| 00:00:04 | preglow | amiconn: yes |
| 00:00:13 | amiconn | ugh... |
| 00:00:14 | preglow | amiconn: the nand is accessed at the lowest level |
| 00:00:21 | shotofadds | yep, there is essentially no hardware flash controller |
| 00:00:38 | * | amiconn wonders why they chose such a bare-bone approach |
| 00:00:58 | shotofadds | it certainly provides an interesting challenge... |
| 00:01:13 | amiconn | Even the ooold Ondio's built-in flash is actually an MMC in a standard chip package (bga) |
| 00:01:20 | preglow | amiconn: well, what other solutions are there, apart from nand with sd controllers and ata bridges? |
| 00:01:32 | | Quit dabujo ("( www.nnscript.com :: NoNameScript 4.2 :: www.regroup-esports.com )") |
| 00:02:14 | amiconn | One of the standard interfaces, i.e. SD, MMC, ATA (or nowadays maybe CE-ATA) |
| 00:02:14 | | Join CyBergRind|w [0] (n=cbr@212.98.160.130) |
| 00:02:28 | preglow | amiconn: i sure as hell would have preferred that |
| 00:02:30 | amiconn | Actually CE-ATA uses MMC protocol |
| 00:03:06 | amiconn | The Sansas use SD (with that proprietary banking extension) |
| 00:03:24 | | Quit jhMikeS (Read error: 104 (Connection reset by peer)) |
| 00:03:35 | | Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS) |
| 00:03:52 | * | shotofadds might just glue a 32Gb SD in the slot and use preglow's driver :) |
| 00:04:46 | | Quit davina (Remote closed the connection) |
| 00:04:59 | preglow | heh |
| 00:05:40 | amiconn | shotofadds: Are such beasts even available yet? |
| 00:05:51 | preglow | i've only seen 16 gig |
| 00:05:56 | amiconn | Actually they might, given that D2 uses full size SD |
| 00:06:14 | amiconn | 32GB CF are definitely available :) |
| 00:06:33 | preglow | in that package, no surprise |
| 00:07:48 | preglow | can you get 16 gig usd cards? |
| 00:07:51 | shotofadds | sandisk's 32gb is apparently available mid-June |
| 00:08:00 | shotofadds | SD, that is |
| 00:08:00 | preglow | at a hefty price, i'd imagine |
| 00:08:06 | shotofadds | naturally :) |
| 00:08:17 | preglow | probably more than i paid for the d2 |
| 00:08:21 | amiconn | Because it's sandisk, or what? |
| 00:08:26 | preglow | both |
| 00:08:27 | preglow | heh |
| 00:08:49 | shotofadds | sandisk SD cards are >50% more expensive than other brands, at least in the UK |
| 00:08:53 | * | amiconn doesn't think his 32GB CF was too expensive |
| 00:09:12 | preglow | sandisk are more expensive here as well |
| 00:09:13 | amiconn | Branded, but not sandisk |
| 00:09:21 | preglow | but then again, the sandisk cards are usually high-quality |
| 00:09:23 | preglow | especially the fast ones |
| 00:09:48 | | Quit ompaul (Client Quit) |
| 00:12:56 | | Join toffe82 [0] (n=chatzill@h-74-0-180-178.snvacaid.covad.net) |
| 00:13:38 | stripwax | is usb software stack still not compiled by default on pp? |
| 00:14:49 | | Quit BlakeJohnson86 (Read error: 110 (Connection timed out)) |
| 00:16:55 | | Quit crope` (Read error: 110 (Connection timed out)) |
| 00:19:01 | | Quit [CBR]Unspoken|w (Success) |
| 00:19:37 | | Join DerPapst [0] (n=Der_Paps@p5B23D3ED.dip.t-dialin.net) |
| 00:21:21 | | Quit shotofadds (Read error: 104 (Connection reset by peer)) |
| 00:21:44 | | Join shotofadds [0] (n=rob@rockbox/developer/shotofadds) |
| 00:24:24 | preglow | anyone see a reason not to commit the license free mod player? |
| 00:25:32 | | Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr) |
| 00:25:43 | preglow | ouch, it actually doesn't handle buffer wraparound? |
| 00:26:01 | | Quit shotofadds (Read error: 104 (Connection reset by peer)) |
| 00:26:28 | * | DerPapst likes it when people answer the questions thierselfes |
| 00:27:17 | pixelma | preglow: is there any other codec that forces you to skip to the next song in the playlist yourself? This is my biggest concern with that one... |
| 00:27:21 | | Join einhirn_ [0] (n=Miranda@p5B031B76.dip0.t-ipconnect.de) |
| 00:28:38 | preglow | pixelma: no, but what solutions are there? |
| 00:29:09 | | Quit einhirn_ (Read error: 104 (Connection reset by peer)) |
