an idea... if i understood well, you are working on user-configured menus, no?
13:10:17c0uttai remember why i wanted to change the menu initialisation now. i had planned that menu creation would be fully dynamic
13:10:19LinusNwhy force the programmer to call menu_item() when he can fill in the info himself?
13:10:21c0uttaquelsaruk, yes
13:10:42LinusN"fully dynamic"?
13:10:42c0uttalinus, to enforce an action based platform
13:11:09c0utta"fully dynamic". using the f1 menu, press f2 to add a menu entry to the f2 menu
13:11:36LinusNah, that one
13:11:40c0utta(the reason behind #configurable=yes)
13:11:58LinusNi think we can leave that for now
13:12:07LinusNi want the basic stuff first
13:12:18LinusNquelsaruk: got it
13:12:18c0uttatrue. i did have it working in my proof of concept. it was beautiful!
13:12:20quelsarukthen, c0utta, how are you going to add *settings* to those menus? a combo? or mark checking those settings and using a "add settings to menu" function?
13:12:38quelsarukLinusN, great.. i was going to use the hammer again :P
13:12:49LinusNgoat blood works fine too
13:12:58c0uttaquelsaruk, i leverage all the existing code. the work i've done is embarassingly small
13:13:38quelsarukc0utta, my graphic menus still have a lot of work to do also ;)
13:14:01LinusNquelsaruk: the entire icons.c file differs
13:14:11LinusNCR/LF maybe?
13:14:11quelsaruki was just thinking that you could use some of my code for that (a more cool interface to add settings to the menu)
13:14:27quelsarukLinusN, yes, but i thought i had repaired that
13:14:40quelsaruki suppose i repaired icons.h and forgot icons.c
13:14:59quelsaruki just added 2 new icons, icon_active and icon_inactive
13:14:59c0uttaquelsaruk, i understand your question
13:15:40c0uttai'm getting the menu entries from a file which has the *setting* as a final parameter
13:16:31c0uttafor example - changing the volume
13:16:56quelsarukLinusN, another thing, i re-wrote the rvf howto yesterday night, would you mind checking if it's easy enought for people or not before commiting it?
13:17:26c0uttathe entry: "0;512;1" in my input file means all contexts (0), f2 (512), volume(1).
13:17:37c0uttathe 1 is a #define
13:18:08LinusNquelsaruk: just commit it, we can always change it again
13:18:10c0uttathis brings up the current "volume setting" option
13:18:41c0uttabut if i use "0;512;1;Volume to 88%;88" it will display "Volume to 88%" in the menu and instantly set the volume to 88
13:19:50quelsarukok :)
13:20:21c0uttait leverages the current volume function, so the code size hasn't really increased that much
13:20:41LinusNc0utta: maybe we could expand a little on this
13:21:07c0uttai'm all ears
13:21:08LinusNtoday, for each setting, you would have to add a function that acts on the argument, right?
13:21:39quelsarukc0utta, but it still uses the menu_draw() function and so on when displaying a custom menu?
13:21:48LinusNquelsaruk: yes
13:21:53c0uttayes. remember my e-mail i sent to you. the way that the "volume" function changed
13:21:56c0uttaquelsaruk, yes
13:22:13c0uttait now has a void pointer
13:22:45c0uttaas the parameter
13:23:30LinusNhmm, never mind, my idea wasn't that good :-)
13:23:32quelsarukyou just send parameters to the function :)
13:23:35 Quit track (Read error: 54 (Connection reset by peer))
13:23:43quelsarukLinusN, i can't believe that :P
13:24:01c0uttaif there are parameters in the file then pass them, otherwise NULL
13:24:31c0uttaquelsaruk, i agree!
13:24:52quelsarukok, i wanted to have a better idea of your code :)
13:25:45quelsaruki think that i will have to send parameters to menu_run() (if i remember well)
13:26:22c0uttathe only change required was to menu_init, but this was committed last week just after 2.2 release
13:26:49quelsarukor maybe merge my splash_menu() with the menu_draw() and send another parameter to the menu_draw()
13:27:07quelsarukso i have to think more about how to implement the splash menus :D
13:27:19LinusNc0utta: which code parses the argument string?
13:27:44c0uttaquelsaruk, sounds good to me. i haven't changed menu_draw, menu_run at all. i have added new functions though
13:28:25c0uttalinus, the code in (for example) volume. i take the parameter and check it against null
13:28:27quelsaruki had to add new items to the menu_items() struct, but i don't mind if i have to re-add them.
13:29:09c0uttaluckily rockbox is so modular that other functions take care of the parsing.
13:29:42LinusNc0utta: but how about "channels", for example?
