pdowling: anyone in this room alive?
xamindar: i mean, no
Tigerchen: airlied: ah, thanks.
sonne: hmmh updating to latest git, xrandr is now fully b0rken...
Tigerchen: narf, someone with access should revert the latest patch, that broke more that it fixed (imo)
udovdh: xrandr even more br0ken?
libv: Tigerchen: ok, will look into it shortly
udovdh: anybody here who knows if Matthias Hopf is back on the job?
udovdh: (or his hobby...)
udovdh: also: updated this one https://bugzilla.redhat.com/show_bug.cgi?id=434695
udovdh: so it appears to be a dual monitor issue
udovdh: no update on the two other bugs...
libv: udovdh: yeah, he is
libv: udovdh: but we had fosdem last weekend
libv: and we're all kind of broken
libv: udovdh: weren't you going to show up as well ;p
udovdh: I was there!
udovdh: saw both radeonhd related presentations
udovdh: was just a bit more than halfway towards the back
udovdh: wearing a greyish sweater
libv: udovdh: and you didn't step up to me to say hello?
libv: udovdh: i have no idea who walked in and out to be honest, so i of course wouldn't have been able to pick you out or remember faces of everybody in our devroom, but it would've been fun to talk to you ;)
libv: it was fun to finally meet ndim for instance
udovdh: oh, didn't want to disturb you too much
udovdh: that was the feeling when I was there
udovdh: because stuff (the direction of radeonhd etc) was so much more focussed, as I felt it
libv: well, on sunday i was mostly making sure that the devroom wasn't too crowded or disturbed too much
libv: so sunday was actually reasonably stress free for me :)
udovdh: sunday I was at the Klik demo
libv: on saturday i was indeed all over the place... this is how it goes with fosdem :)
udovdh: and the TangoGPS talk
udovdh: and at the keysign thingie
libv: anyway, better luck next time :)
udovdh: I have the follow up of `what the hack ` in the picture
udovdh: shiuld be again in 2009?
udovdh: WTh in 2005 was very cool
udovdh: maybe ccc before that?
libv: yeah, i think we have a fully fixed spot on fosdem from now on
udovdh: that's cool!
libv: it works out reasonably well since localhorst (who runs the opensuse room, and is in the same team at SUSE) and i can get some stuff done together... next year we'll try to synchronise some more so that this will run a bit smoother too
udovdh: smoother? what was the problem?
libv: about those other events, something like fosdem takes away a lot of precious time which should've really been used for development
libv: well... our devroom ended at 18:00, opensuse one at 17:00
udovdh: yes, time is a precious thing
libv: we have a 5-6h drive back to nurnberg (thanks to the german autobahns)
udovdh: I was surprised how fast most of the people in Belgium drive on the highway
libv: and we only thought about this difference in time somewhere in the afternoon
udovdh: typically 130+
udovdh: for me it was 1.5 hours to get home
udovdh: but the talks in this edition were very OK
libv: yeah, if the weather is good, 135 is the going speed
udovdh: I got soem insight in how the cards work, how the development is heading
udovdh: and also about the sw development
udovdh: can't ask ATI about cards doing 132x80 or so?
libv: and some people will drive at 150kmph, which often just is too much for belgian highways
udovdh: (character mode)
udovdh: and scroll fast?
udovdh: just look ahead
udovdh: and drive a diesel
libv: oh, i am a diesel person myself
udovdh: Cool! It helps, you don't have to break as often as in a petrol car
libv: but doing 150 on a belgian highway is only possible after 20:00 or so
udovdh: just let go of the pedal
udovdh: and it will slow down
udovdh: I did 150 going home in the afternoon. for much longer stretches than I would have imagined
udovdh: car won't go much faster than that...
libv: before that it is usually too dangerous because too crowded or the road is full of sunday drivers
udovdh: yeah, I know...
udovdh: that's why you should look ahead. but anything over 120 is a bonus over here
udovdh: BTW: about the radeonhd BIOS update:
udovdh: Gigabyte suggests me to use a bootable windows (windowsPE) CD
udovdh: instead of providing a decent DOS flasher for the card
udovdh: of course the winPE might work
udovdh: but it is a strange tactic
udovdh: would the card's bios also have some microcode or whatever for the engine on the card?
