--- Log opened Сбт Авг 26 00:00:23 2006
00:07 -!- Myrizio_ is now known as Myrizio
00:10 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has joined #nouveau
00:31 -!- pmdata [i=patrice@ANantes-154-1-28-224.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
00:32 -!- tibbs|h is now known as tibbs
00:56 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
00:59 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
01:01 -!- KoalaBR [n=KoalaBR@port-83-236-14-126.dynamic.qsc.de] has joined #nouveau
01:01 < KoalaBR> Hi
01:02 < stillunknown> hello Bear
01:03 < KoalaBR> marcheu: Just a short question. Do we know, how a display is setup. I mean which regs get what values, which commands are influenced?
01:03 < KoalaBR> I just started on that, but it would be wasted time, if the algorithm is already known...
01:10 < marcheu> KoalaBR: modesetting ? the "DDX does that already, and airlied is working on that
01:11 < KoalaBR> marcheu: Sorry, display_lists_ it should have read
01:11 < KoalaBR> It's late for me :)
01:12 < marcheu> ah, well it depends on the card
01:12 < marcheu> and on the size of the display list
01:12 < marcheu> nv10 does it with indexed arrays
01:12 < marcheu> nv20 too IIRC
01:13 < marcheu> nv3+0+, I don't know
01:13 < KoalaBR> Yes, NV40 in my case
01:13 < KoalaBR> Well I am just beginning. But : 
01:14 < KoalaBR> If there are less then 30 Triangles, then the display list is encoded directly, no difference to the same commands coded without a DL
01:15 < stillunknown> <marcheu> nv3+Â0+, I don't know <-- how did you get those strange characters?
01:16 < KoalaBR> Otherwise, the vertexes are made unique (all doubles are removed) and inserted from reg 0x0c to 0x0c + (unique *3)
01:17 < KoalaBR> after that, the regs index the vertexlist, 2 index in one register (low and high 16 bit)
01:17 < KoalaBR> command 1824 describes the size of the display list, which may be "fragmented"
01:18 < KoalaBR> How fragmentation is done, I'm still researching. therefore my question :)
01:19 < KoalaBR> marcheu: So, continue or already known stuff?
01:20 < KoalaBR> stillunknown: Hopefully I can start testing ddx tomorrow. Hope you manage to write your "howto compile and test", I would try it out :)
01:20 < KoalaBR> I would have done it myself, but you beat me to it 
01:21 < stillunknown> most my attempts have been adapting the standard builds in gentoo, so i don't know everything
01:21 < stillunknown> (gentoo has a huge so-called eclass which deals with pretty much everything about modular-x)
01:22 < stillunknown> but i'll make some regular instructions too (soon hopefully)
01:23 < KoalaBR> I use gentoo too :)
01:23 < KoalaBR> N
01:24 < stillunknown> http://nouveau.freedesktop.org/wiki/GentooEbuilds
01:24 < KoalaBR> Just need a little help on how to use my own ebuilds (need to setup an overlay?)
01:24 < stillunknown> yes, you need an overlay
01:25 < KoalaBR> Ok, I think I have read a doc on how to set one up
01:25 < marcheu> KoalaBR: I never tried display lists myself, but does renouveau show unknown commands ?
01:25 < KoalaBR> Yes 1824 for example
01:25 < marcheu> ok, I guess that's not done then :)
01:26 < KoalaBR> 2 or 3 IIRC - deal :)
01:27 < KoalaBR> Will try to help you in the future with driver development too, but I'm a newbie in this field 
01:27 < stillunknown> you know more than me :-)
01:28 < KoalaBR> stillunknown: In what regard? 
01:28 < stillunknown> driver development
01:29 < KoalaBR> No, I just tried to find out pattern of renouveaus output and succeeded a few times
01:29 < KoalaBR> but I never wrote anything for mesa, kernel or X11 so .,..
01:30 < KoalaBR> But you can do ebuilds ...