13:29:53LinusNwhere it is a list of options
13:30:56c0uttai checked the code for functions like set_cfg_option - they do all the parsing for me
13:31:26LinusNnot really
13:31:27quelsarukc0utta.. errmmm.. maybe i had to change that, sorry :(
13:31:50LinusNset_cfg_option still needs a list of options to compare with
13:32:14LinusNso you would basically duplicate the code in settings_load_config()
13:32:35c0utta100% correct
13:33:17LinusNso if we change settings_load_config() a little, we could leverage from that
13:33:58LinusNthat code needs to be revised anyway
13:34:00c0uttalinus, i started this a long time ago before i really understood the code
13:34:22c0uttanow that i have a fairly good understanding, i'm keen to review this too
13:34:40LinusNhmm, a lot of work
13:35:02LinusNi want a first shot at the f2/f3 menus forst
13:35:43c0uttawe'll just need to make sure that the default f2 only performs "easy" actions
13:35:44LinusNif we start revamping the settings first, the f2/f3 stuff would be delayed even further
13:36:14c0uttameans that backlight is off the list :(
13:36:15LinusNi have some nice ideas for the settinsg code, btw
13:36:27c0uttai'm all ears
13:36:32LinusNbut let's save that for later :-)
13:37:59LinusNbasically, each setting would be a struct, containing all necessary information, such as the name, the list of options, scaling factors etc etc
13:40:32c0uttai had a similar idea. i also had an idea about changing the way the settings operate
13:40:50c0uttathe current up/down to change i find annoying
13:41:30quelsarukLinusN, that could even make my code easier :)
13:41:32c0uttai like to see all my options at once (if possible) and then scroll down/up to select
13:41:40quelsaruki had that idea :P
13:41:49quelsarukthat's part of my gfx menus
13:41:57quelsaruk(in mi mind, of course)
13:42:22c0uttabut, i also wanted to allow the same setting code to be used in the quickscreens
13:42:22quelsarukthat's why i said i had to change the set_cfg_option
13:44:01quelsarukLinusN, rvf howto commited, hope it's ok :)
13:45:30c0uttaanyway, i seem to be doing a lot of talking and not much programming
13:46:27quelsarukc0utta, i always need a lot of talking to order my ideas
13:47:04c0uttai get most of my ideas from linus anyway
13:50:58quelsarukLinusN, player can only show characters in the rocklatin1 no?
13:51:25quelsaruktranslating to english
13:51:35LinusNyou want to define your own chars?
13:51:45quelsarukno, i want to use some chars
13:52:20quelsaruki think i won't have to define new chars, if i'm lucky :)
13:53:19quelsaruki wanted to port part of the gfx menu code for the player, so they can also use chekboxes (getting rid of the yes/no questions)
13:53:53quelsaruki think it's a good idea
13:55:07quelsarukdid you see my fake gfx menu snapshot? (made a long time ago)
13:55:39LinusNdon't remember
13:55:42quelsaruki have a lot of ideas in my mind..
13:57:25quelsarukif i can materialize this one, i'll tell you :)
13:57:33 Join mattzz [0] (
13:59:03quelsarukhi mattzz
14:00:56mattzzLinusN: I have a little suggestion for the rockbox web pages
14:02:07mattzzLinusN: How about writing what's currently in the pipe? E.g. voicefont, menu keys etc.
14:02:26mattzzKeeps people on track. mainly pure users
14:02:52quelsaruksorry to enter the conversation, but that reminded me something. What happened to the languages in the website?? I tried to find in the site where to download a language and wasn't able :(
14:03:45LinusNmattzz: sure, as long as someone updates it...
14:05:04quelsaruki don't really need any language or font file, but imagine that the only change made to the source is a language update, why should i download a full rockbox package if i only want a 30KB file?
14:05:10mattzzCould be done on a monthly basis. Kind of a summary of current activities in CVS and mailinglist
14:10:15mattzzLinusN: How many people are currently able to update web pages?
14:11:09LinusNquelsaruk: because you don't know that the only change is the lang file?
14:12:03LinusNmattzz: any guy with CVS access can change it
14:12:17quelsarukLinusN, recent cvs activity ;)
14:12:29LinusNmattzz: we have a digest
14:12:46LinusNand rss feed
14:13:31quelsarukbtw, mattzz, LinusN, would you mind reading the rvf howto? i have already found 1 spelling error, and as i'm going to update it, correcting new errors now is less work :)
14:14:54mattzzquelsaruk: Yep, I wanted to check out a linux-flow for rvf creation
14:16:31mattzzLinusN: The thing is: When I visit the rockbox site, I see some very rough news and _rececent_ CVS activities. If I am not too much into rockbox hacking I don't get the idea what is currently cookin'.
14:16:37quelsarukLinusN, another possibility, you download bleeding edge, and find out that a new option has been added, and as you use ancient egipcian language, you would like to see that not in english, but in ancient egipcian. You have to wait until next day to have the daily build tarball :P
14:17:41LinusNmattzz: we need to update the digest/rss feed more regularly, and add a link from the front page
14:18:11LinusNquelsaruk: i didn't know that the languages were updated that often
14:18:18LinusNon the download page
14:18:43quelsarukLinusN, an hypothetical possibility ;)
14:18:56mattzzLinusN: Hm, where can I find the rss feed on the main page?
14:19:14LinusNyou can't (the link to it disappeared)
14:19:45quelsarukLinusN, normally i *fight* against zagor, but as he's not here :D
14:20:40quelsarukmattzz, would you mind sending me that rvf howto for linux??