| 00:29:17 | Xqtftqx | Can somebody send me a dual boot loader for the S? im having trouble using mknkboot |
| 00:30:01 | pixelma | the older codec (the one with the licence problem) could handle that and the author (which was the submitter of the former too) said it should be possible to add that |
| 00:30:49 | pixelma | preglow: though it worked differently... |
| 00:31:33 | Nico_P | stripwax: no, it isn't. there are some remaining issues |
| 00:31:51 | Nico_P | Xqtftqx: does mknkboot work for you? |
| 00:32:30 | | Join jumpatrain [0] (i=juma@gateway/gpg-tor/key-0xB9002659) |
| 00:33:10 | jumpatrain | any one notice that when doing "resume playback" on a very long muspack file you get a very loud white noise result just before the music plays back? |
| 00:33:15 | preglow | pixelma: well, i don't know what mechanism that used |
| 00:33:37 | preglow | pixelma: but i'm not exactly a fan of the looping either |
| 00:35:42 | * | jhMikeS wonders if he should just put a bootloader build up since no further config changes should be needed for awhile |
| 00:36:42 | jhMikeS | of course just a raw .bin and a single boot one |
| 00:38:01 | jumpatrain | do i need voice to speak or somethin ? |
| 00:38:12 | DerPapst | no you don't |
| 00:38:15 | preglow | if we were +m, yeah |
| 00:38:16 | preglow | but we're not |
| 00:38:26 | jumpatrain | so you did see what i wrote |
| 00:38:32 | preglow | "no" :) |
| 00:38:39 | jumpatrain | bah! |
| 00:38:47 | Llorean | jumpatrain: Did you really want everyone in the room who *doesn't* have your problem to respond too, or something? |
| 00:39:00 | jumpatrain | if i remember correctly, youre the guy who wanted to port mpc 8 into rb |
| 00:39:10 | Llorean | jhMikeS: I'd wait until after powermanagement so that charging works in bootloader USB maybe? |
| 00:39:12 | preglow | jumpatrain: does that happen for just long files? |
| 00:39:22 | preglow | jumpatrain: i'd kinda expect that to happen for all musepack files |
| 00:39:25 | jumpatrain | im not sure preglow |
| 00:39:34 | jhMikeS | Nico_P: does that mutex commit address any conditions beyond funky values being shown? |
| 00:39:41 | jumpatrain | preglow then it is normal |
| 00:39:51 | Llorean | preglow: Actually, long Vorbis files have a strange noise for me when resuming, but I'd never thought about it (just in case maybe it's not strictly codec-specific) |
| 00:39:58 | preglow | jumpatrain: well, normal and normal. it's not desirable, but it might very well be happening |
| 00:40:05 | preglow | Llorean: what kind of noise? |
| 00:40:11 | preglow | file length really shouldn't matter |
| 00:40:13 | jhMikeS | Llorean: I suppose a stern warning? |
| 00:40:24 | | Quit petur ("plop") |
| 00:40:24 | Nico_P | jhMikeS: I once had some dropouts that happened at the same time the funky values were shown (after the disable hack commit) |
| 00:40:33 | preglow | white noise sounds too much, though |
| 00:40:38 | preglow | there should just be a pop if anything |
| 00:40:45 | jhMikeS | Nico_P: so that actually fixed that? |
| 00:41:11 | Nico_P | I never had it again, but I guess it's fixed if the funky values issue is actually fixed too |
| 00:41:24 | preglow | aren't tasks i watch supposed to send me mails on changes? |
| 00:41:31 | jhMikeS | thread schedule order changes will happen for sure |
| 00:41:53 | | Quit DerDome ("Leaving.") |
| 00:42:04 | Llorean | preglow: I'm not entirely sure how to describe it. It's very, very short, and considerably louder than the contents of the file, and a bit high pitched. |
| 00:42:46 | Llorean | Pop-like, I guess. |
| 00:43:05 | preglow | pop sounds plausible enough |
| 00:43:18 | jumpatrain | its no pop-like its deafening |
| 00:43:20 | * | Llorean didn't notice jumpatrain had said "white" noise. |
| 00:43:41 | preglow | jumpatrain: well, it would be helpful if you found out if it happened only for long files or not |
| 00:43:42 | | Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/mibbit.com/x-05dda1c5a062a6b6) |
| 00:43:45 | preglow | i can't make any noise happen here |