01:30 < KoalaBR> I can't :)
01:30 < stillunknown> ebuilds are easy, just bash
01:30 < phh> Oo
01:30 < phh> it's python
01:30 < KoalaBR> And you bravely went were no man has gone before and installed a driver known to be a crash risk :)
01:31 < marcheu> hey, that driver doesn't crash
01:31 < stillunknown> i'm a bit mad
01:31 < marcheu> except for me :)
01:31 < stillunknown> it just seems so familiar and the only thing familiar is bash for me
01:32 < KoalaBR> I spoke only about a risk, not of a fact (try to weasel myself out of the problem)
01:33 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has left #nouveau []
01:33 < KoalaBR> Ok, I will report back tomorrow, but for now, it's bed time
01:33 < cptn> phh: it is a bash script; only portage is written in python, ebuilds are bash scipts
01:33 < KoalaBR> good night
01:33 < phh> cptn, hum...
01:33 -!- KoalaBR [n=KoalaBR@port-83-236-14-126.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"]
01:33 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
01:33 < phh> i thaught that emerge was technology full ..
01:34 < stillunknown> emerge is pretty good
01:34 < nael> phh: fmerge is pretty more powerfull :)
01:34 < phh> nael, chhhhuuuuutttttt
01:35 < stillunknown> fmerge looks like some kind of diff
01:36 < nael> doh ! fmerge exists ...
02:00 -!- hanno [i=TR6q3O2a@nat-wh-1.rz.uni-karlsruhe.de] has joined #nouveau
02:11 < hanno> is there anything already working? I have some nvidia-cards I can do tests on
02:11 < hanno> (one nv60 and one very old, I think tnt2 was it)
02:12 < phh> hanno, reverse engineering tools are working
02:12 < marcheu> the 2D driver has working EXA for some people
02:12 < marcheu> btw nv60 ?
02:12 < phh> yeah i don't know nv60 ...
02:12 < hanno> nv6?
02:12 < hanno> something with a 6 in it
02:12 < phh> nv6 is really old too :p
02:12 < marcheu> nv6 is tnt2 
02:12 < hanno> little moment ;-)
02:12 < marcheu> tnt2m64
02:13 < hanno> 05:00.0 VGA compatible controller: nVidia Corporation NV43 [GeForce 6600 GT] (rev a2)
02:13 < hanno> nv43 ;-)
02:13 < phh> Need help /o\
02:13 < hanno> no "runs some simple glvertex commands" yet?
02:14 < marcheu> nope, we have to get the 2D driver into a workable state
02:14 < marcheu> then we can move on to 3D
02:14 < phh> i have a segfault in renouevua
02:14 < phh> renouveau*
02:14 < phh> and looking more precisly it looks....strange
02:15 < phh> just look!
02:15 < phh> huym
02:15 < phh> pastebin
02:16 < phh> marcheu, http://pastebin.ca/149501
02:16 < phh> look at addresses, renouveau says 0xadc76000, but gdb says  0xadb76000 .........
02:20 -!- hanno [i=TR6q3O2a@nat-wh-1.rz.uni-karlsruhe.de] has quit [Remote closed the connection]
02:25 < phh> ...;;;
02:25 < phh> forget
02:25 < phh> i need sleeping
02:25 < phh> so why
02:30 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Remote closed the connection]
02:30 < dagb> marcheu: looking at http://nouveau.freedesktop.org/wiki/RequiredFunctionality, there is certainly progress. Do you have a vague idea about how long it is going to take to get a workable 2D driver? and a 3d driver?
02:31 < phh> dagb, 2D works ..
02:31 < dagb> January 07? July 07?
02:31 < phh> (it's based on nv driver which already works :)
02:31 < dagb> phh: 00:19 < marcheu> nope, we have to get the 2D driver into a workable state
02:31 < phh> but... it works
02:32 < marcheu> workable for 3D
02:32 < marcheu> it's already workable for 2D
02:32 < phh> marcheu, ho ok
02:32 < dagb> ah. context is everything.
02:32 < marcheu> yeah
02:33 < dagb> so.... a non-commital guesstimate taken from medium-thin air: how long?
02:35 < marcheu> hmm I don't really know, but yeah around 1 year probably
02:37 < dagb> what would/could speed up the process, apart from documentation? people? skills? money? hardware? publicity?
02:38 < dagb> or is this one of the problems where it takes a certain amount of time no matter what resources you throw at it?
02:38 < phh> peolpe/skills ?
02:38 < phh> people*
02:38 < marcheu> yeah, people with skills
02:38 < marcheu> I'm sitting on a pile of cards, cards are cheap (ebay has lots of them)
02:39 < marcheu> but people who can write driver, these are a rare breed
02:39 < dagb> I have that impression, yes.