14:20:48mattzzLinusN: OK, that _is_ exactly what I've been missing ;-) You guys have been there allready...
14:21:15quelsarukso i can add it to the website
14:21:17mattzzquelsaruk: I want to fiddle with that today. If I find out something I will let you know
14:21:58quelsarukif i'm not here, please send me a mail :)
14:22:16quelsaruklunch time!!!
14:22:26quelsaruki thought it was earlier!
14:22:32quelsarukcu l8r!!!
14:23:14 Quit quelsaruk ("someday i'll lost my head!")
14:23:26 Quit c0utta (Read error: 60 (Operation timed out))
14:36:28***Saving seen data "./dancer.seen"
14:39:42 Join AciD` [0] (
14:40:59 Quit joshn (Read error: 60 (Operation timed out))
14:49:04 Join silencer_ [0] (
14:52:07 Quit AciD` (Read error: 54 (Connection reset by peer))
14:57:36 Quit silencer (Read error: 60 (Operation timed out))
15:10:12 Join Ka__ [0] (~tkirk@
15:14:52 Part LinusN
15:14:52 Quit mattzz (Read error: 54 (Connection reset by peer))
15:22:22 Join Lethal|Bert [0] (
15:37:54 Join Domin [0] (
15:47:59 Join Hermann1 [0] (
15:49:00Hermann1Hi guys, has someone ever installed a 60/80GB HD in the Archos Recorder 20?
15:49:21Hermann1I wonder if the firmware can handle this Hard disks
15:49:32Dominas far as i know it can
15:50:35Hermann1Do you have any sources?
15:51:44Hermann1I would like to know it for sure, because I wont have a bad suprise after I bought the new HD...
15:52:15 Join mattzz [0] (
15:55:19Dominits for the player but it shouldt be the same for the recorder
15:59:17Hermann1Thanks Domin
15:59:56 Join quelsaruk [0] (~mxf@
15:59:57Hermann1Have you seen anything about 60/80GB HD? I haven`t. The page seems to be a bit outdated
16:00:00quelsarukhi again
16:00:26Hermann1Hi quelsaruk
16:01:39DominHermann1, her someone talk about it in here, so i think it works, cant se why it shouldent
16:02:01quelsarukHermann1, it's possible
16:02:12quelsarukdwihno has a 80GB hd afaik
16:02:16quelsarukzagor also has one
16:02:27dwihnoNot 80
16:02:42quelsarukthat's why i said as far as i know
16:02:50dwihnoNow you know more than far :) ? :)
16:03:10quelsarukthen you have to buy a 80GB hd
16:03:36Dominif you have to use any of the patches, you have to compile them your selvf, right ???
16:05:01quelsarukDomin, yes, if the patch hasn't been implemented in the code, you have to patch a source and compile it
16:06:44quelsarukHermann1, so the answer to your question about handling those hd, is _yes_
16:10:46Lethal|Berthmm.. i've converted quite a few avi files to rvf files so far and with a lot of them i have the problem that the sound doesnt fit to the video. is there anything i can do about it? maybe something with the framerate?
16:11:17quelsaruki have jsut updated the rvf manual in the web, if that helps :)
16:11:42quelsarukbut as i'm not the video guru.. i don't really know
16:11:56quelsaruki read yesterday something about that, i think it was in the mail list
16:12:05Hermann180GB question: Ok, good to know, thanks all! I also found some messages saying that people use 60GB HD
16:12:28mattzzquelsaruk: where can I find the rvf manual?
16:12:52quelsarukmattzz, in rockbox->documentation->how to create rvf files
16:13:37mattzzquelsaruk: shoot me... I thought there was another one....
16:14:26*quelsaruk shoots mattzz
16:17:43quelsarukmattzz, btw, i think i'm going to add a setting_items[] struct
16:18:31*mattzz wants fancy icons, whistles and bells in the sound settings dialog
16:19:01quelsaruksorry, i talked about that with c0utta
16:19:20mattzzno prob
16:19:21quelsarukhehehe... that must wait a bit
16:19:47quelsaruki thought i was talking with you, but we spoke about the rvf howto :)
16:19:58quelsarukas i said before, i'm gonna loose my head one day
16:20:17Dominanyone knows if the search function works ???
16:23:56quelsarukthe search function??
16:24:08Dominsearch function for mp3
16:24:26quelsarukbut that's not in the code right now, isn't it?
16:24:39quelsaruka patch :)
16:24:41quelsarukwell.. dunno
16:24:45quelsaruktry :)
16:24:46Dominthere are 2
16:24:56Dominhehe, dont know anything about compiling
16:25:07quelsarukwindows or linux?
16:25:58quelsarukcheck this link:
16:26:26quelsarukafterwards, check
16:26:35quelsarukand i think that's all you need to compile rockbox
16:26:55Hermann1Does anyone knows if (where) I can buy the warrenty seal which is on the screw of the jukebox? just in case I accidently scratch it..... ;-)
16:26:57 Join methangas [0] (
16:27:15Dominnice quelsaruk
16:27:21Dominthanks, will look into that
16:27:26quelsarukHermann1 i don't think so... but you could try to scratch it with a lot of care ;)
16:27:49Hermann1do you think that is possible?
