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 2012-10-04

00:23:14 Quit prof_wolfff (Ping timeout: 248 seconds)
00:24:23 Join factor [0] (
00:27:02 Quit pamaury (Ping timeout: 246 seconds)
00:27:35 Quit bertrik (Remote host closed the connection)
00:38:06 Nick froggyman is now known as freshman (~froggyman@unaffiliated/froggyman)
00:38:36 Nick freshman is now known as froggyman (~froggyman@unaffiliated/froggyman)
00:55:54 Join mystica555 [0] (
00:55:56 Quit mystica555 (Excess Flood)
00:56:31 Join amayer [0] (
01:08:15 Quit XavierGr (Ping timeout: 240 seconds)
01:17:31 Quit amayer (Quit: Going ~/)
01:17:45 Join Thra11_ [0] (
01:17:45 Join amayer [0] (
01:25:59 Part amayer
01:34:39 Quit lebellium (Quit: ChatZilla 0.9.89 [Firefox 16.0/20120925201946])
01:36:17 Nick LittleCreature is now known as SpiderFromMars (~fearofmus@unaffiliated/fearofmusic)
01:54:27***Saving seen data "./dancer.seen"
02:07:14 Quit sakax (Ping timeout: 248 seconds)
02:08:20 Quit eckoit (Quit: eckoit)
02:21:13 Join sakax [0] (
02:28:25 Join saratoga [0] (123e0cca@gateway/web/freenode/ip.
02:28:43saratogawe should really make the codec profiling flags and instrumentation function calls a configure time option
02:28:53saratogasilly to have to add them manually when the preprocessor could do it
02:29:49derfsaratoga: So, if I've done things right, the actual amount of memory needed for the global stack, in fixed-point mode, assuming you aren't using the multistream API, is 49806 bytes.
02:30:14saratogaplus the stuff needed for ogg? or including that?
02:30:19derfThat will go down somewhat by moving the MDCT scratch buffers into IRAM instead of using the global stack.
02:30:26derfsaratoga: libogg does not use the pseudostack.
02:30:43saratogaFWIW, the targets that are most memory constrained do not use IRAM
02:31:01derfThat's good, because figuring out exactly how much they go down is complicated.
02:31:12saratogai would just ignore IRAM
02:31:26saratogaanything with IRAM has plenty of memory
02:32:38derfAnyway, here's the analysis:
02:32:55derfThat's for all API entry points, in both fixed-point and float mode.
02:33:14derfSo the only number you should actually care about is opus_decode: Fixed:.
02:34:01derfThere's also some stupidity we can eliminate to make that go down a bit.
02:36:35 Join eckoit [0] (
02:37:15 Quit mc2739 (Ping timeout: 256 seconds)
02:39:06 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
02:42:20 Quit perrikwp (Ping timeout: 246 seconds)
02:47:30 Join perrikwp [0] (
02:52:02 Quit perrikwp (Ping timeout: 248 seconds)
02:57:35 Join perrikwp [0] (
03:05:12 Quit perrikwp (Ping timeout: 260 seconds)
03:08:07 Join someguy [0] (
03:08:21 Quit eckoit (Quit: eckoit)
03:08:45someguyRockbox utility can't find my clip zip
03:09:13someguysays it's not mounted even though it clearly says it's on D
03:22:22derfI like that rockbox keeps changing the duration of this file as I listen to it.