| 00:43:51 | Llorean | jhMikeS: A warning works. I was just responding to a "no further config updates should be needed" part of the statement. :) |
| 00:43:58 | jumpatrain | preglow just a second |
| 00:45:01 | jumpatrain | preglow indeed, only long files |
| 00:45:03 | jhMikeS | Llorean: I just meant that the system control coprocessor setup should be final unless the memory mapping is changed |
| 00:45:27 | jumpatrain | something in the order of 1 hour 30 minutes is what causes this white noise over here, preglow |
| 00:45:28 | preglow | jumpatrain: weird |
| 00:45:32 | Llorean | jhMikeS: Gotcha. So "future build compatibility should remain unbroken"? |
| 00:45:32 | | Quit ender` (" Kids. You gotta love them. I adore children. A little salt, a squeeze of lemon--perfect. -- Harry Dresden") |
| 00:45:51 | Llorean | preglow: That lines up with about what causes a pop in my speex files. |
| 00:45:53 | | Quit nicktastic (Read error: 113 (No route to host)) |
| 00:46:09 | preglow | buschel: for the logs, i don't have any long files, could you be bothered to check up on the mentioned bug? |
| 00:46:18 | jhMikeS | Llorean: for quite awhile barring any breakthrough on how to change the driver framebuffer address |
| 00:46:32 | Llorean | jhMikeS: Alright then, I'd say a warning about charging is good enough. :) |
| 00:46:50 | preglow | Llorean: well, if it happens for all codecs, i have no idea what causes it |
| 00:48:15 | | Quit Mathiasdm ("Yuuw!") |
| 00:50:06 | * | preglow wants his sd card to stop feeding bad crcs :/ |
| 00:50:59 | Llorean | preglow: I haven't experienced it with MP3, but my longest ones are about 1:20, so they may fall just under the mark. |
| 00:51:08 | * | Llorean isn't exactly sure where it is. |
| 00:51:15 | Llorean | I hadn't even thought twice about it until now. |
| 00:51:55 | | Quit herrwaldo (Remote closed the connection) |
| 00:53:32 | preglow | extremo-weird, the response as i get it is almost certainly correct, but cpu ardently insists it's corrupted |
| 00:53:36 | jumpatrain | preglow its not occuring on mp3 |
| 00:54:08 | preglow | well, i'm wondering why the hell file length would matter |
| 00:54:34 | jhMikeS | Llorean: what sort of header/legal should go in a README.txt file? |
| 00:56:11 | Llorean | jhMikeS: I think just installation instructions and the GPL (or reference to it). Doesn't the GPL more or less say "we offer no warranty of reliability or even usability for any purpose whatsoever" kinda stuff? |
| 00:56:58 | preglow | yes |
| 00:57:17 | Llorean | Covers all the necessary warning to make it a "if you bricked it, don't blame us" kinda thing. |
| 00:57:58 | jhMikeS | np - just a header without the $Id$ bit then I suppose |
| 00:59:46 | | Join Shaid [0] (n=adam@dsl-202-45-112-116-static.VIC.netspace.net.au) |
| 01:00 |
| 01:00:19 | | Join BitTorment [0] (n=martin@87-194-94-92.bethere.co.uk) |
| 01:01:32 | | Join crope` [0] (n=crope@dyn3-82-128-186-160.psoas.suomi.net) |
| 01:04:49 | | Join csc` [0] (n=csc@archlinux/user/csc) |
| 01:06:36 | | Quit mf0102 ("Ex-Chat") |
| 01:19:07 | | Quit n1s () |
| 01:22:58 | saratoga | I don't think IDE DMA makes up the difference in runtime for PP |
| 01:23:18 | saratoga | the Sansa has no hard disk and still gets below the retail firmware unless a very fast codec like flac or mpc is used |
| 01:23:29 | saratoga | DMA would probably help, but its not going to close the gap |
| 01:23:43 | saratoga | we waste far to much power on codec decoding |
| 01:24:44 | saratoga | and IRAM does make a big difference for PP in codecs, since most codecs just sequentially load/store memory which is still quite slow since every single cache line must be loaded and then stored |
| 01:26:00 | saratoga | for instance, when doing the TDAC part of the IMDCT windowing process in WMA, I got a ~6% speed up just by putting the samples in IRAM, even though the process only accounted for < 15% of total run time |
| 01:26:02 | | Quit Nico_P (Remote closed the connection) |
| 01:26:12 | saratoga | just because I didn't have to keep fetching and flushing cache lines |