16:27:56DominHermann1, try heating it a bit before you tear it off
16:28:18quelsarukDomin, i think you wont have any problems with that. Ask here if need any help
16:28:31Dominquelsaruk thanks ;-)
16:28:40quelsarukok, now i must go
16:28:59Hermann1yeah, thanks Domin, thats a good idea
16:29:17DominHermann1, works on the seal on the PS2
16:29:20quelsarukHermann1, but be careful with the box ;)
16:29:34DominHermann1, dont heat it to much thoug
16:29:37quelsarukcu l8r
16:29:45DominLater mate
16:30:17 Quit quelsaruk ()
16:32:13Hermann1ok, I will try my operation "seal" now.... B-), cu all
16:33:08Dominhave fun with it :-)
16:33:31Hermann1yes, i wil
16:33:33 Quit Hermann1 ("Leaving")
16:33:45mattzzdammit. transcode segfaults
16:36:22 Join mecraw__ [0] (~mecraw@
16:36:30***Saving seen data "./dancer.seen"
16:42:01 Join Hermann1 [0] (
16:42:33Hermann1Operation "seal" failed! :-/
16:42:41Hermann1but is doesn`t matter
16:43:02Hermann1Ive only 2 month warranty anyways left
16:43:09Dominahh, ok
16:43:19Dominthen it probably dont matter :-)
16:44:02Hermann1it`s made that way thats impossible to not viod to seal (probably because its a seal...;-)
16:46:04Lethal|Bertwhat is jörgs nickname in irc (or when is he usually around?)
16:46:05 Quit Hermann1 (Client Quit)
16:50:32 Quit Ka__ ("Leaving")
16:50:54DominHave non idea
16:52:02 Join AciD` [0] (
17:15:17 Quit adi|home (Client Quit)
17:47:21 Quit mattzz ("Client exiting")
18:22:48Lethal|Bertis there any option to turn off the function of the F1 - F3 buttons on the display (in the latest build)?
18:23:12Lethal|Berti mean the stuff that is shown there (MENU, OPTION, LCD)
18:36:33***Saving seen data "./dancer.seen"
18:38:24 Join edx{off} [0] (
18:38:44 Nick edx{off} is now known as edx{code} (
18:43:51 Nick edx{code} is now known as edx{eats} (
19:02:37 Nick edx{eats} is now known as edx{code} (
19:20:38 Join amiconn [0] (
19:21:12amiconnHi all
19:42:51 Join cjnr11 [0] (
19:43:04 Part cjnr11
19:52:18 Join Neurosupherot [0] (
19:52:50Neurosupherotim going mad with bmp2rb
20:05:26 Join quelsaruk [0] (~f74rty@
20:05:41amiconnHi quelsaruk
20:06:06quelsarukany new code to improve rockbox amiconn? :)
20:07:05DBUGEnqueued KICK quelsaruk
20:07:05quelsaruk[Domin] were you able to compile rockbox?
20:07:55amiconnquelsaruk: yep, a new bitswap (sped up about 17% - again!)
20:08:31quelsaruksomeday, we will be able to even boot rockbox before pressing the ON key ;)
20:14:12quelsarukchanging irc client
20:14:17 Quit quelsaruk (" »¡« Scøøp Script 2003 »!« pegue o seu! Grátis em <>")
20:15:37 Join Quelsaruk [0] (~Swordmast@
20:17:11 Part Neurosupherot
20:17:15 Quit methangas (" HydraIRC -> <- \o/")
20:21:03 Nick amiconn is now known as amiconn|afk (
20:22:34 Quit matsl (Remote closed the connection)
20:24:54QuelsarukLethal|Bert, jorg's nickname is [idc]dragon :)
20:25:15Lethal|Bertok.. thx :)
20:25:37Quelsarukwhat is that about turning off the functions in the F2/3 menus??
20:25:44Quelsaruki don't really understand you
20:25:45Lethal|Berti got the latest build
20:25:49Lethal|Bertbecause of the video playback
20:25:52Lethal|Bertin the main screen
20:25:58Lethal|Bertwhere it shows the root directory
20:26:05Lethal|Bertthere are this 3 new buttons on the display
20:26:14Lethal|Bertlike in the original archos firmware
20:26:25Lethal|Bertand they show the function the F1 - F3 keys have
20:26:37Lethal|Bertand i was just wondering if you can turn them off maybe
20:26:41Lethal|Bertbecause i dont need them ;)
20:26:43Quelsarukthere's an option
20:27:02 Join methangas [0] (
20:27:11Quelsarukmenu->general settings->display->status-/scrollbar->button bar
20:27:21Quelsarukjust follow that path :)
20:27:39Quelsaruki think linus forgot adding the button bar in that submenu
20:27:39Lethal|Bertoh ok
20:27:41Lethal|Bertdidnt see it
20:27:45Lethal|Bertthank you :)
20:27:56Quelsarukyou are welcome
20:28:36Quelsarukthat button bar is intented to work quite well when we have the new menus working
20:29:04Lethal|Bertnew menus?