03:22:44 Join mikroflops_ [0] (
03:24:37someguyI'm stupid, forgot to change to msc
03:24:44someguyNever had that happen before
03:25:45 Quit mikroflops (Ping timeout: 260 seconds)
03:26:45 Quit someguy ()
03:30:56 Quit Thra11_ (Ping timeout: 252 seconds)
03:34:24 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
03:46:46 Quit [Saint] (Remote host closed the connection)
03:50:10 Join [Saint] [0] (~saint@rockbox/user/saint)
03:53:38 Join perrikwp [0] (
03:54:30***Saving seen data "./dancer.seen"
03:58:03 Join eckoit [0] (
04:04:09 Join perrikwp_ [0] (
04:06:44 Quit perrikwp (Ping timeout: 248 seconds)
04:10:55 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:10:56 Quit amiconn (Disconnected by services)
04:10:57 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:10:58 Quit pixelma (Disconnected by services)
04:10:59 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
04:27:05 Quit eckoit (Quit: eckoit)
04:38:21 Join TheSphinX^ [0] (
04:41:23 Quit TheSphinX_ (Ping timeout: 245 seconds)
04:42:14 Join amayer [0] (
04:46:22 Join mystica555 [0] (
04:46:24 Quit mystica555 (Excess Flood)
04:56:39 Join mikroflops [0] (
04:59:45 Quit mikroflops_ (Ping timeout: 244 seconds)
05:13:13 Quit Epicanis (Quit: bedtime)
05:18:34 Quit TheSeven (Disconnected by services)
05:18:43 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
05:44:16 Quit pedro_angelo (Read error: Connection reset by peer)
05:54:31***Saving seen data "./dancer.seen"
05:59:12 Quit factor (Read error: Connection reset by peer)
06:00:39 Join factor [0] (
06:10:10 Part amayer
06:37:54 Join eckoit [0] (~ryan@
07:18:47 Join mortalis [0] (
07:32:49 Join stoffel [0] (
07:54:32***Saving seen data "./dancer.seen"
08:09:41 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
08:11:19 Join Buschel [0] (
08:18:03Buscheln1s: thanks for the hint. now test_codec takes ages for the 64 kbps file. so, i expect to have some result in a few minutes
08:22:28 Quit Prodicus (Ping timeout: 252 seconds)
08:27:16 Join LinusN [0] (
08:33:36 Join jsge [0] (
08:34:03Buschelhmm, got the output file, but it seems i do not have any arm-elf-objdump? where should this be located?
08:34:26 Quit jsge (Client Quit)
08:36:58 Join Zagor [0] (
08:36:58 Quit Zagor (Changing host)
08:36:58 Join Zagor [242] (~bjst@rockbox/developer/Zagor)
08:47:21Buschelgot it ->
08:52:29Buschelmain parts: 27% for mdct pre-/post rotation, deshuffling and mirroring + 40% for ifft
08:57:16 Join ender` [0] (
09:01:55 Join Rower85 [0] (husvagn@
09:02:58 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
09:07:27 Join petur [0] (~petur@rockbox/developer/petur)
09:08:46 Quit Provel (Ping timeout: 245 seconds)
09:11:12 Join Provel [0] (
09:15:04Buscheli did an new profiling with all ifft-butterfly functions as non-static. you can clearly see that the N=4 and N=5 butterflies dominate the ifft ->
09:16:02Buscheli will later try toi break down clt_mdct_backward into single steps to gather where the most cpu time is consumed
09:17:15Buschelbtw, all measurements were done based on the open −− not yet pushed −− patch
09:22:07 Quit Buschel (Ping timeout: 240 seconds)
09:52:23 Quit gxk (Quit: leaving)
09:54:36***Saving seen data "./dancer.seen"
09:57:52 Join XavierGr [0] (XavierGr@rockbox/staff/XavierGr)
10:05:06 Quit [Saint] (Quit: No Ping reply in 180 seconds.)