02:39 -!- hanno [i=lQmNywaF@nat-wh-1.rz.uni-karlsruhe.de] has joined #nouveau
02:40 < dagb> maybe you could attract potentially skilled people if you/someone wrote an article for lwn.net or some other geeky publication?
02:41 < dagb> with a really low-barrier level to get started.
02:41 < dagb> I don't know. Just rambling.
02:41 < marcheu> the barrier is not low, that's a fact
02:42 < hanno> marcheu, you could definitely need more marketing
02:42 < dagb> any realistic way to lower it?
02:42 < marcheu> you can always attract more people, but I think all those who are able to handle graphics driver already know about us
02:42 < marcheu> and that's the point really
02:43 < marcheu> the result of wider communication would be us spending all our time explaining what we are doing to a number of newbies
02:43 < dagb> you know, to find and develop the really skilled people, you need a certain volume of people.
02:44 < marcheu> which ends up being counterproductive
02:44 < dagb> yes. there is a cross-over point somewhere.
02:46 < marcheu> right now, I already spend a non-neglibigle amount of time explaining what happens/how to help to newcomers
02:46 < dagb> do you document it everything in the wiki?
02:47 < hanno> you should put up an faq and a news-section
02:47 < hanno> and a "state of things"
02:47 < dagb> there is a "state of things"
02:47 < hanno> okay
02:47 < hanno> but the rest
02:47 < hanno> faq with "is there anything I can try out?" "what can I, no/rare programming knowledge, do to help?"
02:48 < hanno> if the answer is "no", it's okay, but then you can just say "look at the faq" if someone asks
02:48 < marcheu> you see
02:48 < marcheu> I have to work from time to time
02:49 < marcheu> so, if you want some up to date doc/state of things/whatever
02:49 < dagb> is it possible to automate more of the RE process?
02:49 < hanno> yeah, but if you say you spend a big amount of time explaining people this over and over again
02:49 < hanno> it's probably better to put it online
02:49 < marcheu> I'm afraid the best solution is to sit here, look at what happens, digest and publish the info yourself
02:49 < hanno> I can think of doing that
02:50 < hanno> it's a wiki, so it should be doable ;-)
02:50 < marcheu> dagb: it's automated as far as we could go. we can't really have a machine figuring the semantics
02:55 < phh> marcheu, let_it_snow() function looks.... strange :)
02:55 < phh> what should it do ? 
02:55 < phh> (don't want to try :D)
02:56 < marcheu> well, the whole visible screen is mapped into every opengl process with the proprietary driver
02:57 < phh> "nice"
02:57 < marcheu> that function just draws white pixels on that area :)
02:57 < phh> marcheu, /dev/nvidia0 acts as /dev/mem would do, but accessible to users and limited to graphic card so ?
02:58 < marcheu> yeah, it's mostly it
02:59 < phh> nael, tien tu vois.
02:59 < phh> marcheu, with agpgart, sharing ressources is in kernel side, no longer in libGL i hope ?
03:01 -!- phh [n=phh@moi.phhusson.info] has quit ["Quitte"]
03:10 -!- `Duke` [n=gnu@86.72.170.74] has quit ["Segmentation fault"]
03:16 < stillunknown> hanno: getting more info onto the wiki from a user perspective is already happening, but the people who write drivers should focus on that, not "waste" time on writing wiki stuff
03:17 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit [Remote closed the connection]
03:19 < stillunknown> hanno: you forgot that the project is in an early stage, a day ago someone told me i was the first tester of the 2d driver, just to give an example
03:20 < stillunknown> the key is people who write stuff on the wiki if they see it in irc
03:21 -!- Myrizio [n=Myrizio@host27-102.pool80104.interbusiness.it] has quit ["Goodbye Ruby Tuesday (Perl Friday)"]
03:22 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
03:27 -!- Duke` [n=gnu@ANantes-251-1-99-188.w86-203.abo.wanadoo.fr] has quit [Read error: 110 (Connection timed out)]
03:28 -!- Duke` [n=gnu@ANantes-251-1-83-142.w86-203.abo.wanadoo.fr] has joined #nouveau
03:41 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
03:46 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