20:29:35Quelsaruki don't know how to explain in english.. i think they are called context menus (if that word makes sense)
20:30:05Quelsaruki mean, afaik, for example F3 will call different menus, depending in what you are displaying right now
20:30:20Quelsarukif you are in WPS you'll get a menu, in the browser, another one...
20:30:29Lethal|Bertthat's cool
20:30:30Quelsarukis that called context menu?
20:30:52Lethal|Bertyou can say that. but i'm german by the way :D
20:31:12Lethal|Bertthe last time i was talking german in here i was told to talk english so everybody can understand it
20:32:09Quelsarukif i speak my alternative spanish.. nor even spanish guys would understand me ;)
20:32:20Lethal|Bertoh my
20:32:32Lethal|Berti've been going through the new video feature all day
20:32:35Lethal|Berti encoded a couple of videos
20:32:41Lethal|Bertsome work perfectly well
20:33:01Lethal|Bertand some don't (audio / video desynchronizing after a few mins)
20:33:06Lethal|Bertand i don't know why
20:33:15Lethal|Bertbecause i'm doing exactly the same when encoding them
20:34:12Quelsaruksame framerate?
20:34:24Quelsarukmaybe you changed video framerate?
20:34:34Lethal|Bertthe videos i encode have a different framerate
20:34:39Lethal|Bertbut i look it up every time
20:34:43Lethal|Bertmost have 29.97
20:34:47Lethal|Bertanother one has 20.00
20:35:03Lethal|Bertbut if i set that as my input framrate it should work (shouldnt it?)
20:35:18Quelsarukafaik, video/audio desincronization appears when you change original framerate, 'cause you change the images framerate, but not the audio (you can't speak faster or slower)
20:36:15Lethal|Berti have this movie that has 20fps.. so i set my input framerat to 20 and my output framerate to 67 for the lcd
20:36:17Lethal|Bertand it works fine
20:36:34***Saving seen data "./dancer.seen"
20:36:35Lethal|Bertand then i have another one that has 29,97 (NTSC). so i do the same with 29,97 and it doesn't work
20:36:39Lethal|Bertand i just cant figure it out
20:36:41Lethal|Bertit makes me go mad ;)
20:37:03Quelsarukthe tool is european, so it doesn't like ntsc format ;)
20:38:20Lethal|Berthmm... i could try to convert the ntsc movie to 20fps before i encode it and see if that works
20:39:07Quelsarukbut that should desincronize the video/audio
20:39:19Quelsarukfor sure
20:40:00Lethal|Bertnot if you convert it. if you just change it then it does so. but with virtual dub you can "convert" it. and with this way it makes sure video/audio wont be desincronized
20:40:18Lethal|Berti'm doing it right now
20:40:19Quelsarukyou use virtual dub??
20:40:30Lethal|Bertmaybe that is the problem?
20:40:38Lethal|Berti dont know how to resize the movie with other tools
20:40:40Quelsarukan you can use avitoyuv later??
20:40:47Lethal|Berthmm yeah
20:40:54Lethal|Berti save it as an uncompressed avi file
20:41:01Lethal|Bertbut maybe it's virtual dub
20:41:02Lethal|Berti dont know
20:41:21Quelsaruki ask because i tried virtualdubmod (the same that virtual dub) but can't make the avitoyuv work
20:41:39Quelsarukit always say "cant find frame in avi file" or something like that
20:41:52Lethal|Berti've used both so far and i got no errors but a lot of problems with the video/audio thing
20:42:04Lethal|Bertmaybe i should try something else
20:42:21Quelsarukthat's the only freeware tool i've found for windows
20:42:35Quelsaruki wanted to make a step-by-step manual for windows users
20:42:44Quelsarukbut as it doesn't work for me...
20:43:10Lethal|Bertin the rock video thing it says something about avisynth for resizing? i have avi synth but i dont know how to use it
20:43:28Quelsarukis that a freeware tool?
20:44:00Lethal|Berti think so
20:44:12Lethal|Berti once downloaded it with a software bundle which was freeware
20:44:30DominQuelsaruk, dident have time to test it today
20:44:39Dominhave to go to work in 1 houre
20:45:17DominAnd then im off for the entire weekend, but guess i will try it on sunday afternoon or perhaps monday
20:45:33Quelsarukok, just wanted to know if you managed to make it work
20:46:02QuelsarukLethal|Bert, it's freeware, so i'm going to try later
20:46:09Quelsaruki have it :P
20:47:17 Quit Nibbler (Read error: 104 (Connection reset by peer))
20:47:51Lethal|Bertwith what program do you resize the videos then?
