--- Log opened Чтв Авг 10 00:00:10 2006
00:00 -!- Duke` [n=gnu@ANantes-251-1-144-35.w86-210.abo.wanadoo.fr] has quit [Connection timed out]
00:01 < swany_> nice... http://intellinuxgraphics.org/
00:01 -!- Duke` [n=gnu@ANantes-251-1-100-25.w86-203.abo.wanadoo.fr] has joined #nouveau
00:05 < pmdata> sorry for being late, will try to make commits anyway :)
00:07 < marcheu> pmdata: hi
00:07 < marcheu> swany_: yeah, too bad they don't have a nice memory manager :)
00:11 -!- Unbeliever [n=hazel@212.145.81.31] has quit [Read error: 104 (Connection reset by peer)]
00:13 < swany_> marcheu, oh, I thought you meant to base yours on that memory manager?
00:13 < marcheu> swany_: well, I've begain doing so with the via/sis one
00:14 < marcheu> and it's not good enough for heavy use, especially if you want to allocate everything (including all those pixmaps) with it
00:15 < marcheu> the intel/radeon memory managers are really just able to take blocks and slice them, not much interest
00:20 < marcheu> ah well, I bet you just don't get the code right the first time
00:21 < marcheu> pmdata: btw did you see the changes I did to the nv28 box ?
00:21 < pmdata> I read the log, you added an nv06
00:21 < marcheu> yup, and using renouveau on the nv28 changed a bit in the move
00:21 < pmdata> nv40[0x370/4] seems to be an extra color mask for the other color buffers
00:21 < marcheu> anyway, if you read the log, you know
00:22 < pmdata> bit 4-7 for color buffer 1, bits 8-11 for color buffer 2, bits 12-15 for color buffer 3
00:22 < pmdata> we'll need someone doing dumps on a stereo nv40
00:23 < marcheu> ajax has one...
00:26 -!- K [i=hazel@tor/session/external/x-bd84ce549ed13635] has joined #nouveau
00:28 -!- K [i=hazel@tor/session/external/x-bd84ce549ed13635] has quit [Remote closed the connection]
00:37 < pmdata> marcheu> I would like to know if nv28 has a second modelview matrix for vertex weight stuff
00:38 < pmdata> it would require you install an old driver (3xxx or 4xxx), and run the test ext_vertex_weighting (disable the extension check before)
00:40 < marcheu> ok, I'll look at that one of these days
00:42 < pmdata> disabling the extension check was necessary (at least for 4xxx), because vertex_weight functions where still present in the libGL albeit not announced in the gl extension string
00:43 < marcheu> oh btw
00:43 < marcheu> the nv28 has some GLSL support that way
00:43 < marcheu> it's not anounced in the extension strings, but you can run some programs
00:44 < marcheu> of course it often fallbacks in software, but I already saw simpler programs making use of the hw
00:44 < pmdata> nv20 seems to be an improved nv10, whereas nv30+ tcl engines are completely different
00:46 < marcheu> yes, but maybe we can figure out more stuff for nv20 vertex/fragment shaders by using that hidden GLSL stuff
00:47 < pmdata> do as you feel
00:48 < marcheu> do what ? I'm doing the memory manager now, and nothing else until it's done
00:48 < pmdata> wouldn't it be better to have a single memory manager, whatever the hw (and whatever the vendor)?
00:49 < pmdata> they all do the same
00:51 < marcheu> yeah, except none does what we need. the issue is that no one has dared allocating all his pixmaps from the drm yet
00:51 < marcheu> I've been looking at existing memory managers, and started leveraging the sis one, but the amount of work seems huge to handle everything we need
00:52 < pmdata> ouch
00:55  * pmdata should try npot non-rectangle_nv textures, it should show different stuff
00:55 < pmdata> hum, requires arb_texture_non_power_of_two
00:56 < marcheu> hmm ?