10:05:32 Join [Saint] [0] (~saint@rockbox/user/saint)
10:06:48 Quit TBCOOL (Quit: hrm)
10:19:52 Join Buschel [0] (
10:21:56 Quit advcomp2019__ (Read error: Connection reset by peer)
10:22:20 Join advcomp2019__ [0] (~advcomp20@unaffiliated/advcomp2019)
10:34:11 Join JdGord [0] (~AndChat80@
10:36:04 Quit mgottschlag (Ping timeout: 246 seconds)
10:39:26 Quit JdGord (Read error: Connection reset by peer)
10:41:30 Join TBCOOL [0] (
10:42:56 Join JdGord [0] (~AndChat80@
10:45:29 Quit JdGord (Read error: Connection reset by peer)
10:47:17 Join pamaury [0] (
10:47:17 Quit pamaury (Changing host)
10:47:17 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
10:47:52 Join wodz [0] (
10:48:11 Join JdGord [0] (~AndChat80@
10:50:16 Quit JdGord (Read error: Connection reset by peer)
10:53:46 Join JdGord [0] (~AndChat80@
10:55:50 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
10:55:51 Quit JdGord (Read error: Connection reset by peer)
10:59:12 Join lebellium [0] (
11:04:10 Join JdGord [0] (~AndChat80@
11:06:38 Quit JdGord (Read error: Connection reset by peer)
11:11:15wodzpamaury: Your atj tool is a bit ackward to use. If you want to unpack afi/fw you need to specify prefix with -o which is appended to *every* unpacked filename so to 'simply' unpack you have to pass -o ''
11:12:17pamauryI did that because if you run it without knowing, you end up with 20+ files on the directory and that's a pain to remove
11:12:26pamaurybut if you prefer I can change this behaviour
11:12:28 Join mikroflops_ [0] (
11:12:39 Join JdGord [0] (~AndChat80@
11:12:39pamauryor put a "extract here" flags
11:12:45 Join gxk [0] (
11:14:06wodzpamaury: ideally would be to specify dir with -o for afi/fw so you end up with all files from archive in single place
11:14:50pamauryand to extract in the current dir for fwu file ?
11:15:13pamaurynote that you can always do -o ./
11:15:29wodzyes, fwu is descrambled actually
11:15:53 Quit mikroflops (Ping timeout: 252 seconds)
11:15:55wodzit will not produce bilion files
11:16:06pamauryok, so what should be the default file name then ?
11:18:03wodzmaybe append .descrabled or something to input filename
11:19:40pamauryI'll do that and we keep the behaviour for the other archives right ?
11:21:55wodzmy proposal is to 1) unpack afi/fw archives to directory (and not mangling filenames) 2) descramble .fwu in current dir with appended .something unless -o is explicitly given
11:24:52 Quit kevku (Remote host closed the connection)
11:27:29 Quit petur (Quit: *plop*)
11:29:29 Quit y4n (Quit: We're fucking 3LN!)
11:32:58 Quit JdGord (Read error: Connection reset by peer)
11:36:19 Join JdGord [0] (~AndChat80@
11:38:36pamaurywodz: you mean unpack to current directory ?
11:38:37 Quit JdGord (Read error: Connection reset by peer)
11:39:35wodzit should default to something like afi dir for AFI and FW dir for FW files which could be changed with -o
11:41:06pamaurywhat about the filename without extension ?
11:41:52 Join JdGord [0] (
11:42:12wodzyou inspect the content and determine the type anyway, no?
11:43:00pamaurybut for the archives we could extract by default in a directory named after the file
11:43:21pamauryif you extract fwimage.bin then it's unpacked into fwimage/
11:44:00wodzyes thats smart
11:44:18pamauryok, action now
11:45:33wodzdo you support FW only of the type found in e150 FW or all 4 types you mentioned?
11:46:55pamauryonly the e150
11:47:04 Quit bluebrother (Disconnected by services)
11:47:06pamauryI can"t support the others without files to test on it
11:47:09 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
11:47:14pamaurybut if you have some, feel free to send them to me
11:47:39pamauryby the way, I haven't found the e100 firmware on internet
11:48:55wodzpamaury: I was asking just in case it will be something s1mp3 people may be interested
11:49:18pamaurythe iriver website mentions it but I havent't found the actual link, maybe I'm dumb; sure it would be nice to support other firmwares
11:50:14 Quit fs-bluebot (Ping timeout: 260 seconds)
11:51:39 Join fs-bluebot [0] (
11:54:39***Saving seen data "./dancer.seen"
11:55:12wodzbut it IS tricky to find
11:55:50pamauryoh god, it will take one hour to download at this speed; thanks
11:57:06wodz <- this one is probably slightly newer version
11:57:55pamauryhow did you find those ?