20:51:39Quelsaruki used BPS Video converter
20:51:43Quelsarukbut it was shareware
20:51:57Quelsarukand as you can imagine, i won't buy a program if i can use another one for free
20:52:42Lethal|Bertbut i'll try it with this program if it doesnt work now with my next test
20:52:51Lethal|Bertjust to see if virtual dub was the reason it didnt work all the time
20:55:50Lethal|Bertoh my god
20:55:53Lethal|Berti think i finally got it
20:56:00Lethal|Berttook my all day
20:56:17Lethal|Bertyes yes yes :)
20:56:19Quelsarukit's easy to use
20:56:29Quelsarukyou won't have any problem with that
20:56:42Lethal|Berti mean i got it with this test (with virtual dub still)
20:56:45Lethal|Berti dont know why
20:57:15Lethal|Bertbut it seems that either virtual dub or one of the converters have problems with numbers AFTER the comma in the fps
20:57:25Lethal|Bertwhen i convert it to 20,00 fps before i encode it
20:57:27Lethal|Bertthen it works
20:57:37Lethal|Bertbut when i have something lik 29,97 or anything
20:57:40Lethal|Bertthen it wont work
21:01:33Lethal|Berti dont know why it is so
21:01:41Quelsarukso converting to 25 fps (i think that's the PAL rate) will make that work :)
21:01:50Lethal|Bertshould work
21:02:02Lethal|Bertill try it with another framerate with ,00
21:02:06Lethal|Bertto see if i was right
21:06:03Quelsarukjust one thing
21:06:27Quelsaruktry using 29,97 fps again, but viewing the file _before_ using avitoyuv
21:06:42Lethal|Bertoh i did that..
21:06:44Quelsarukso we can see if it's a problem in virtual dub or in the tools
21:07:01Lethal|Bertif i watch it before encoding it the audio/video fits perfectly well
21:07:13Lethal|Bertwith any framerate
21:07:28Lethal|Bertonly on the player the rvf file doesnt play correctly then
21:11:52Quelsarukthen, the problem is in 1 tool
21:12:30Quelsaruki think we should ask in the mailing list, to see if anyone reports the same proble
21:12:43Quelsarukso i can add that in the rvf manual
21:12:55Quelsarukdon't you think so?
21:14:22Lethal|Berti'm downloading BPS Video converter right now and i'll try to encode a file with 29,97 fps and to play it on the player
21:14:57 Nick edx{code} is now known as edx{off} (
21:16:40 Join Neurosupherot [0] (
21:17:04 Quit Neurosupherot (Client Quit)
21:17:05 Join Neurosupherot [0] (
21:17:09 Quit Neurosupherot (Client Quit)
21:19:05Quelsaruki think i'm leaving
21:19:10Quelsarukdinner time
21:21:36 Quit Quelsaruk ("Leaving")
21:28:49 Join [IDC]Dragon [0] (
21:33:09Lethal|Bertyou're the man i've been waiting for ;)
21:33:22Lethal|Berti've been going through this video plugin stuff all day today
21:33:25Lethal|Bertand i have some problems
21:33:50Lethal|Berti have this movie that has 20fps.. so i set my input framerat to 20 and my output framerate to 67 for the lcd
21:33:50Lethal|Bert[20:36] <Lethal|Bert> and it works fine
21:33:50Lethal|Bert[20:36] <Lethal|Bert> and then i have another one that has 29,97 (NTSC). so i do the same with 29,97 and it doesn't work
21:33:50DBUGEnqueued KICK Lethal|Bert
21:33:50Lethal|Bert[20:36] <Lethal|Bert> and i just cant figure it out
21:33:50Lethal|Bert[20:36] <Lethal|Bert> it makes me go mad ;)
21:33:53[IDC]Dragonsaw something in the log
21:33:58Lethal|Bertoh ok
21:34:18[IDC]Dragonare you sure you've got the framerate correct?
21:34:29Lethal|Bertbeen doing this all day
21:34:35Lethal|Berti have made a lot of checks
21:34:46Lethal|Berti checked the avi file before converting it
21:34:53Lethal|Bertand i also tried several framerates
21:34:56[IDC]Dragonwith what?
21:34:57Lethal|Berti just don't know
21:35:14Lethal|Berti'm doing exactly the same thing with every single movie. for some files it works perfectly well. and for some files it wont work
21:35:18Lethal|Berti'm using virtual dub
21:35:21 Join amiconn [0] (
21:35:22Lethal|Bertmaybe that is the problem i dont know
21:36:09[IDC]Dragonmaybe that file has skipped frames? I don't know how the toolchain deals with that.
21:36:44Lethal|Bertthat's possible.. right now im converting a movie which i have resized with another program than virtual dub
21:37:01Lethal|Bertso in a few ill see if the error was caused by the file which vd created
21:37:08[IDC]Dragonbut #1 rule is: use the correct framerate, up tho tha last fractional digit.
21:37:15Lethal|Berti did that
21:37:20Lethal|Bertall day long
21:37:27[IDC]DragonOK, OK
21:37:56Lethal|Bertbtw... this is a GREAT plug in
21:38:03Lethal|Berti totally flipped out when i saw that yesterday :)
21:38:53[IDC]DragonI have a new converter tool in the pipeline, that may work better.
21:39:59Lethal|Bertcool :)
21:40:49[IDC]DragonIt's intended to do all at once, no intermediate files.
21:41:37 Join choichoi [0] (
21:41:40[IDC]DragonMaybe I should do a pre-release of the video part, this works already.
21:41:54choichoii am french
21:42:11Lethal|Bert[IDC]Dragon> Maybe I should do a pre-release of the video part, this works already.