00:56 < marcheu> that one is the mipmapped one IIRC
00:56 < pmdata> well, it's using gl_texture_2d target in texture_format test, but change width and height to be npot
00:57 < marcheu> the nv_texture_rectangle doesn't do mipmaps and shoud be available even on nv10 IIRC
00:57 < pmdata> yep, texture_rectangle is present on nv10
00:57 < marcheu> (and arb_texture_rectangle which is the same)
00:57 < pmdata> but real npot textures is different, and simpler
00:58 < marcheu> yes, but not available before nv40
00:59 < pmdata> that's what I was wondering
00:59 < pmdata> qfire will have to try this
01:00 < pmdata> qfire> change TEXTURE_WIDTH and TEXTURE_HEIGHT to be npot values (15x13 for example), keep GL_TEXTURE_2D as target in texture_format test
01:02 -!- K [i=hazel@tor/session/external/x-76130086d6b7248b] has joined #nouveau
01:14 -!- atcl [n=Universe@L994f.l.pppool.de] has quit [Read error: 104 (Connection reset by peer)]
01:19 -!- Myrizio [n=Myrizio@host47-101.pool80104.interbusiness.it] has joined #nouveau
01:25 -!- Duke` [n=gnu@ANantes-251-1-100-25.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
01:30 -!- darktama [n=darktama@gentoo/contributor/darktama] has quit [Read error: 113 (No route to host)]
01:40 < pmdata> I should also try playing with ext_texture_filter_anisotropic, should be set somewhere in texture format when mipmapping enabled
01:56 -!- pmdata [i=patrice@ANantes-154-1-41-159.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
02:19 -!- tibbs|h is now known as tibbs
02:24 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Remote closed the connection]
02:30 -!- K [i=hazel@tor/session/external/x-76130086d6b7248b] has quit [Remote closed the connection]
02:30 < qfire> pmdata: http://nouveau.sourceforge.net/tests/g71/card_10de-0291_test_texture_format_2d_15x13.txt
03:02 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Read error: 131 (Connection reset by peer)]
03:04 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
04:04 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau
04:28 < airlied> marcheu: you really need to look at the TTM stuff.
04:29 < airlied> marcheu: it will be the new DRI memory manager.
04:29 < airlied> it isn't finished but would be much more useful to put effort into it than wasting time on another one..
04:52 -!- swany_ [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit [Remote closed the connection]
04:53 -!- shenki [n=shenki@ppp174-101.lns3.adl4.internode.on.net] has joined #nouveau
05:04 -!- shenki [n=shenki@ppp174-101.lns3.adl4.internode.on.net] has quit [Read error: 60 (Operation timed out)]
06:39 -!- darktama [n=darktama@gentoo/contributor/darktama] has joined #nouveau
07:06 -!- Myrizio [n=Myrizio@host47-101.pool80104.interbusiness.it] has quit [Read error: 113 (No route to host)]
07:26 -!- shenki [n=shenki@ppp173-18.lns3.adl4.internode.on.net] has joined #nouveau
09:06 -!- Netsplit calvino.freenode.net <-> irc.freenode.net quits: qfire, nano-
09:08 -!- Netsplit over, joins: nano-
09:08 -!- Netsplit over, joins: qfire
10:08 -!- Duke` [n=gnu@ANantes-251-1-100-25.w86-203.abo.wanadoo.fr] has joined #nouveau
10:15 -!- Duke` [n=gnu@ANantes-251-1-100-25.w86-203.abo.wanadoo.fr] has quit [Read error: 104 (Connection reset by peer)]
10:15 -!- Duke` [n=gnu@ANantes-251-1-100-25.w86-203.abo.wanadoo.fr] has joined #nouveau
10:32 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has quit [Read error: 110 (Connection timed out)]
10:34 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has joined #nouveau
12:00 -!- katakombi [n=kombrisn@dslb-084-056-146-087.pools.arcor-ip.net] has joined #nouveau
12:00 -!- Unavowed [n=silent@host81-158-183-201.range81-158.btcentralplus.com] has joined #nouveau
12:12 < marcheu> airlied: we need something that works now, and can be used by the DDX & DRI. not a user space prototype
13:08 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
13:31 -!- Myrizio [n=Myrizio@host99-96.pool80104.interbusiness.it] has joined #nouveau
13:41 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
14:49 -!- katakombi [n=kombrisn@dslb-084-056-146-087.pools.arcor-ip.net] has quit [Remote closed the connection]