03:54 -!- shenki_ [n=shenki@121.44.69.112] has quit [Read error: 60 (Operation timed out)]
05:21 -!- hanno [i=lQmNywaF@nat-wh-1.rz.uni-karlsruhe.de] has quit ["Verlassend"]
05:46 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
05:58 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has quit [Read error: 104 (Connection reset by peer)]
07:30 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
07:32 -!- shenki_ [n=shenki@121.44.69.112] has joined #nouveau
08:40 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has joined #nouveau
08:45 -!- skrull [i=skr@200.96.100.169] has quit [Read error: 110 (Connection timed out)]
08:52 -!- shenki_ [n=shenki@121.44.69.112] has quit [Read error: 110 (Connection timed out)]
09:08 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has quit [Read error: 104 (Connection reset by peer)]
09:10 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
09:20 -!- Gentle [n=DasTier@HSI-KBW-085-216-054-105.hsi.kabelbw.de] has joined #nouveau
09:31 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has joined #nouveau
10:48 -!- pef_aw is now known as pef
10:53 < pef> I'm beginning the work on texturing for nv20, hope nobody is currently working on it :) (features matrix says no)
11:05 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has quit [Read error: 104 (Connection reset by peer)]
11:19 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has joined #nouveau
12:28 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
12:48 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
12:53 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
12:58 -!- shenki [n=shenki@121.44.69.112] has quit [Read error: 60 (Operation timed out)]
13:09 -!- shenki [n=shenki@121.44.75.187] has joined #nouveau
13:14 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
13:18 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
13:18 < stillunknown> my Xorg.0.log sais RandR is loaded, while i have not specified that in my xorg.conf
13:19 < stillunknown> does this mean i automaticly loose all 2d acceleration?
13:23 < stillunknown> brb
13:23 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
13:25 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
13:26 < stillunknown> the extension wasn't loaded, because when i forced my xorg stopped working ok :-)
13:27 -!- shenki [n=shenki@121.44.75.187] has quit [Read error: 60 (Operation timed out)]
13:28 -!- dagb [i=dagb@97.80-202-99.nextgentel.com] has quit [Read error: 104 (Connection reset by peer)]
13:38 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
13:42 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has quit [Read error: 113 (No route to host)]
14:05 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has joined #nouveau
14:26 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
14:29 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
14:34 -!- nael [n=nael@ras75-1-81-57-62-96.fbx.proxad.net] has quit [Remote closed the connection]
14:47 -!- stringfellow [n=stringfe@a80-100-96-220.adsl.xs4all.nl] has joined #nouveau
14:48 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
15:00 -!- fatal__ [i=gem@gamezone.fjortis.info] has joined #nouveau
15:39 -!- shenki [n=shenki@121.44.130.22] has joined #nouveau
15:41 -!- K [i=hazel@1.pool85-57-128.dynamic.uni2.es] has joined #nouveau
15:49 -!- shenki_ [n=shenki@121.44.130.22] has joined #nouveau
15:53 -!- shenki [n=shenki@121.44.130.22] has quit ["Leaving"]
15:55 -!- shenki [n=shenki@121.44.130.22] has joined #nouveau
17:13 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit [Remote closed the connection]
17:13 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
17:31 -!- KoalaBR [n=KoalaBR@port-83-236-14-217.dynamic.qsc.de] has joined #nouveau
17:36 < KoalaBR> marcheu: I downloaded both drm (CVS) and DDX (git). Compiling drm works - but trying to compile ddx I get In file included from nv_bios.c:24:
17:36 < KoalaBR> nv_include.h:76:25: nouveau_drm.h: No such file or directory#
17:36 < darktama> copy nouveau_drm.h to xf86-video-nouveau/src/
17:37 < darktama> the "make install" target for libdrm should install it aswell
17:39 < KoalaBR> Thanks.