21:42:12Lethal|Bert[21:41] <choichoi> helo
21:42:14Lethal|Bertoops sry
21:42:31Lethal|Berta pre-release would be nice as well
21:42:51 Quit amiconn|afk (Read error: 60 (Operation timed out))
21:42:51 Nick amiconn is now known as amiconn|afk (
21:43:08 Join scott666 [0] (
21:43:11[IDC]Dragonamiconn|afk: still afk?
21:47:54Lethal|Berthmm. i just tried it with an avi file (29,97fps) that was resized with BPS Video Converter Pro (so no virtual dub at all here) and i got exactly the same error. seems to be the framerate then
21:48:46 Nick amiconn|afk is now known as amiconn (
21:49:08amiconn[IDC]Dragon: I'm a bit busy doing non-computer stuff :(
21:53:49amiconn[IDC]Dragon: I've prepared a new bitswap.S (not yet posted as a patch) that is 20% faster in theory, 17% in the real world. Is this worth additional 20 bytes of IRAM?
21:56:53amiconn[IDC]Dragon: Did you have a look at my 2 last patches (memset/memcpy and memcmp) yet?
21:57:16[IDC]DragonI was doing that right now, yes.
22:02:27amiconnI discovered a bug related to the "handover" of the mpeg buffer to a plugin and back: If a plugin calls plugin_get_mp3_buffer() while music is playing, the playback stops (obviously).
22:04:24amiconnBUT: If the plugin exits and you try to continue the music (it's set to pause then), this is not possible. Every time you hit PLAY/PAUSE, it re-pauses. You first have to press OFF to stop, then ON, PLAY to resume to get it playing again!
22:05:31[IDC]DragonHmm, the intention was to stop it, not to pause.
22:06:10[IDC]Dragonthe buffer has to be treated as trashed.
22:08:19amiconnYou can try it: Start some mp3 playing, call a plugin (e.g. your test_mp3.rock), exit the plugin, go to the wps (it shows the pause symbol at the status bar) and hit PLAY: the play symbol shows up for a fraction of a second, then pause again and no music.
22:09:02[IDC]DragonDo you happen to have a suggestion on how to improve the code?
22:11:49amiconnHmm, strange: I tried it again, this time it worked (playback was stopped) !?!?
22:19:33amiconnHmm: could not trigger that behaviour again, was apparently just a glitch while I testet my memxxx() routines.
22:21:05[IDC]DragonI'm looking at your memset()
22:21:44[IDC]DragonIs this working for <4 bytes?
22:21:51[IDC]Dragonfor 0 bytes?
22:22:17[IDC]DragonI don't see deflects from the long loop
22:22:46amiconnYes, works for all lengths (even 0) and all alignments. "100% tested and certified (TM)" by a special test plugin.
22:23:42amiconn[IDC]Dragon: At the very top there is a check if there are at least 12 bytes to set. If not, the first byte loop and the main loop are completely skipped.
22:24:02[IDC]Dragonah, I see.
22:25:38amiconnThis is even shorter than what the compiler produces: memset.c compiles to 51 instructions, memset.S are only 35 instructions.
22:25:44 Quit choichoi ()
22:27:19amiconnBtw: did you notice that bitswap() (both the version by Magnus and my first improvement) does not always work for 0 bytes?
22:27:54amiconnIf there are 0 bytes to flip _and_ the start address is odd, it always flips 1 byte!
22:28:38[IDC]Dragonlet's hope it was never called that way...
22:29:01amiconnI did not correct this behaviour for my new version, don't know if it's necessary.
22:29:21[IDC]Dragonwould it be extra cost?
22:30:15amiconnYes, 2 additional instructions.
22:30:48amiconn(a compare and a conditional branch)
22:30:59[IDC]Dragonobviously, yes
22:31:30amiconnShould I incorporate that?
22:31:38[IDC]DragonI would do it, it's not in the loop
22:32:06amiconnOk, then. My new patch will include it.
22:32:11[IDC]Dragonwho knows how that buffer wraps
22:32:22[IDC]Dragonthank you!
22:35:44amiconnSo the whole bitswap takes 2 clock cycles longer ;)
22:36:36***Saving seen data "./dancer.seen"
22:36:47 Join Nibbler [0] (
22:37:55 Quit edx{off} ()
22:49:17amiconn[IDC]Dragon: New bitswap() patch is out.
23:07:17[IDC]DragonI'm facing the task to delete files from cvs.
23:07:36[IDC]Dragon(memcpy.c / memset.c)
23:11:56amiconn[IDC]Dragon: Meanwhile I think that you can forget about my improved memcmp(), it was a nice exercise but not worth the IRAM, since it isn't used much.
23:11:56Lethal|Bertis there any program with which i can test rvf files on my pc (emulator, rvf-player or something)?
23:12:37[IDC]DragonLethal|Bert: for video-only files, yes
23:12:51[IDC]Dragonsomebody made a Java app
23:13:59Lethal|Bertoh ok... found the link on the rvf manual page.. thx
23:14:44amiconnWhich patch category should I use for fonts?