| 01:29:45 | | Quit csc` (Remote closed the connection) |
| 01:31:40 | preglow | sure, it does matter |
| 01:31:49 | preglow | but yeah, codec efficiency might just be the thing |
| 01:32:05 | preglow | a good mp3 decoder can be twice as fast as our libmad performs on arm |
| 01:33:19 | | Quit nedd1 ("Leaving.") |
| 01:36:59 | saratoga | i need to look at mad |
| 01:38:57 | * | amiconn hates the ugly colour sequence when booting a colour target with current svn |
| 01:39:22 | preglow | it's invisible here |
| 01:40:04 | preglow | what i do hate is the ugly backlight blinking that happens when i shut down my nano |
| 01:40:45 | *** | Saving seen data "./dancer.seen" |
| 01:41:12 | | Nick fxb__ is now known as fxb (n=felixbru@h1252615.stratoserver.net) |
| 01:43:40 | toffe82 | amiconn: you never see the gigabeat X booting :) |
| 01:45:15 | | Quit Xqtftqx ("http://www.mibbit.com ajax IRC Client") |
| 01:45:21 | amiconn | whoa |
| 01:45:39 | amiconn | MPC: 482% -> 615% realtime on X5 with Buschel's patch |
| 01:46:03 | amiconn | (+ changing the #ifdef to apply on MCF5250) |
| 01:46:22 | preglow | doesn't surprise me |
| 01:46:44 | preglow | almost all the codec optimization on coldfire was done thanks to iram |
| 01:47:12 | amiconn | + EMAC |
| 01:48:25 | saratoga | wow |
| 01:48:51 | saratoga | it really is amazing how much ARM7 sucks |
| 01:49:01 | | Join csc` [0] (n=csc@archlinux/user/csc) |
| 01:49:08 | preglow | indeed |
| 01:49:12 | saratoga | i wonder how people get MP3 so fast on this damn CPU |
| 01:49:23 | saratoga | libmad already looks amazing well optimized |
| 01:49:27 | saratoga | its nearly all assembly |
| 01:49:51 | saratoga | i need to figure out a way to profile it |
| 01:51:51 | | Part toffe82 |
| 01:55:55 | jhMikeS | they take shortcuts so they can boast about decoding speed? |
| 01:57:35 | jhMikeS | had anyone compared the quality of output of libmad to retailos in any meaningful way? |
| 01:58:07 | preglow | i'm quite sure libmad's quality is better |
| 01:58:22 | preglow | but we have no good way of accessing a raw retailos decode |
| 01:58:31 | preglow | but yes, they can take shortcuts, like use 32 bit multiplies |
| 01:58:54 | preglow | i'm almost certain that almost all mp3 player firmwares keep to 16 bit processing |
| 01:59:05 | | Quit OlivierBorowski (Remote closed the connection) |
| 02:00 |
| 02:00:08 | amiconn | Well, at least on iriver H1x0 we do |
| 02:00:10 | amiconn | Just record the s/pdif signal |
| 02:01:07 | * | amiconn thinks that the coldfire mp3 decoder used in the iriver is inferior to libmad |
| 02:01:17 | saratoga | MAD can actually cheat and do lower precision multiplies too with (supposidly) only a little loss of precision |
| 02:01:20 | jhMikeS | quite honestly, I think the rockbox output sounds better and I really have no bias to prefer one or the other in evaluating it. |
| 02:01:24 | saratoga | i don't know if we use it though |
| 02:01:40 | preglow | we don't |
| 02:01:50 | amiconn | No idea about any apple OF though (as I can't try them) |
| 02:01:59 | preglow | i really don't want to cut corners like that in rockbox unless we absolutely have to |
| 02:02:13 | saratoga | some of those tricks might make sense though, since the default mad output is good to something like 120 or 130dB, which is rather absurd for MP3 |
| 02:02:16 | preglow | i think the potential for sound quality we have is one of our defining features |
| 02:02:27 | jhMikeS | if somhow you can convert some multiplies to shift + add/sub/rsb sequences, you may very well gain speed, at least if it's 32-bit |
| 02:03:00 | saratoga | the trick in mad is mostly to replace 32x32=64 multiplies with 32x32=32 multiplies |
| 02:03:03 | | Join Me [0] (n=Me@76.226.23.98) |
| 02:03:09 | saratoga | by carefully prescaling certain constants to avoid overflow |
| 02:03:26 | Me | Hey I have a question for you guys which everyone here probably gets a lot. |
| 02:03:43 | preglow | then ask it again |
| 02:04:08 | preglow | jhMikeS: most of the muls in mad are 64 bit, hard to change those with shift/add seqs |