11:59:24wodzgoogle magic :-)
11:59:27 Quit JdGord (Quit: Bye)
12:08:37 Join kevku [0] (
12:28:29pamaurywodz: done
12:36:21Buscheli now did a profiling with opus mdct split into seperate functions (pre-rotate, post-rotate, de-shuffle, mirroring part 1 and mirroring part 2). it is very clear that −− on PP502x −− only pre- and post-rotate are of interest. ->
12:38:57 Join Topy [0] (
12:39:49 Join Thra11_ [0] (
12:42:04 Quit T44 (Ping timeout: 246 seconds)
12:44:00pamaurywodz: indeed, the files are version 10.40 and 10.50; I can successfully extract fwu, afi and fwimage files
12:45:13pamaurywodz: for you information, the iriver E40 is imx233 based without encryption. So if you want to work on irivers, this one is a safe bet :)
12:45:42pamaury*for your
12:46:02wodzpamaury: I would like to get in touch with MIPS - thats why E1x0 is interesting for me
12:46:16pamauryok fin
13:23:55 Quit mgottschlag (Ping timeout: 246 seconds)
13:32:17 Join Prodicus [0] (
13:44:05 Join larsie [0] (
13:48:12 Quit larsie (Client Quit)
13:50:40 Quit stoffel (Ping timeout: 246 seconds)
13:54:41***Saving seen data "./dancer.seen"
13:57:12 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
14:08:45wodzcrap, ld 2.17 doesn't support −−gc-sections and -r at the same time. It doesn't support −−print-gc-sections also so I could tell what is actually discarded :/
14:09:15Torneit doesn't?
14:09:37Tornedoes it actually just error, or do you just mean it doesn't gc anything?
14:09:55wodzit errors
14:10:20wodzIt seems more recent ld doesn't have this limitation
14:13:25wodzyep, our arm toolchain is happy with −−gc-sections and -r together.
14:14:10wodzso SH and MIPS toolchains are problematic
14:18:49 Join amayer_ [0] (
14:21:54 Quit factor (Read error: Connection reset by peer)
14:23:38 Quit Topy (Ping timeout: 260 seconds)
14:24:44 Join Topy [0] (
14:26:44 Part amayer_
14:28:38 Join T44 [0] (
14:29:47 Quit hype (Ping timeout: 240 seconds)
14:30:14 Join hype [0] (~hype@
14:31:13 Quit Topy (Ping timeout: 240 seconds)
14:35:23kugelfinally a reason to upgrade sh!
14:35:44 Quit Prodicus (Read error: Connection reset by peer)
14:36:13 Join Prodicus [0] (
14:42:51 Join pedro_angelo [0] (
14:44:25 Join dfkt [0] (dfkt@unaffiliated/dfkt)
14:46:36 Quit Zagor (Quit: Clint excited)
14:50:25 Quit hype (Quit: ["Textual IRC Client:"])
14:54:23 Join amayer_ [0] (
15:01:40 Quit wodz (Read error: Connection reset by peer)
15:04:03 Quit pedro_angelo (Remote host closed the connection)
15:12:50 Quit sciopath (Read error: Connection reset by peer)
15:13:17 Join sciopath [0] (
15:15:29 Quit Buschel (Quit: ChatZilla [Firefox 15.0.1/20120905151427])
15:45:50 Quit mortalis (Quit: Leaving)
15:51:02 Part LinusN
15:54:43***Saving seen data "./dancer.seen"
15:58:56 Join stoffel [0] (
16:12:25 Part amayer_
16:29:47 Join amayer_ [0] (
17:22:56 Join hype [0] (~hype@
17:54:44***Saving seen data "./dancer.seen"
17:54:59 Quit stoffel (Ping timeout: 248 seconds)
18:07:20 Join einhirn [0] (
18:10:26 Nick desowin_ is now known as desowin (
18:12:33 Join bootlfdhfrsa [0] (~Prmhfhfx@
18:13:43 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
18:21:27 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
18:37:46 Quit AlexP (Ping timeout: 256 seconds)
18:38:09 Join AlexP [0] (~alex@rockbox/staff/AlexP)
18:43:38 Join n1s [0] (
18:43:43 Quit n1s (Changing host)
18:43:43 Join n1s [0] (~n1s@rockbox/developer/n1s)
18:44:21 Quit XavierGr (Ping timeout: 244 seconds)