udovdh: or are the functions hardcoded?
udovdh: on the chip?
libv: udovdh: ASICInit sets up the engine, the ramcontroller, etc, so yes, this might change some things
udovdh: set up == uploade code to the engine and start the code there?
udovdh: my card in the dual screen box is at F2
udovdh: this one is F4
udovdh: at least there is a difference
udovdh: system-config-display uses Xinerama for dual screen
udovdh: (just let it generate a conf on the single display system)
dmb: guess the cia.vc bot i have in here isn't working
Obscene_CNN: whoa!! lots of up dates to git
egbert: Obscene_CNN: i pushed some bug fixes i had in my queue.
Obscene_CNN: I guess, 11 of them :D
egbert: i didn't want to push anything before i went to fosdem as i wasn't sure how i can react to bug reports while i was there.
Obscene_CNN: good idea
udovdh: In file included from rhd_atomwrapper.c:33:
udovdh: ./AtomBios/includes/CD_Common_Types.h:50:95: error: operator '||' has no right operand
udovdh: I get this on both machines
udovdh: small typo?
egbert: i got this patch from dragonfly bsd.
udovdh: it looks like something is missing, from the error
egbert: i thought i had tested it before pushing.
udovdh: I just did a git pull
udovdh: and make
udovdh: on both boxes
egbert: arrgh. tested wrong tree.
udovdh: but we caught it quickly!
egbert: udovdh: fixed pushed
udovdh: ok will test
rbmorse: me too.
udovdh: thanks for the quick fix
egbert: udovdh: i should have tested more carefully in the first place.
udovdh: it's not that bad when the support is this quick!
dmb: its not support, its just egbert :D
udovdh: haha, oh well..
udovdh: compare it to an issue with a commercially provided driver
egbert: the patch that was on radeonhd@ was bogus already. wonder how the new os support was tested in the first place with this patch.
egbert: udovdh: there you only get binaries. to create them the sources have to compile first :p
udovdh: yes... but any issue would take longer to fix
libv: egbert: did you check whether your VGA change still works when you null VSYNC before entering the X server?
libv: egbert: this was an issue we had some months ago
dmb: that is true, large proprietary corporations tend to take longer to incorporate fixes
dmb: thats because there is no source code
egbert: libv: why would this matter here?
libv: egbert: i am not sure, but it could influence things
egbert: libv: i was not dealing with this. if there is anything touchy there it should be commented in the code.
udovdh: Silken mouse enabled
libv: commit 1a1d110f49
libv: it probably still is solid, but it's worth testing for this
libv: egbert: i bring it up exactly because you weren't dealing with it :)
libv: at the time
libv: and that you don't know about this because of that
egbert: libv: i don't understand the order changes you did here well.
egbert: these things should be commented in more detail.
egbert: libv: and i should have done this, too.
libv: well, let's see how this changes things, i do think i checked reordering as a possible solution to the issue, but it's been too long for me to remember this distinctly
egbert: libv: your change involved moving restoring crtc past vga restore.
egbert: libv: so it looks like this did have something to do with it.
egbert: if this is really an issue fixing this won't be so easy.
libv: just null the VSYNC in console and test
libv: it's rather easy to reproduce
egbert: libv: not atm. i've got to make some progress on other things first.
rbmorse: I may be self-delusional, but did the latest commit boost performance a bit. Things seem snappier.
libv: rbmorse: hrm... wouldn't think so
rbmorse: just a feeling. Not sure how one would tell. Me likee.
libv: rbmorse: let's hope it remains as fast then :)
libv: because if not, something else is going on
the-me: which acceleration method should I use with my mobile X1250pro?