17:41 < KoalaBR> BTW; I have written a small summary what happens within this project: http://www.ping.de/sites/koala/companion.txt   (If consensus is that it is ok, I will add it to the Wiki with correct Markup
17:48 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
17:50 < stillunknown> KoalaBR: the proper way is to install libdrm, before the ddx
17:52 -!- shenki [n=shenki@121.44.130.22] has quit [Read error: 104 (Connection reset by peer)]
17:52 -!- shenki [n=shenki@121.44.130.22] has joined #nouveau
17:52 < stillunknown> KoalaBR: oh darktama already said that
17:52 < KoalaBR> stillunknown. I hesitate because I don't want to mess up my gentoo install
17:53 < KoalaBR> You know, i need my working xorg 7.0
17:53 < stillunknown> i think XOrg 7.1 might be needed
17:53 < KoalaBR> The new driver is not in ~x86 I fear
17:53 < stillunknown> i just want my core utils stable, the rest i can fix if it breaks :-)
17:54 < stillunknown> the new driver is there, just not enough testing yet, so the xorg-server still blocks it
17:54 < KoalaBR> Well, I will wait for it
17:55 -!- Myrizio [n=Myrizio@host234-102.pool80104.interbusiness.it] has joined #nouveau
17:56 < stillunknown> xorg breakage doesn't scare me
17:56 < stillunknown> breaking binutils or gcc is bad
17:56 -!- shenki [n=shenki@121.44.130.22] has quit [Read error: 104 (Connection reset by peer)]
17:57 -!- shenki [n=shenki@121.44.130.22] has joined #nouveau
17:59 < KoalaBR> stillunknown: I would like some opinions on whether this writeup would be useful for the homepage: http://www.ping.de/sites/koala/companion.txt
17:59 < KoalaBR> Markup will be added, of course
18:00 < stillunknown> (reading)
18:02 < stillunknown> you need to say something about the combination of DDX, DRI and DRM (why is a drm needed for example?)
18:02 < stillunknown> also explain what XAA and EXA is, because that doesn't mean anything to most people
18:02 < stillunknown> (i didn't know anything about it until 2 days ago)
18:03 < KoalaBR> Ok, will take that into account
18:06 < stillunknown> KoalaBR: once xorg 7.1 becomes stable, you should be able to install the ebuilds, try them, if it goes wrong, then remove the -* keywords and install the normal stuff again
18:06 < KoalaBR> Thanks
18:06 < stillunknown> doing things portage way should leave no mess
18:07 < KoalaBR> True
18:08 < stillunknown> a person named pef started on texture related things for nv20, do you know if anyone is working on that?
18:10 < KoalaBR> No, darktama did some initial work on NV40 but stalled
18:15 < KoalaBR> I need to go, will be back later
18:16 -!- KoalaBR [n=KoalaBR@port-83-236-14-217.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"]
18:53 < zoeloelip> :q
19:00 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
19:14 -!- shenki_ [n=shenki@121.44.130.22] has quit ["http://xkcd.com/comics/fourier.jpg"]
20:51 -!- arachnist [i=arachnis@hive.bsic.pl] has joined #nouveau
20:51 < arachnist> hi
20:51 < stillunknown> boo
20:52 < arachnist> hmm
20:52 < stillunknown> my poor sense of humour :-)
20:52 < arachnist> how to make a "card dump"? i could provide one from a gf7900gt (pci-e) if it'd be possible ;>
20:53 < phh> download renouveau CVS first
20:53 < phh> stillunknown, or it has been sent to git ?
20:53 < stillunknown> only the ddx is in git
20:54 < phh> k
20:54 < phh> is there a particular reason or it's only to try git ?
20:55 < stillunknown> probably easier to branch and merge
20:55 < stillunknown> drm already has the original module in git
20:55 < stillunknown> eventually it will be merged back in
20:55 < phh> ah
20:57 < stillunknown> arachnist: card dumps are not click and run
20:57 < stillunknown> you need to get it from cvs, change main.c to choose the tests and compile
20:58 < stillunknown> renouveau docs: http://www.ping.de/sites/koala/
20:59 < arachnist> nice
21:02 < stillunknown> http://nouveau.freedesktop.org/wiki/
21:04 < arachnist> uncomment this one in main.c at line 137, right? //   test_startup();
21:05 < arachnist> or should i say, all of the test_*, right?
21:05 < phh> for all test*
21:10 < stillunknown> there is also an option somewhere for multiple log output
21:11 < stillunknown> because i redirected the output into one file and it was more than 190.000 lines output
21:11 < zoeloelip> in the file test.c 
21:11 < arachnist> yeah, in re.c
21:12 < zoeloelip> indeed, my mistake
21:12 < zoeloelip> they told me to uncomment these lines: #define OUTPUT_MULTIPLE_FILES
21:12 < zoeloelip> #define DONT_DUMP_CHANGED_REGS
21:12 < zoeloelip> #define DONT_DUMP_ABOVE_40000
21:15 < phh> yp
21:15 < arachnist> and just run renouveau
21:15 < arachnist> ? :>
21:15 < phh> yep
21:16 < arachnist> (a nice square appeared in the lower right corner
21:16 < zoeloelip> then just zip those file and make them available somewhere
21:16 < arachnist> ok, will do
21:16 < phh> when it will end :p
21:17 < arachnist> naturally ;>
21:17 < Gentle> what would you say? Is it currently dangerous to "just try out" nouveau?