18:45:43 Quit sciopath (Read error: Connection reset by peer)
18:46:15 Join sciopath [0] (
18:49:37 Join XavierGr [0] (XavierGr@rockbox/staff/XavierGr)
18:56:38 Join Epicanis [0] (
18:58:20 Quit shamus (Quit: chaos reigns within reflect repent and reboot order shall return)
18:58:44 Quit Thra11_ (Ping timeout: 256 seconds)
18:59:54 Quit mgottschlag (Ping timeout: 246 seconds)
19:00:15 Join shamus [0] (
19:06:42 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
19:16:05 Quit n1s (Read error: Connection timed out)
19:17:06 Join pamaury_ [0] (
19:17:25 Quit pamaury (Ping timeout: 244 seconds)
19:18:35 Join Thra11 [0] (
19:25:38 Join pretty_function [0] (~sigBART@
19:34:19 Join n1s [0] (
19:34:19 Quit n1s (Changing host)
19:34:19 Join n1s [0] (~n1s@rockbox/developer/n1s)
19:41:59 Join stoffel [0] (
19:54:45***Saving seen data "./dancer.seen"
19:55:58 Join Horscht [0] (
19:55:59 Quit Horscht (Changing host)
19:55:59 Join Horscht [0] (~Horscht@xbmc/user/horscht)
20:09:38 Quit stoffel (Remote host closed the connection)
20:28:49 Quit georgeha1k (Ping timeout: 246 seconds)
20:35:43amayer_i dont have git on my current machine is someone avaliable to make a commit?
20:35:45amayer_it would be to the default cabbiev2
20:36:09amayer_i guess to WPSLIST file
20:37:58amayer_we need to add: "show icons: on"
20:39:04 Join georgehank [0] (
20:40:21amayer_actually i think we need to add it to "" too
20:40:48amayer_its in the manual but not set by WPSLIST or
20:41:46amayer_its not a huge thing but i wont be able to get to it until this weekend
20:46:28 Join Keripo [0] (
20:49:15n1sBuschel: i can't quite convince myself that the reordering of operations you did in comb_filter can't introduce overflows
20:57:24 Join mystica555 [0] (
20:57:24 Quit mystica555 (Client Quit)
21:03:54 Join lorenzo92 [0] (
21:06:23lorenzo92kugel: what about RDS patch? I think it's fine, it works perfectly, apart the text in lebellium's theme (who, by the way, will release a new version that's not affected by the crash)
21:12:44lorenzo92*well he has done a workaround
21:13:01lorenzo92not really a solution :)
21:13:03n1sif themes can crash rockbox we have serious bugs
21:13:56lebelliumnls: FS #12759
21:13:57fs-bluebot Various RDS issues (Clip Zip, YP-R0) (bugs, unconfirmed)
21:14:08lorenzo92n1s: indeed, but I think it's due malformed strings of whatever...not really "theme engine" issues
21:14:08lebelliumthat's the truth :(
21:15:15lorenzo92n1s: i.e. memory corruption...
21:17:14n1swell, somethings' causing the memory corruption, and there seems to be usb issues related to theme as well, suggesting the themes are capable of causing it
21:40:51 Join petur [0] (~petur@rockbox/developer/petur)
21:54:48***Saving seen data "./dancer.seen"
21:59:09kugellorenzo92: i asked you to separate out the gpio changes
21:59:44lorenzo92I did that, but I did also a mistake ... doubled the patches on gerrit :S
22:00:07lorenzo92mainly just needed to restore the gpio file
22:00:10lorenzo92nothing more
22:00:14lorenzo92then it's done
22:00:33 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
22:00:38 Join pretty_f_ [0] (~sigBART@
22:00:47 Quit pretty_function (Read error: Connection reset by peer)
22:01:14kugellorenzo92: that doesnt look right
22:01:46kugel g#321 reverts g#320
22:01:48fs-bluebotGerrit review #321 at : This patch adds RDS capability to Samsung YP-R0 target. by Lorenzo Miori (changes/21/321/1)
22:01:53kugelif you have git related questions just ask
22:06:07lorenzo92indeed, sorry ...