Obscene_CNN: XAA I think
libv: XAA is the default anyway on r5xx and rs6xx
Obscene_CNN: Would x11perf be a better benchmark than gtkperf at testing XAA improvements?
libv: Obscene_CNN: i guess so, as it probably was used to optimise xaa
Obscene_CNN: I just can't get gtkperf to give very consistent results.
rx__: carl worth uses several benchmarking utilities for comparing exa, xaa
rx__: you should check out his site
beerockxs: could someone take a look at my xorg.conf and tell me why static randr multimonitor is not working?
beerockxs: I recently reinstalled my system, and I think I used the same settings
Tigerchen: beerockxs: very simple
Tigerchen: beerockxs: http://wiki.x.org/wiki/radeonhd#head-99961806d359be473c98eb244c8eab9101523abf
beerockxs: Tigerchen: Thanks
beerockxs_: Tigerchen: that didn't work.
beerockxs_: Tigerchen: the second display still clones the first
Tigerchen: beerockxs_: have you changed the name of your monitor section accordingly?
beerockxs_: Tigerchen: In the device section, I now have "Option "monitor-DVI-I_1/digital" "Asus""
beerockxs_: and also one for the other monitor
beerockxs_: the monitor ids are "Asus" and "BenQ", just like in the device section
Tigerchen: could you nopaste your xorg.log?
Tigerchen: brb in 5 min, gotta bring the dog out
beerockxs_: np, thanks for your help :)
Tigerchen: beerockxs_: at first look it should just work. can you manually setup the monitors the way you want?
Tigerchen: beerockxs_: i think i got it. (perhaps)
Tigerchen: (II) RADEONHD(0): Using 1680x1680 Framebuffer with 1728 pitch
Tigerchen: beerockxs_: maybe you have to set a virtual-line
beerockxs_: Tigerchen: doing xrandr --output DVI-I_2/digital --right-of DVI-I_1/digital gives me xrandr: screen cannot be larger than 1680x1680 (desired size 2960x1050)
Tigerchen: that's it
beerockxs_: Tigerchen: I thought that was for when setting it dynamically via xrandr, and not for the static setup?
Tigerchen: you need a virtual fb that's large enough to fit both monitors
Tigerchen: doesn't hurt to try it does it?
beerockxs_: Tigerchen: that's true, only the wiki says in the example that it's actually harmful to set a virtual size when using the static setup
Tigerchen: hmm, it's my only guess that it could be this, otherwise i have no idea. and that way you can at least configure it the way you want after x-start
beerockxs_: Tigerchen: with that virtual line, i can set it manually via the xrandr call, but it doesn't get set automatically
Tigerchen: beerockxs_: hmm, i have no further idea, perhaps someone on the mailinglist can help, but as a first workaround I'd keep the virtual line and put the xrandr-call in an xinitrc or something
beerockxs_: Tigerchen: yeah, that sounds good
beerockxs_: Tigerchen: I might check if it works with an earlier driver revision or sth. like that, because it did work befor my reinstall
beerockxs_: i don' think that me now being in 64 bit should affect that
Tigerchen: i don't think so but it might be
^^MAg^^: Hi, what is the current status of DRI under radeonhd? I've got Mobility FireGL V5200
agd5f: ^^MAg^^: no 3D yet
^^MAg^^: ah ok, so no Xv for now
airlied: ^^MAg^^: radeon git master has Xv now
jadamcze: The link to Keith's Packard's talk at FOSDEM on radeonhd.org gives me a 404 (http://www.radeonhd.org/fosdem-2008/fosdem_2008_intel.ogg)
jadamcze: linked from http://www.radeonhd.org/?page=fosdem_videos
airlied: jadamcze: I don't t hink it got uploaded yet
jadamcze: airlied: I had no problem with other vids on that page. I should just try again later?
airlied: jadamcze: yes as I said it hasn't been uploaded yet and I think its in transit over the atlantic in a plane now
jadamcze: That's some expensive international bandwidth.
aneas: but its easier than mailing a hexdump printout
gusta1: the vids has been working for over a day, and keth's has not been available for that long (I remember checking yesterday). I don't think it's a filename error or something.
gusta1: but sure, wait, if you find that fun ;)
jadamcze: gusta1: Fun or not, I'm not aware of any other options.
gusta1: talk to the owner of the site or wait a couple of years until they're available on fosdem's own web page
gusta1: i.e., keep reloading this monthly: http://www.fosdem.org/2008/media/video
airlied: as I said the owner is in the air over the atlantic somewhere.
airlied: so when he lands and gets setup he'll finish the uploads..
gusta1: something's terribly wrong with most of these conferences since the videos are often extremely hard to find, ususally on some third-party blog-referenced-only web page.