| 02:04:09 | Me | How far is the progress in v2s? |
| 02:04:14 | preglow | Me: not far at all |
| 02:04:27 | Me | =[ no time soon? |
| 02:04:28 | jhMikeS | preglow: those are difficult indeed |
| 02:04:41 | DerPapst | i'm sure if it gets asked a lot it's documented somewhere because people are tired of answering the same question over and over again</smart_ass> ;-) |
| 02:04:44 | preglow | Me: not that i know, i think some work has tasted on deciphering the firmware format, but that's that |
| 02:04:45 | jhMikeS | It's sort of a per-context thing |
| 02:04:47 | saratoga | are 32x32 multiplies a lot faster then the 32x32=64 type? |
| 02:05:03 | preglow | saratoga: you do shave off two-three cycles by using them |
| 02:05:05 | Me | :[ |
| 02:05:19 | saratoga | preglow: thats pretty impressive |
| 02:05:24 | | Quit Me (Client Quit) |
| 02:05:24 | saratoga | it would certainly add up |
| 02:05:32 | preglow | and i'd really, really we don't use them |
| 02:05:33 | jhMikeS | preglow: "work has tasted" <==?? :p |
| 02:05:43 | saratoga | anyway, I tend to think the precision of decoders is greatly overvalued |
| 02:05:53 | preglow | if we do, i would at least expect some very favorable comparisons |
| 02:06:01 | preglow | saratoga: even when doing dsp on the output? |
| 02:06:02 | saratoga | markun broke the WMA decoder last fall in such a way that reduces precision by about 40dB |
| 02:06:08 | saratoga | no one has ever noticed |
| 02:06:26 | preglow | saratoga: that was fixed rather quick, wasn't it? |
| 02:06:33 | saratoga | no i haven't bothered |
| 02:06:40 | jumpatrain | how exactly do you do dsp on the output ? |
| 02:06:59 | amiconn | saratoga: Actually multiplication speed on arm depends on how many bytes of one of the operands are pouplated. Here are the details: I tried the patch on X5 (Coldfire MCF5250, also having 128KB or IRAM). It sped up decoding of my test track from 482% realtime to 615% realtime! |
| 02:06:59 | amiconn | I've attached the modified patch, which also corrects the checks for PP5022 and PP5024. Those macros don't work like the CPU_* group macros. They are always defined, which would cause v1 to break on any other CPU than PP5022/PP5024. You're supposed to check CONFIG_CPU. |
| 02:06:59 | amiconn | Regarding IRAM distribution - I'd prefer to keep even distribution between core and codecs/plugins, like it's already done in SVN for MCF5250. While IRAM might be quite useful for codecs, it might also be quite useful in the core (e.g. for better DSP code), and there are often also some target specific tasks in the core which profit from extra IRAM. |
| 02:07:02 | saratoga | i wanted to look at other ways to attempt his optimization before i reverted it |
| 02:07:09 | jhMikeS | "exactly"? I think the source is needed for that. |
| 02:07:12 | amiconn | Bah, what was that! :( |
| 02:07:23 | preglow | amiconn: looks like a paste to me :) |
| 02:07:27 | amiconn | yeah |
| 02:07:42 | amiconn | But not the one I wanted to do |
| 02:07:52 | saratoga | amiconn: since core uses only a very tiny percentage of total CPU cycles, I'm curious what the use of more IRAM would be? |
| 02:08:03 | | Quit kennyj (Remote closed the connection) |
| 02:09:05 | saratoga | i mean, buschel and I's patches, each save more CPU cycles the entire core uses for typical decode (ignoring resampling and EQ I suppose)\ |
| 02:09:19 | amiconn | I already mentioned the main purpose (DSP). |
| 02:09:34 | preglow | i don't know how wise it is to keep reserving iram we might never use, though |
| 02:09:36 | amiconn | There are target specific things like framebuffer |
| 02:09:54 | saratoga | do any of our targets with 128k need that though? |
| 02:10:16 | amiconn | Even the 16K extra for codecs are currently unused on MCF5250 |
| 02:10:31 | amiconn | I only used them once for an APE test |
| 02:11:03 | saratoga | for PP at least, I don't believe we use even the 48k we've reserved now, so I don't see much point in adding more |
| 02:11:19 | saratoga | and for coldfire, it would be nice to use the same split, so that optimizations apply to the X5 as well |