21:17 < arachnist> is it ok if some other window temporairly gets above that fancy square?
21:17 < phh> arachnist, i think that it isn't
21:17 < phh> and i think you should do anything with the 3D card
21:18 < arachnist> ok
21:20 < arachnist> having nvidia drivers (8774) loaded is ok?
21:21 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit [Remote closed the connection]
21:21 < zoeloelip> arachnist: that is the whole point of that dump, see how the nvidia driver use the gpu
21:21 < arachnist> and last one: is it amd64-safe?
21:23 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
21:25 -!- pmdata [i=patrice@ANantes-154-1-90-97.w86-210.abo.wanadoo.fr] has joined #nouveau
21:31 < arachnist> here it is, hope i've done everything correctly |> http://hive.bsic.pl/~arachnist/gf7900gt-dump.tar.bz2
21:34 < arachnist> hmm, different results than on 'http://nouveau.sourceforge.net/tests/g71/'
21:54 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit ["Mike Patton"]
21:54 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
21:58 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit [Client Quit]
21:58 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
22:12 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
22:24 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit [Read error: 60 (Operation timed out)]
22:26 < pmdata> when I see NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID = 0x00000093
22:26 < pmdata> does it mean that the following data is to setup the register 0x93? and it is always the same, for the same usage?
22:27 < phh> pmdata, same for same usage i think
22:28 < pmdata> so we can make an array that maps a const_id to a name?
22:28 < darktama> pmdata, no I don't think we can.
22:29 < pmdata> why not?
22:29 < darktama> some consts may well be used for the same purpose each time, but it doesn't necessarily mean they are
22:29 < darktama> ie. nvidia sometimes puts the modelview-projection matrix in 0-3... sometimes in 0x2c->0x2f
22:29 < darktama> basically they can be used for whatever you want
22:30 < pmdata> but the gpu must know what it should be used for
22:30 < darktama> yes, a vertex program will reference them
22:31 < pmdata> so there must be something that say const_id = 0x93 is point parameter
22:32 < darktama> I'm not familiar with point parameters.. what are they, and do you have a dump? :P
22:33 < pmdata> arachnist posted some earlier, I'm reading them, there is arb_point_parameter dump
22:35 < pmdata> for me const_id 0x2c to 0x31 are clip planes
22:35 < darktama> yes, but they don't have to be.. you could upload them anywhere you want
22:35 < pmdata> const_id 8 seems to be projection or modelview matrix
22:35 < darktama> and just change the offset in the vertex program
22:38 < pmdata> const_id 0 is another matrix
22:38 < pmdata> I don't see why it should be more complex than that
22:39 < darktama> my point is that the hardware doesn't say  "constant 0xXX is for this purpose.. etc".. you can make them anything you want
22:39 < darktama> I just ran the arb_point_parameters test here.  Take a look at the vertex program, you'll see it referencing c[0x93] (etc)
22:40 < pmdata> yes I see that, and?
22:41 < darktama> well, there's really nothing else except that I think it's misleading to start naming general-purpose vtxprog constant regs
22:43 < pmdata> I agree that on a different chip, they maybe different values, like nv20/nv30 tcl differences, but not much more
22:43 < pmdata> and from what I see c,10,14,18 are the texture matrix
22:44 < darktama> it could change between driver versions even..
22:44 < pmdata> till I see a different value for same dump on same chip, I think I got it right
22:46 -!- pmdata is now known as pmdata_tv
22:52 < stillunknown> does the current driver (and nv?) accelerate XRender?
22:53 < darktama> nvidia accelerate PictOpOver and PictOpSource, but that's it afaik
22:54 < darktama> nv (EXA) accelerates the same, but has a lot of fallback cases (mask pictures, alpha channel)
22:55 < stillunknown> do you mean it accelerates it enough or is very incomplete?
22:55 < darktama> very incomplete
22:55 < darktama> it'll hit fallbacks when using xcompmgr with ARGB windows
22:57 < stillunknown> is there any decent information of the XRender spec?