gusta1: and often only in the world's worst format ever - ogg theora
gusta1: in most countries there is nothing wrong with mpeg4 regarding patent issues etc, so why that couldn't be the default is mystery to me..
airlied: gusta1: fosdem doesn't do the recordings..
airlied: gusta1: it was just an audience member who had the equipment
airlied: gusta1: most confs don't video at all, LCA is the only one I've seen do it successfully.
gusta1: that equals (to me) that "fosdem doesn't care about what people outside the meeting wants - that they want to see what was said"
airlied: gusta1: the whole point of confs is for people who go to them..
jadamcze: gusta1: If people want to see what is said, they should attend the conference :p
gusta1: uh... right...
airlied: conferences are not for giving info to the Internet, thats what irc/web/mail is for
aneas: not everyone can fly to brussel to attend some talks ;)
gusta1: i, and some of my friends (which means thenths of thousands world-wide at least) watches many of these videos
gusta1: exactly, aneas
dli: gusta1, what's wrong with ogg? my mplayer plays http://www.fosdem.org/2008/media/video
airlied: but you shouldn't expect a volunteer run conf to go through all the extra effort
gusta1: but no one is talking to us, I guess. and since there's no other talk being done, what we can expect is releases and a git source... yay.
airlied: and videoing is a lot of extra effort..
airlied: gusta1: its a driver, you getr sources and releases, phoronix occaisonally does updates..
gusta1: dli omg, the quality is the worst ever. it's not seekable, the timing is way fucked (mplayer usually says a talk is 120 hours or so)
airlied: I don't think you'd learn much more from the talk than you would just hanging around here..
gusta1: i'm relying on phoronix, ok. well that's great i guess..
dli: airlied, if the mplayer window is not on top, the mplayer window messed up. an Xv bug from r500?
jadamcze: IRC is not good at charts and pretty demos ;)
airlied: dli: possibly I'll have to look into it :)
airlied: jadamcze: what pretty demos, hey I can turn on a monitor looks the same for all cards ;)
dli: airlied, I guess I couldn't make screenshot of Xv output
airlied: dli: should be able to
gusta1: what i'm saying is, there's lots of corporate backing. most of you people who do things seems to either be working (for intel, ati, suse etc etc) or a university... usually, the companies pays for these such events, no?
gusta1: but in open source, you're/we're still relying on volunteers? that's just sad.
airlied: gusta1: FOSDEM isn't paid for by companies..
airlied: it might pay to send people, but all the planning is done by volunteersw..
airlied: I don't see the point of you argument though..
gusta1: you don't need to
airlied: large companies don't organise conferences..
airlied: RH rarely funds my conf attendances.
gusta1: wow ok.
airlied: all the big Linux developer confs are volunteer run..
airlied: they get sponsorship money in a lot of cases to cover the costs.
timofonic: There will be subtitles for FOSDEM X.org 2008 Videos? Please, my spoken English is totally unuseful and probably I'm not the alone
gusta1: and that's just great, you think. ok.
airlied: but nobody gets paid to organise them..
gusta1: perfect perhaps
airlied: gusta1: yes it keeps the confs from having corporates taking over
gusta1: taking over yeah...
gusta1: what does that mean, anyway?
airlied: gusta1: OLS does make a bit of money and does pay the organisers company alright..
gusta1: good quality mpeg4 videos? ;)
airlied: gusta1: it means that Intel runs a conf, you don't hear anything abuot AMD.
airlied: or AMD runs it ans nobody presents anything on nouvueau
gusta1: I see no problem with RH, intel, ati or any other company or group of companies paying for an event. it would be great to inform people what's going on, and to give them good seekable watchabe videos.
gusta1: right now, I can't see the point with the confs. do any of you "learn" anything that you couldn't learn here on irc? I'm just following your argument that _I_ wouldn't...
airlied: everything I've seen at confs I'mve mostly known about on irc well before hand..
gusta1: I guess what's most important on those meetings are actually the social parts, getting to know each other and talking Real business over a beer.
airlied: gusta1: yes.. most people don't go to confs to watch talks..
airlied: or give talks..
airlied: they are just a means to get funding to go to the confs.