| 02:11:31 | amiconn | So I'd keep even distribution for now, and if it later turns out that it's better used for codecs (or for the core), we can change that |
| 02:11:35 | jumpatrain | i can hear noise when im listening to silence on this sansa |
| 02:11:38 | jumpatrain | is that normal ? |
| 02:11:56 | saratoga | amiconn: I've got a good use for the extra 32KB right now |
| 02:12:00 | saratoga | the WMA codec . . . |
| 02:12:15 | saratoga | i can put the output buffer in IRAM and speed up both decoding and DSP operations |
| 02:12:25 | kkurbjun | Is a 12% power savings worth a little extra LCD startup time after it turns off? |
| 02:12:49 | kkurbjun | I mean how much would that bug people on the gigabeat F do you think? |
| 02:12:53 | preglow | how much? |
| 02:13:12 | saratoga | presumably the same trick would be applicable to AAC too, since it uses nearly the same windowing process |
| 02:13:14 | scorche|sh | 12% is pretty big.. |
| 02:13:18 | pixelma | jumpatrain: yes, unfortunately the Sansas are a bit noise (a bit depends on your exact player) |
| 02:13:26 | pixelma | s/noise/noisy |
| 02:13:36 | kkurbjun | it's a split second after the backlight fades in.. you see a white screen for a brief time and then the normal screen comes up |
| 02:13:38 | preglow | saratoga: aac would love iram |
| 02:13:54 | jumpatrain | pixelma sansa e280. |
| 02:13:55 | jumpatrain | :( |
| 02:13:57 | saratoga | i mean pretty much any pure IMDCT codec should be able to very profitablely use more then 64 kbps of IRAM (as in use it 4 or 5 times per sample instead of DRAM) |
| 02:14:14 | amiconn | saratoga: If you reduce core iram to 48KB, the M5 will be quite tight (X5 has a bit more room) |
| 02:14:23 | saratoga | because of it's frame buffer? |
| 02:14:56 | saratoga | for a pure IMDCT codec, you optimally need 48KB of IRAM just for the IMDCT |
| 02:15:07 | saratoga | and thats not counting any of the FFT constants or windowing or any of that |
| 02:15:21 | saratoga | 48KB just for the sample data, since each pass of the IMDCT processes 48KB worth of data |
| 02:15:41 | kkurbjun | preglow, scorche, I tried to eliminate the startup time, but it seems to be a limitation of the on glass LCD controller, not the one in the SOC. I could commit it first and see how people take to it |
| 02:15:57 | preglow | kkurbjun: yeah, but how much extra start time? |
| 02:16:02 | scorche|sh | kkurbjun: if it is a split second for 12%..... |
| 02:16:08 | saratoga | with 64 KB, you can start to put things like FFT constants, Windowing constants, and trig tables in IRAM too |
| 02:16:09 | preglow | split second can be anything :> |
| 02:16:15 | kkurbjun | it's probably on the order of 15 ms |
| 02:16:21 | kkurbjun | about |
| 02:16:21 | saratoga | at 80KB you can fit virtually everything and avoid DRAM pain altogether |
| 02:16:24 | preglow | *shrug* |
| 02:16:27 | preglow | see what people thinj |
| 02:16:47 | saratoga | 15 ms is nothing |
| 02:17:21 | | Quit dan_a (Success) |
| 02:17:37 | amiconn | saratoga: M5 iram end address is currently 0x1000bcd4, so it's only 812 bytes away from the 48K border |
| 02:17:55 | amiconn | All other cf targets are lower than that though |
| 02:18:13 | pixelma | jumpatrain: I actually meant small differences in the production, so your unit. Some people hear something, some not, probably has to do with earphones and ears too... |
| 02:18:16 | kkurbjun | ok, I'll commit it and see if anyone really hates it :-D |
| 02:18:19 | saratoga | amiconn: how much would you save by putting the buffer in DRAM like on color targets? |
| 02:18:26 | amiconn | I'm not entirely sure, but I think this is because both the main and remote framebuffers are in iram |
| 02:19:02 | amiconn | While it's the same situation as on H1x0, the M5's remote framebuffer is larger than the H10's |
| 02:19:04 | * | jhMikeS meant to recheck if digital or analogue volume is attenuated first on AS3514 |
| 02:19:07 | amiconn | *H1x0's |
| 02:19:21 | | Nick fxb is now known as fxb__ (n=felixbru@h1252615.stratoserver.net) |