22:12:14 Join prof_wolfff [0] (
22:12:45 Quit alexbobp (Ping timeout: 260 seconds)
22:13:13 Quit pretty_f_ (Ping timeout: 260 seconds)
22:13:16 Join alexbobp [0] (
22:40:00kugellorenzo92: also g#320 still contains the changes
22:40:02fs-bluebotGerrit review #320 at : This patch adds RDS capability to Samsung YP-R0 target. by Lorenzo Miori (changes/20/320/2)
22:50:46n1sBuschel: the opus guys didn't seem entirely confident that the operation order changes would not introduce overflows so i'd like to skip them for now, keeping just the changes skipping the calculations for 0 gains and using OPUS_COPY instead saves 7MHz on cf and 7.4MHz on pp
22:54:31 Join lebellium_ [0] (
22:54:51 Quit n1s (Quit: Ex-Chat)
22:56:18 Quit lebellium (Ping timeout: 246 seconds)
22:56:26 Nick lebellium_ is now known as lebellium (
23:05:03 Quit Horscht (Read error: Connection reset by peer)
23:05:26 Join Horscht [0] (
23:05:26 Quit Horscht (Changing host)
23:05:26 Join Horscht [0] (~Horscht@xbmc/user/horscht)
23:09:39 Quit lorenzo92 (Quit: ChatZilla 0.9.89 [Firefox 15.0.1/20120907231657])
23:11:21 Quit petur (Quit: Leaving)
23:13:55 Quit Horscht (Quit: Verlassend)
23:24:04 Part amayer_
23:26:43 Quit prof_wolfff (Ping timeout: 260 seconds)
23:35:35 Nick SpiderFromMars is now known as LittleCreature (~fearofmus@unaffiliated/fearofmusic)
23:40:04 Quit scorche (Disconnected by services)
23:40:08 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
23:42:13 Quit benedikt93 (Quit: Bye ;))
23:43:21saratogawhat is that werid "works because the cos is nearly one" bit in the Opus MDCT for?
23:44:08 Join Buschel [0] (
23:45:47Buschelsaratoga: i suspect they just do not do the multiplicstion with the cosinus term for yr and yi because it is nearly 1
23:46:02Buschelsaratoga: the sine term is also assumed to be sin(x) = x
23:46:37saratogawhat does that save them?
23:46:43saratogahalf the table entries?
23:47:05Buschellike, the real solution would be *fp++ = yr*cos(x) + yi*sin(x) -> they do *fp++ = yr + yi*x
23:47:20Buschelthey save two muls per loop
23:47:43saratogaisn't the yr already the dot product of the trig terms?
23:49:09saratogathe usual implementation is 4 multiplies per loop, but this one looks like 6 ?
23:49:44Buscheln1s: i can understan the concerns. if you expand the terms like "MULT16_32_Q15(MULT16_16_Q15(f0,g01),(x[idx0-1]+x[idx0+1]))" to "MULT16_32_Q15(MULT16_16_Q15(f0,g01), x[idx0-1]) + MULT16_32_Q15(MULT16_16_Q15(f0,g01), x[idx0+1])" i can live with for now. but it would be good to keep idx0/idx1/f0/f1 stuff.
23:50:47Buschelthis is what puzzled me as well... compared to the mdct pre-/post-rotation in faad
23:51:50saratogais it actually an optimization?
23:53:23saratogathe ffmpeg version for instance looks very similar but probably a lot faster:;a=blob;f=apps/codecs/libwma/mdct.c;h=e66be0e825e38bb120002c03af0f407de51cdffb;hb=132bc634edfb54aa857230eb6976309c8d4a2360#l74
23:53:30saratogaalthough maybe its not comparable?
23:54:09Buschelto me it looks like totally different implementations (opus vs. faad)
23:54:38 Quit eckoit (Quit: eckoit)
23:54:52***Saving seen data "./dancer.seen"
23:55:28Buscheln1s: when the comb_filter() is later switched to multiply-add such overflow cannot occure anymore
23:56:47 Quit pamaury_ (Ping timeout: 245 seconds)

Previous day | Next day