gusta1: so why have them? why the fuzz? why are lots of people interested in what is being said, you might ask.
airlied: because they provide a focal point around which people can gather..
gusta1: well, because we like to watch these talks, because most of us don't work with the projects, and likes to see some short abstract for an hour or so. that's why so many people likes good videos.. but this is all weird to you, or?
gusta1: I think you're just absolutely insanely wrong.
sickPUPs: hi, whats the best videocard for HD editing?
sickPUPs: sorry for the newb question
airlied: gusta1: no I understand you want to watch videos, I'm just saying the conf presentations aren't the place for that..
gusta1: you only, only, look at it from you dev's point of view. but there are very interested users out here..
airlied: it would be better if people would do specific videos for the Internet, just talking to a camera..
gusta1: airlied: I haven't found any other great place, but I'll keep looking
aneas: gusta1, cool down
airlied: when you have an audience its a pain to have to remember the recording as well..
gusta1: that's the whole point... *sigh*
sickPUPs: whats this channel about?
airlied: gusta1: are you going to fix the presenters?
gusta1: i mean. you say that you like the confs, no? they have a purpose, and they won't dissapear. now, you say informing people what's going on is better done by someone talking into a camera? my point is, mix these together.
airlied: gusta1: yes I assume you've presented at lots of confs with camera?
gusta1: the vids I've seen have been great, although with really crap quality and non-seekable, and hard to find etc... but I really don't think some dude talking into a camera alone would be better. I've seen those vids to, and they're boring as ...
airlied: sonne: radeonhd
airlied: sickPUPs: radeonhd
gusta1: airlied. i'm not blaming you, or anyone else who was at the conf.
airlied: gusta1: the things is large companines with money aren't going to sovle the problem..
gusta1: I just say, I think something's wrong with the confs being held (fosdem, lca etc)
airlied: gusta1: LCA has proven you can do it nearly properly and even then they get so many problems with the recording quality
sickPUPs: im looking for a channel to do with al things film/video production or crit
airlied: sickPUPs: seen topic I doubt that is here.
timofonic: Going to sleep, bye
gusta1: airlied: have those who attend these meetings, working for these companies tried to get corporate backup (for 1 person filming, and encoding to decent codec)?
airlied: gusta1: LCA does it with a team of about 30 I think.
airlied: gusta1: one person can't do it as has been proven many times..
airlied: gusta1: most of the confs use the sponsorship money for other things which are more important..
airlied: like givign the people who go to the conf things and organising activities.
gusta1: I can't believe it's 30 people when the vids are not just mediocre at best, and all in ogg format...
gusta1: I take back what I said previously, I must have been completely wrong about filming. I now realise it's a very very hard thing to do.
airlied: gusta1: they have two people per room doing the recordings, one camera, one sound person
airlied: so they have 5 rooms running at once, so 10 ppl..
airlied: then they have to give those people breaks, so they need around 20 to cover that.
airlied: then they have another group of people taking the recordings, encoding them and uploading them
gusta1: heh, i gotta say, i'm surprised.. really.
airlied: also LCA last year took like a month post-conf to get a lot of videso out..
airlied: this year they managed to get them out during the conf..
airlied: maybe next year who knows..
marcheu: just getting speakers amplified is already a big business... in confs I've been there was one or two guys per room for just that...
gusta1: live? ;)
airlied: gusta1: hmm I don't think Tasmania has the bandwidth..
airlied: marcheu: LCA had the problem of amping speakers and recording them at the same time without feedback.
airlied: marcheu: hence why my talk sound levels are so bad..
gusta1: can't be more dejected when, in 2008, I get web server directory listing of ogg files.. but I'll wait for the conf which has .torrent's to xvid's.. it's like all the work, in vain..
marcheu: uh ? ogg doesn't work ?
gusta1: I downloaded a dozen of the vids the other day, and started to watch, seek and bam! back to position zero in mplayer..
marcheu: I don't think you'd encode a free software conf in a non-free format :)
sickPUPs: what is the best codec for HD?
gusta1: it could be my mplayer, I don't know. but I have _never_ managed to watch them properly. they're never seekable and always bad qual
TGEN: #define best
gusta1: marcheu: what's non-free about xvid?
gusta1: and don't remember, you're french.
cbecht: So, I started X with the radeonhd driver, but forgot to remove the fglrx kernel module first. I got a warning in the log file, a black screen that I had to reboot out of, and a card that no longer POSTs. Oddly timed hardware failure, or have I managed to do damage from software?
gusta1: don't forget*
marcheu: gusta1: what does that have to do with being french ?