| 02:19:59 | saratoga | is anyone here familar enough with FFT algorithms to tell me what type of FFT fft.c is in libwma? |
| 02:20:13 | | Quit gevaerts ("bedtime") |
| 02:20:25 | preglow | not really |
| 02:20:35 | preglow | i'm usually more than satisfied with just using them |
| 02:20:42 | amiconn | saratoga: simple calculation: Main fb is 160*128*2/8 == 5120 bytes. Remote fb is 128*96*2/8 == 3072 bytes |
| 02:20:44 | jhMikeS | hmmm, HP is primary, then it's mixer |
| 02:21:59 | saratoga | i think its radix 2 ? |
| 02:22:37 | preglow | anything else would be foolish |
| 02:22:49 | preglow | wma only uses blocks that are power of two, afaik |
| 02:22:55 | saratoga | well theres radix 4 and split radix |
| 02:23:06 | saratoga | you can in theory do quite a bit better then radix 2 |
| 02:23:37 | preglow | split radix is usually the best on ordinary cpus |
| 02:23:48 | preglow | well, whatever this fft is, it surely doesn't look optimal |
| 02:24:39 | saratoga | yes |
| 02:25:45 | | Join DaCapn [0] (n=dacapn@c-67-170-177-22.hsd1.or.comcast.net) |
| 02:26:22 | | Join toffe82 [0] (n=chatzill@adsl-70-235-225-95.dsl.frs2ca.sbcglobal.net) |
| 02:27:58 | | Nick JdGordon|zzz is now known as JdGordon (i=jonno@rockbox/developer/JdGordon) |
| 02:28:07 | preglow | bedtime |
| 02:28:56 | | Part pixelma |
| 02:29:14 | JdGordon | oh bloody joy.. 60 forum views and not a single comment :/ |
| 02:32:29 | | Join kennyj [0] (i=kennyj@72.214.22.184) |
| 02:40:04 | | Join Gletob13 [0] (n=Gletob@c-71-206-137-72.hsd1.va.comcast.net) |
| 02:41:54 | Llorean | JdGordon: Maybe try the mailing list? |
| 02:42:04 | | Join z35 [0] (n=z@149.123.33.65.cfl.res.rr.com) |
| 02:42:43 | JdGordon | na, that list is never fun :) |
| 02:44:01 | Llorean | Misticriver? :-P |
| 02:44:20 | JdGordon | AAAAAAAAAAAAAAAAAAAAAAAHHHHHHHHHH! |
| 02:44:25 | JdGordon | *runs away crying* |
| 02:48:24 | | Quit Jon-Kha (leguin.freenode.net irc.freenode.net) |
| 02:48:24 | NSplit | leguin.freenode.net irc.freenode.net |
| 02:48:24 | | Quit Slasheri (leguin.freenode.net irc.freenode.net) |
| 02:48:24 | | Quit suom1 (leguin.freenode.net irc.freenode.net) |
| 02:48:24 | | Quit Hadaka (leguin.freenode.net irc.freenode.net) |
| 02:49:15 | NHeal | leguin.freenode.net irc.freenode.net |
| 02:49:15 | NJoin | Jon-Kha [0] (i=jon-kha@xdsl-83-150-91-127.nebulazone.fi) |
| 02:49:15 | NJoin | Slasheri [0] (i=miipekk@rockbox/developer/Slasheri) |
| 02:49:15 | NJoin | suom1 [0] (i=suom1@irssi.mobi) |
| 02:49:15 | NJoin | Hadaka [0] (i=naked@naked.iki.fi) |
| 02:51:08 | kugel | JdGordon: You want me to test your patch? |
| 02:52:48 | kugel | JdGordon: I'm not entirely aware what it does though |
| 02:54:10 | | Quit Gletob13 (Read error: 104 (Connection reset by peer)) |
| 02:54:21 | | Join Gletob13 [0] (n=Gletob@c-71-206-137-72.hsd1.va.comcast.net) |
| 02:54:35 | Llorean | kugel: It really sorta needs people who are already familiar with the screen to see if any old functionality is broken, etc. |
| 02:55:55 | DerPapst | poor JdGordon... |
| 02:55:56 | | Quit saratoga ("CGI:IRC (Ping timeout)") |
| 02:56:13 | | Quit Gletob13 (Read error: 104 (Connection reset by peer)) |
| 02:56:16 | | Join Gletob13 [0] (n=Gletob@c-71-206-137-72.hsd1.va.comcast.net) |
| 03:00 |
| 03:05:08 | | Quit jumpatrain (Remote closed the connection) |
| 03:05:48 | kkurbjun | :), I lied, I should have said up to 17% less power, I made an assumption on the current without the LCD disable stuff, and it turns out it was higher :P. |
| 03:07:23 | | Join jumpatrain [0] (i=juma@gateway/gpg-tor/key-0xB9002659) |
| 03:07:50 | | Quit Gletob13 ("Leaving") |
| 03:13:03 | | Quit jumpatrain (Remote closed the connection) |
| 03:14:23 | | Join jumpatrain [0] (i=juma@gateway/gpg-tor/key-0xB9002659) |
| 03:18:23 | DerPapst | anybody wanna have a dump of the flash of a 2nd gen nano? http://home.gna.org/linux4nano/index.html |
| 03:27:23 | | Quit homielowe () |
| 03:40:25 | | Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb) |
| 03:40:50 | *** | Saving seen data "./dancer.seen" |