23:17:41[IDC]Dragonhmm, how about LCD?
23:17:59[IDC]Dragonwhat do you have?
23:18:50amiconn2 improved fonts, one of the you may already know from my ata test. These are improved versions of other fonts and all I ever use.
23:19:05amiconns/one of the/one of them/
23:19:21[IDC]DragonI use a modifier proportional.
23:19:48[IDC]Dragonbased on uwe_prop or so
23:20:10amiconnSee my patch, will use category=LCD then.
23:21:49amiconnBtw.: these were laying aroundon my hd for 5 months :(
23:22:40[IDC]Dragoname with my font, never filed that. I think I lost the .bdf, too
23:23:31amiconnpfaedit is really a mess under cygwin: it crashes a lot.
23:26:17[IDC]Dragonnever tried, I didn't know it can run there
23:26:38[IDC]Dragonso you're bitswapping signed now?
23:27:11amiconnWhat did you use to edit the .bdf's then? I didn't find another (free) tool for that.
23:27:28amiconn[22:49] <amiconn> [IDC]Dragon: New bitswap() patch is out.
23:27:29[IDC]Dragonpfaedit on Linux
23:27:44[IDC]Dragonyes, looking at the patch now.
23:27:48amiconnHmm, I don't have a linux box.
23:30:12[IDC]Dragonso the new bitswap trick is signed lookup and one more unroll, am I correct?
23:31:32amiconnThe signed lookup is only for saving 2 instructions for the 1 byte swapping (1 in the leading byte swap and 1 in the trailing loop). The speedup comes from unrolling only.
23:33:10[IDC]Dragonyou still do 16 bit access
23:33:22[IDC]Dragonany benefit from 32?
23:34:49amiconn16bit is faster here: 32bit access is 2 clock cycles anyway for DRAM and it would take more effort to disassemble/reassemble the bytes.
23:36:15amiconndisassemble: 16bit - mov.w, swap.b, mov.w, swap.b vs. 32bit - mov.l, swap.w, swap.b, swap.b
23:38:04amiconnreassemble: 16bit - shll8, extu.b, or, mov.w, shll8, extu.b, or, mov.w vs. 32bit - shll8, extu.b, or, shll8, extu.b, or, shll8, extu.b, or, mov.l
23:40:14[IDC]Dragonbrr, yes
23:40:36amiconnDid you notice my special trick in memcpy() in the main loop for word aligned src and dest?
23:40:56[IDC]Dragonnot yet
23:41:30amiconnHint: there is an "xtrct" instruction in it.
23:43:18amiconnBtw: For memcpy, I also tested further unrolling to 4 longwords per pass in order to make better use of the fast page mode as we've discussed recently.
23:44:29amiconnThis gives another speedup of ~20%, but also makes it significantly larger (~30 instructions). It may also cause slowdown for small blocks.
23:44:51[IDC]Dragontradeoffs, yes...
23:45:43amiconnI think it's not worth it for memcpy(), since it isn't used as much as our beloved bitswap() ;)
23:46:59[IDC]DragonI don't get your trick, I'd say you mess up the data here
23:48:15[IDC]Dragonyou store 16, 32, 16
23:48:58amiconnYes, this is the trick. At the start of the main loop for word aligned destination, I adjust the destination address to long alignment (-2).
23:49:01[IDC]Dragoninstead of 32, 32
23:49:18Lethal|Berti'm off for today. thx again to everyone who helped me today. nighty night :)
23:49:29[IDC]Dragondid we help?
23:49:54Lethal|Bertsure you did
23:50:04Lethal|Bertand the guy from spain, i dont remember his nickname atm ;)
23:50:25Lethal|Bertat least i got 5 episodes of my favorite show working on the box now :)
23:50:36amiconnLethal|Bert: Do you mean quelsaruk?
23:50:36[IDC]Dragonwhich is
23:50:42 Quit methangas (" HydraIRC -> <- Get hot chicks here!")
23:50:56Lethal|Bertyes, qualsaruk
23:51:09Lethal|Bertthe show is 8 simple rules with john ritter, who did last yr :/
23:52:05Lethal|Bertanyways.. im tired, so im goin to bed now. good night :)
23:52:31 Quit Lethal|Bert ("lat0rz")
23:52:33amiconn[IDC]Dragon: I cannot do 32,32 here, since the destination isn't long aligned. So I have to store the first and last word of the 2 longs as words.
23:53:40amiconnBut, the adjustment mentioned above enables me to store the middle 2 words of the 2 longs as one long. To get at these. I use the xtrct instruction.
23:54:37[IDC]Dragonyes, got that. I overlooked this was 16 bit align only, not 32 bit.
23:55:35amiconnIf I unroll this to 4 longwords (already tested!), there would be 3 xtrct instructions, so I could store 16/32/32/32/16.
23:57:06amiconnThere is another tricky thing: In order to not destroy register contents that I would need later (this would require additional register copying) I write each block backwards.
23:57:29[IDC]Dragonah, that again.
23:58:19amiconnIf you mean memset(): There the whole area is filled backwards.

Previous day | Next day