15:06 -!- shenki [n=shenki@ppp173-18.lns3.adl4.internode.on.net] has quit ["http://xkcd.com/comics/fourier.jpg"]
15:29 -!- Koala_BR [n=KoalaBR|@195.227.227.59] has joined #nouveau
16:33 -!- barnaby [n=barnaby@mailgate.echelonl.com] has joined #nouveau
16:36 -!- barnaby [n=barnaby@mailgate.echelonl.com] has quit ["using sirc version 2.211+KSIRC/1.3.12"]
16:36 -!- barnaby [n=barnaby@mailgate.echelonl.com] has joined #nouveau
16:39 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit [Remote closed the connection]
16:53 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
16:57 -!- philv [n=bleep@cowpig.ca] has quit [Read error: 110 (Connection timed out)]
17:29 -!- philv [n=bleep@cowpig.ca] has joined #nouveau
17:31 < marcheu> darktama: did you make use of the map that covers the whole FB ? because I'm about to remive it
17:31 < marcheu> remove
17:32 < darktama> hmm, I don't think I did.. I guess we'll find out if anything breaks :)
17:37 -!- atcl [n=Universe@L994f.l.pppool.de] has joined #nouveau
17:47 -!- Koala_BR [n=KoalaBR|@195.227.227.59] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.12/20050920]"]
17:53 < barnaby> Hi all, just wanted to offer to test anything on my NV18GL [Quadro4 NVS AGP 8x
17:53 < barnaby> ]
17:55 < barnaby> I am currently running the binaries on xorg-7.0 with TwinView and am happy to do any testing that would help (can't really help with code though)
18:00 -!- Myrizio_ [n=Myrizio@host22-98.pool80104.interbusiness.it] has joined #nouveau
18:01 < marcheu> barnaby: you could try some renouveau tests on it, alghouth I think my softquadroed geforce 4 mx 4000 is the same
18:01 -!- Myrizio [n=Myrizio@host99-96.pool80104.interbusiness.it] has quit [Read error: 110 (Connection timed out)]
18:04 < barnaby> Just downloaded and run renouveau-daily.tgz, flashs a white triangle on black background and a lot of text that meens little to me?
18:04 < barnaby> Should I post the output somewhere?
18:05 < marcheu> nope
18:05 < marcheu> instead open re.c
18:05 < barnaby> ok?
18:05 < marcheu> and uncomment OUTPUT_MULTIPLE_FILES
18:05 < marcheu> then open main.c
18:05 < marcheu> and uncomment all test_*()
18:06 < marcheu> it'll then output one file for each test, and you can put that online, or tar it & send it to me
18:09 < barnaby> had to leave test_fragprog2 and test_vtxprog3 commented to get it to compile
18:14 < barnaby> The results are at http://learningmatters.com/renouveau.tar.gz
18:17 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has quit ["Leaving"]
18:20 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
18:22 -!- philv [n=bleep@cowpig.ca] has quit ["leaving"]
18:22 < marcheu> ok, if you plan on updating it with newer tests I can give you access to the sf.net site where we put our test results
18:29 < barnaby> Happy to post newer results if that is helpful, my sf.net user is bware
18:29 < marcheu> well, we add new tests all the time
18:29 < marcheu> so newer results are always welcome :)
18:34 < marcheu> ok, you're added
18:34 < marcheu> path to the tests on shell.sf.net is /home/groups/n/no/nouveau/htdocs/tests
18:34 < marcheu> yours sit in nv18gl
18:35 < marcheu> maybe I should give you ownership of those files, btw, so that you can overwrite them. one moment
18:36 < marcheu> hmm, I can't chown
18:36 < marcheu> I'll remove them, and let you upload them yourself
18:54 -!- philv [n=bleep@cowpig.ca] has joined #nouveau
18:55 < barnaby> okay the files are on sf.net now, I will try and keep them up to date
18:55 -!- atcl [n=Universe@L994f.l.pppool.de] has quit [Read error: 113 (No route to host)]
18:56 < marcheu> cool
18:58 < barnaby> ls
18:59 -!- shenki [n=shenki@ppp175-80.lns3.adl4.internode.on.net] has joined #nouveau
18:59 < philv> porn/ bad_porn/
18:59 < darktama> .reall_bad_pr0n/
19:05  * shenki leaves
19:06  * darktama debates going to bed or finishing NV20 vtxprogs
19:07 < marcheu> darktama: btw did you read the logs ?
19:07 < darktama> hm?
19:07 < marcheu> darktama: the nv28 can do some glsl, which is not exposed in the extension list
19:07 < marcheu> but the entrypoints are here
19:07 < darktama> ah, yes I did see that
19:07 < darktama> I guess it can only do glsl vtx progs?