TGEN: cbecht: did you let the card bleed capacitors and any data from RAM?
gusta1: well, some people say xvid isn't "free" in countries like the US, but in europe (most of the countries anyway) it is.
TGEN: cbecht: unhook power, leave it off for a couple of minutes, then boot it again
gusta1: i'm a bit tired of people saying only theora is "free" and xvid is not and so on...
cbecht: TGEN: You mean leave it unpowered for a suitably long period of time before rebooting? Yes.
cbecht: TGEN: I'll try again though. Nothing to lose.
TGEN: cbecht: where unpowered also means no standby power
cbecht: Incidentally, I can sort of get it to work by booting off of a PCI video card and then starting X. But there is all sorts of weirdness.
TGEN: (unhook the power cord :))
cbecht: TGEN: Where unpowered meant physically removed from the machine because otherwise I couldn't boot. =/
cbecht: For the record, weirdness includes only being able to start X once. If I exit, then restart the server it hangs. Also, any resolution over 1024x768 includes unusual visual distortion. The screen is divided into vertical bands maybe 10 pixelss wide that repeat. So for example, you see shadow copies of the mouse cursor.
TGEN: cbecht: that smells of VRAM death
cbecht: Also, starting X with fglrx just hangs. Only radeonhd works even a little.
cbecht: TGEN: Hmm. Yeah, that would make sense.
cbecht: TGEN: Thoughts on testing that hypothesis? maybe change the color depth and see if it changes the resolutions that work?
TGEN: cbecht: boot Windows? :)
cbecht: TGEN: It doesn't POST. =/
cbecht: TGEN: Well, I suppose I could try to boot windows from the PCI card but then I'd have to jump through multi-monitor hoops.
cbecht: TGEN: Also, I don't have a windows install handy.
TGEN: still under warranty?
cbecht: TGEN: Sadly, no. Or at least, not that I can tell. I got it from newegg, and their 30-day return period has passed. I can't find any information on Gigabyte's site about warranties. I would expect there would be something though.
cbecht: TGEN: Guess I'll have to rummage around and see if I can find a registration card or something...
TGEN: cbecht: hmm, at least here in Europe 2 years are mandatory
cbecht: TGEN: Yeah, I don't think we (USA) have particularly good consumer protection laws in that regard, but I guess I'm stuck looking into it.
cbecht: TGEN: I had managed to necromancy this thing up to 1024x768, so I was hoping maybe I could fix it the rest of the way somehow. Earlier efforts included attempting to flash the video BIOS.
cbecht: So, this is suspicious. I can do 1024x768 24bpp, but I can't do 1280x1024 8bpp.
cbecht: If it were an issue with VRAM, I'd expect that means I have X MB of vram before I hit the b0rken bits.
cbecht: But that doesn't seem to be the case. =/
aneas: my xchat windows has rarely been so full ... start using 4 letter lines like normal people on irc do ^^
sickPUPs: does anybody like this monitor??.......http://accessories.euro.dell.com/sna/productdetail.aspx?c=uk&l=en&s=dhs&cs=ukdhs1&sku=142660
cbecht: Ok, so after testing 1280x1024 @8bpp, I'm inclined to believe it's something other than vram.
cbecht: There is an error message at one point about not being able to read the full rom, but google turned that up in a few other places and it seems mostly harmless.
cbecht: Incidentally, the wiki says that using the radeonhd driver after the fglrx driver results in "shimmery" video. Something about the line buffer not being reset or cleaned up or something properly.
cbecht: That sounds a bit like what I am seeing, and this did start when I used the radeonhd driver with the fglrx kernel module loaded. And on reboot the card didn't initialize. So the card has not successfully "rebooted" since then. Maybe it's just got a wedged line buffer?
cbecht: I really wish I knew what a line buffer was right about now.