| 03:41:12 | | Quit Spiffelight ("if ($Temp > ('pants');") |
| 03:44:54 | | Join staen [0] (n=59f5c5c8@gateway/web/cgi-irc/labb.contactor.se/x-998298973a71e5f9) |
| 03:45:14 | | Join fdinel [0] (n=Miranda@modemcable002.173-131-66.mc.videotron.ca) |
| 03:45:35 | staen | Hi! |
| 03:46:41 | | Quit staen (Client Quit) |
| 03:48:08 | | Quit Hadaka (Read error: 145 (Connection timed out)) |
| 03:49:39 | | Join Hadaka [0] (i=naked@naked.iki.fi) |
| 03:53:14 | | Quit PaulJam (".") |
| 03:54:40 | | Quit linuxstb (Read error: 110 (Connection timed out)) |
| 04:00 |
| 04:00:37 | | Join dan_a [0] (n=dan-mirc@217.23.173.156) |
| 04:04:50 | | Quit Thundercloud (Remote closed the connection) |
| 04:07:58 | | Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey) |
| 04:16:04 | | Quit MU{lappy} (Read error: 113 (No route to host)) |
| 04:16:43 | | Join inigomontoya [0] (n=621cdcfd@gateway/web/cgi-irc/labb.contactor.se/x-eaacae9ad4d81a42) |
| 04:17:16 | | Quit dan_a () |
| 04:17:24 | | Join jhulst [0] (n=jhulst@unaffiliated/jhulst) |
| 04:23:56 | | Join djh_ [0] (i=zippy@c-71-224-103-75.hsd1.pa.comcast.net) |
| 04:28:06 | | Quit inigomontoya ("CGI:IRC (EOF)") |
| 04:33:36 | | Join miepchen^schlaf_ [0] (n=miepchen@p54BF55CC.dip.t-dialin.net) |
| 04:43:23 | | Join toffe82_ [0] (n=chatzill@adsl-70-137-196-185.dsl.frs2ca.sbcglobal.net) |
| 04:48:02 | | Quit miepchen^schlaf (Read error: 110 (Connection timed out)) |
| 04:51:27 | | Quit XavierGr () |
| 04:58:31 | | Quit toffe82 (Read error: 110 (Connection timed out)) |
| 05:00 |
| 05:01:43 | | Quit krazykit (Remote closed the connection) |
| 05:03:48 | | Join krazykit [0] (n=kkit@adsl-76-240-192-86.dsl.ipltin.sbcglobal.net) |
| 05:09:52 | | Quit tchan (Read error: 104 (Connection reset by peer)) |
| 05:10:19 | | Part jumpatrain ("Leaving") |
| 05:11:01 | | Join mackes [0] (n=root@cpe-24-198-43-238.buffalo.res.rr.com) |
| 05:11:14 | | Join tchan [0] (n=tchan@lunar-linux/developer/tchan) |
| 05:12:41 | | Quit Horscht ("User was distributing pornography on server; system seized by FBI") |
| 05:40:52 | *** | Saving seen data "./dancer.seen" |
| 05:44:47 | | Quit csc` (Remote closed the connection) |
| 05:49:46 | | Join MU{lappy} [0] (n=Militant@pool-72-70-183-80.hrbgpa.fios.verizon.net) |
| 05:55:02 | | Quit fdinel ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org") |
| 05:58:17 | | Join nuonguy [0] (n=john@c-24-6-175-26.hsd1.ca.comcast.net) |
| 05:58:36 | | Quit DerPapst (Nick collision from services.) |
| 05:58:46 | | Join DerPapst_ [0] (n=Der_Paps@p5B23F651.dip.t-dialin.net) |
| 06:00 |
| 06:14:58 | | Quit kkurbjun (Read error: 104 (Connection reset by peer)) |
| 06:15:45 | | Quit mackes ("Mackes is out") |
| 06:23:44 | | Quit kugel ("ChatZilla 0.9.82 [Firefox 3.0b5/2008050509]") |
| 06:35:44 | | Join linuxstb__ [0] (n=linuxstb@i-83-67-212-170.freedom2surf.net) |
| 06:45:49 | | Join m0f0x [0] (n=m0f0x@189-47-0-88.dsl.telesp.net.br) |
| 06:49:02 | | Quit linuxstb_ (Read error: 110 (Connection timed out)) |
| 06:57:08 | | Join cool_walking_ [0] (n=notroot@203-59-129-195.perm.iinet.net.au) |
| 07:00 |
| 07:07:39 | | Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb) |
| 07:07:57 | | Part toffe82_ |
| 07:15:14 | | Join pondlife [50] (n=Steve@rockbox/developer/pondlife) |
| 07:15:41 | | Quit jhulst ("Konversation terminated!") |
| 07:21:57 | | Quit linuxstb__ (Read error: 110 (Connection timed out)) |
| 07:38:01 | | Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb) |
| 07:40:56 | *** | Saving seen data "./dancer.seen" |
| 07:53:27 | | Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul) |
| 07:54:16 | | Quit BHSPitLappy (Remote closed the connection) |
| 08:00 |
| 08:05:24 | | Quit ompaul (Client Quit) |
| 08:05:28 | | Quit DerPapst_ (Read error: 113 (No route to host)) |
| 08:07:50 | | Join Buschel [0] (n=abc@p54A3DEC1.dip.t-dialin.net) |
| 08:08:58 | Buschel | amiconn: your results are pretty impressive :o) +27% |
| 08:09:26 | amiconn | Hardly surprising.... |
| 08:09:33 | Buschel |
icon identifies that the person is a core developer (has commit access).