19:07 < marcheu> no, it can do frag progs as well
19:08 < marcheu> but they sometimes end up being sw emulated
19:08 < darktama> really? I didn't know it had fragprogs
19:09 < marcheu> I guess it's emulated trough texture_shader
19:21 -!- shenki is now known as sHenKi_h4TeS_cAp
19:21 -!- sHenKi_h4TeS_cAp is now known as shenki
19:28 -!- lraptor [n=laurens@host-84-235-192-222.satlynx.net] has joined #nouveau
19:39 -!- shenki [n=shenki@ppp175-80.lns3.adl4.internode.on.net] has quit ["http://xkcd.com/comics/fourier.jpg"]
19:53 -!- Unavowed [n=silent@host81-158-183-201.range81-158.btcentralplus.com] has quit ["leaving"]
19:57 -!- barnaby [n=barnaby@mailgate.echelonl.com] has quit ["using sirc version 2.211+KSIRC/1.3.12"]
20:10 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
20:11 -!- ajmitch [n=ajmitch@ubuntu/member/ajmitch] has quit [Connection timed out]
20:14 -!- lraptor [n=laurens@host-84-235-192-222.satlynx.net] has left #nouveau ["Konversation terminated!"]
20:57 -!- pmdata [i=patrice@ANantes-154-1-58-138.w81-53.abo.wanadoo.fr] has joined #nouveau
21:15 < pmdata> hello
21:22 < qfire>  hi 
21:22 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
21:22 < qfire> pmdata: not sure if you saw in the logs, but I ran the npot version of test_texture_format
21:23 < qfire> http://nouveau.sourceforge.net/tests/g71/card_10de-0291_test_texture_format_2d_15x13.txt
21:23 < pmdata> yep, I saw that, it seems the texture unit accept npot stuff directly
21:23 < pmdata> marcheu> I made more dumps using fsaa yesterday
21:23 < qfire> it rounded the 15 to 16, or am I reading that wrong?
21:24 -!- KoalaBR [n=KoalaBR@port-83-236-14-17.dynamic.qsc.de] has joined #nouveau
21:25 < pmdata> NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT(1)       = width = 15 | height = 13
21:25 < KoalaBR> Hi
21:27 < qfire> I missed that before, was looking at the pitch, which makes sense they would round the memory buffer.
21:28 < qfire> still getting used to reading the dumps
21:30 < pmdata> fsaa is done on nv10 by defining a texture_rectangle from color buffer
21:32 < KoalaBR> marcheu: I have a newer version of the doc online at http://www.ping.de/sites/koala/ please have a look at the first page including the "Using renouveau" Link and tell me what you think
21:32 < KoalaBR> Comments from others are welcome too
21:32 < pmdata> and I have a new nv10 unknown command "NV10_TCL_PRIMITIVE_3D      [0x03f0/4] = 0.000000 | UNKNOWN = 00000002"
21:32 < pmdata> only used when fsaa enabled
21:34 < pmdata> hum, maybe 3f0 is some kind of 'render to texture', will have to check that
21:34 -!- K [i=hazel@85-57-132-109.gij1.adsl.uni2.es] has joined #nouveau
21:36 < pmdata> hum EXT_pixel_buffer_object, should make a test function for this
21:54 -!- `Duke` [n=gnu@ANantes-251-1-137-54.w86-210.abo.wanadoo.fr] has joined #nouveau
22:01 -!- Duke` [n=gnu@ANantes-251-1-100-25.w86-203.abo.wanadoo.fr] has quit [Read error: 110 (Connection timed out)]
22:12 < pmdata> renouveau segfault when fsaa enabled and I use a small resolution
22:14 < marcheu> pmdata: conceptually, nothing should be needed for render to texture
22:15 < marcheu> just point the framebuffer at the texture adress
22:15 < marcheu> if everything was linear, that is
22:15 < marcheu> but, since most cards have the framebuffer and/or the texture tiled, usually you have to tell the engine to tile/untile stuff
22:15 < marcheu> so maybe 3f0 does something like that
22:17 < marcheu> that, or it has native FSAA (other than multisampling, I mean)
22:17 < marcheu> if you want to try render to texture, you should also look at pbuffers
22:19 < marcheu> KoalaBR: looks pretty good
22:20 < marcheu> KoalaBR: there are some minor changes, like "=> fifo" can also be "=> quadro fifo"
22:20 < KoalaBR> marcheu: ok
22:21 < KoalaBR> marcheu: still adding to it, though
22:21 < KoalaBR> far from finished
22:21 < marcheu> well, it's good to have something to point newcomers at :)
22:21 < KoalaBR> That was the intention, nothing more
22:22 < marcheu> btw I like the new wording :)
22:22 < KoalaBR> :) 
22:23 < KoalaBR> Currently I'm writing down the clip planes for NV40 and NV30
22:23 < KoalaBR> NV30 seems complex, though
22:23 < marcheu> pmdata: I can send pbuffer code your way if you like. pbuffers are a mess to setup
22:24 < marcheu> KoalaBR: could you check that there are actually 2 sets of clipping planes on nv30 ?