22:57 < darktama> hmm, there's the protocol doc
22:57 < darktama> that's the most complete I know of
23:02 -!- pef [n=loic@ubuntu/member/pef] has quit [Remote closed the connection]
23:05 < stillunknown> it's not the easiest thing to find, do you have a doc?
23:05 < stillunknown> darktama: forgot to trigger
23:06 < darktama> yes, I did.. just got to find it again :S
23:06 < stillunknown> i mean a link
23:06 < darktama> http://webcvs.freedesktop.org/xlibs/Render/protocol?revision=1.9
23:07 < darktama> it might be out of date (couldn't find it in git).. but it's still relevant
23:09 < stillunknown> http://gitweb.freedesktop.org/?p=xorg/doc/xorg-docs.git;a=blob_plain;h=65750ac67e8824bc1274639961695e63f08f9d5a;f=specs/Render/protocol
23:09 < stillunknown> bit of searching
23:09 < stillunknown> still the same dox
23:09 < stillunknown> *doc
23:10 < darktama> ah, thanks :)
23:11 < stillunknown> is this(poor Xrender implementation) the reason the nvidia driver is better?
23:13 < darktama> not necessarily, but their driver *does* Render better than we do.  as for other stuff, I haven't checked the difference yet
23:13 < stillunknown> it looks like it's missing 99% of the protocol
23:13 < darktama> yeah, nvidia are too
23:14 < darktama> most of the other open-source drivers do render a *lot* better than nvidia
23:14 < stillunknown> is this a problem of lacking hardware interface?
23:15 < stillunknown> or is simply a case of undefined protocol
23:15 < darktama> I don't think the 2D engine is capable of doing a decent implementation, but we can't be sure without docs
23:15 < darktama> radeon/intel open-source drivers use the 3D engine
23:17 -!- phh [n=phh@moi.phhusson.info] has quit [Remote closed the connection]
23:18 < stillunknown> this seems like a shitload of things to implement even for an experienced programmer (which i'm not)
23:18 < darktama> EXA breaks the Render protocol down.. I'm not sure of the details, but the bulk of the work for Render is done in the EXA Composite hooks
23:19 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
23:19 < stillunknown> it sounds bad that you don't know :-)
23:20 < darktama> well, I know how to write the composite hook.. I don't know how much of Render EXA supports accelerating
23:31 < stillunknown> btw. XRandr is broken
23:33 < darktama> hm, in what way?
23:35 < phh> stillunknown, it ever worked Oo
23:35 < phh> arf
23:35 < phh> xrandr doesn't mean rotation
23:36 < stillunknown> oh, well i just turned it on earlier today and my screen looked like 2 small screens and distortion in one
23:36 < phh> huhu
23:36 < stillunknown> what is xrandr precisely?
23:37 < darktama> is this rotation, or the mode-change part?
23:37 < darktama> X Resize and Rotate
23:37 < stillunknown> Option Rotate "xrandr" is all i know
23:37 < stillunknown> *did
23:38 < stillunknown> and on startup isn't wasn't looking usefull, so i killed the x server
23:38 < darktama> yeah, that falls back to complete software-rendering anyway.. I guess I broke that somehow (or it was already broken? never tested it)
23:40 < stillunknown> first time i tested it, only to make sure it wasn't on by default (which would have meant losing the little bit of 2d accel that exists)
23:41 < darktama> I'm going to start cleaning up the ddx from the mman changes tomorrow, I'll take a look at it at the same time
23:42 < darktama> anything else broken (that works in nv) that you've noticed?
23:42 < stillunknown> i didn't test RandR in nv
23:43 < stillunknown> just mentioned it's broken
23:43 < darktama> I was just guessing it works in nv :)
23:44 < stillunknown> i'd wish i could help with the driver writing, but sourcecode can be so intimidating
23:45 < darktama> hehe :)
23:45 < stillunknown> just looking at the comments in the EXA header makes me go ...
23:46 < darktama> yeah, it can be a bit like that at times
23:47 < stillunknown> if there's anything specific in the sourcecode that needs repeated cleanup, just yell and maybe i can help (i can mimic things pretty good sometimes)
23:48 < stillunknown> away for a while
23:48 < darktama> sure :)
--- Log closed Вск Авг 27 00:00:23 2006