22:24 < KoalaBR> marcheu: Have only that one dump available currently, so it's mostly guess work
22:24 < marcheu> any nv30 owner in here ?
22:25 < marcheu> (nv30 is any kind of geforce 5x00, you know you can help ! :)
22:26 < KoalaBR> Regarding the doc: I will maintain this if interest exists either as a separate patch (as there are a long text only parts added to the source code) or I can add it to the current cvs version - whatever you like
22:28 < KoalaBR> Perhaps we should a page "Small pleas for help" and add everything to it, we still need (like quadro owners who are willing to run tests, NV3x owners for clip tests...)
22:29 < marcheu> as you want. as long as the wording is correct, I have no issue with it in cvs or on the website
22:29 < KoalaBR> Seems, as if I would need a Wiki account then? :)
22:38 < pmdata> marcheu> if you have some test function for pbuffer, I welcome it
22:41 < marcheu> KoalaBR: everyone can create one
22:42 < marcheu> pmdata: actualyl what I have comes from mesa, glxpbdemo.c
22:45 < pmdata> I have this: ENGINE_SW[0] = eb60aa55 = class = 0a55 (UNKNOWN) | page entry linear | dma_access = wo | dma_target = NV mem | dma adjust = eb6
22:45 < KoalaBR> marcheu: It was a somewhat theoretical question, a kind of joke. I will add the pages or the link to the wiki and add the "Help: Small task" too. But first I describe the few things I found out about clipplanes on NV30/NV40
22:45 < pmdata> does it mean I have an unknown object 0x55 (for nv 0x0a = nv10) ?
22:47 < pmdata> this is DX6_MULTITEX_TRIANGLE
22:47 < marcheu> pmdata: usually that means a variant of 0x55 for the NV0A
22:47 < marcheu> but NV0A doesn't exist
22:48 < marcheu> KoalaBR: well try to find the login screen on the wiki, and you'll see why I said that :)
22:48 < marcheu> KoalaBR: even better, try to login to another name afterwards
22:48 < KoalaBR> ?
22:49 < marcheu> pmdata: it happens a lot with later cards. some card revisions feature an object with a different class
22:49 < pmdata> where do I mask the 0x0a part in renouveau?
22:49 < marcheu> KoalaBR: there is no login screen :)
22:49 < marcheu> pmdata: I think there's a bug somewhere, because NV0A doens't exist
22:51 < KoalaBR> ;)
22:51 < pmdata> 0x0a = 10
22:52 < marcheu> pmdata: yes, but usually it uses BCD
22:52 < pmdata> maybe nvidia changed later
22:52 < marcheu> maybe
22:53 < marcheu> but even the nv18 has a 0x17?? variant
22:54 < KoalaBR> marcheu: Is it possible to setup renouveau for the case that the same object [0x1478/4] does different things for different card families (in this case NV40 and NV30)?
22:54 < marcheu> KoalaBR: yes, but again, I think it's the same
22:54 < marcheu> KoalaBR: you just have to find a nv30
22:55 < marcheu> anyway, have to go, bbl
22:59 < KoalaBR> Thanks bye
23:03 -!- pmdata [i=patrice@ANantes-154-1-58-138.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
23:47 -!- KoalaBR [n=KoalaBR@port-83-236-14-17.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"]
23:51 -!- KoalaBR [n=KoalaBR@port-83-236-14-17.dynamic.qsc.de] has joined #nouveau
23:52 < KoalaBR> Hi, does anyone here has an idea how to use the Nouveau Wiki?
23:55 -!- KoalaBR [n=KoalaBR@port-83-236-14-17.dynamic.qsc.de] has quit [Client Quit]
--- Log closed Птн Авг 11 00:00:11 2006
