--- Log opened Срд Авг 02 00:00:04 2006
00:00 -!- Colbert [n=colby@adsl-69-211-9-63.dsl.chcgil.ameritech.net] has joined #nouveau
00:00 < Colbert> Hey
00:01 < Colbert> pmdata. need nv17 dumps still? updating from cvs
00:03 < marcheu> nano: I cared about eating, I'm doing the file stuff now
00:03 < pmdata> of course, yes
00:03 < pmdata> I love nv1x dumps
00:03 < Colbert> what do you need run?
00:04 < nano> :-)
00:04 < Colbert> look like i've missed alot lately
00:06  * pmdata would like dumps for all tests, each in a separate file
00:06 < pmdata> and a coffee
00:08 < Colbert> making coffee
00:08 < pmdata> thanks
00:17 < Colbert> hmm, my code seems to be causing trouble with this again, it will take me a little while to sort this out
00:26 < Duke`> hi people
00:27 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
00:30 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
00:30 < Duke`> strange compilation problem:
00:30 < Duke`> re2.o: In function `dump_after':/home/henry/Desktop/renouveau/re.c:584: undefined reference to `glFlush'
00:30 < marcheu> don't use make -j
00:30 < Duke`> ah
00:30 < Duke`> what is -j ?
00:30 < marcheu> we don't have depencies right :)
00:30 < marcheu> -j is when you build multiple files at once
00:30 < marcheu> are alternatively, redo a make and it will work
00:31 < marcheu> and someone shoudl fix the makefile
00:31 < Duke`> ok
00:35 < Duke`> does someone need some tests with nv34?
00:36 < Colbert> had the same prob as duke, thought it was my fault
00:36 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has joined #nouveau
00:38 < Colbert> does the non zero sized triangle give better data?
00:47 -!- Gloubi [n=Gloubi@ABordeaux-253-1-59-152.w82-125.abo.wanadoo.fr] has quit ["I'll be back"]
00:50 < pmdata> duke> I got some nv34 dumps 2 weeks ago, but if new stuff have been added to renouveau, then redumps are useful, to see unknown stuff
00:51 < pmdata> each week, a full dump of the tests should be done on each card, to see where progress have been made
00:51 < marcheu> pmdata: cool, as soon as the 25 tinderboxes I ordered and the matching 25 nvidia cards arrive in the data center, I'll setup that :)
00:53 < pmdata> I was able to run renouveau with 6106, I had to add '-lGL', and set the gl lib filename to NULL (in this case, SDL load pointers from static lib)
00:54 < marcheu> no, when the gl lib filename is NULL, it's dynamically loaded from a predefined libGL
00:55 < marcheu> see src/video/x11/SDL_x11gl.c for example
00:55 < pmdata> yes, I read the src, but the comment says so
00:56 < marcheu> which comment ?
00:56 < pmdata> anyway I wonder why I have to statically link libGL to get function pointers
00:56 < marcheu> you can't/must not link to libGL statically
00:56 < marcheu> it's a really bad idea
00:57 < pmdata>  Passing a NULL path means load pointers from the application */ just before x11_gl_loadlibrary
00:58 < marcheu> yeah and then...
00:58 < marcheu>         if ( path == NULL ) {
00:58 < marcheu>                 path = SDL_getenv("SDL_VIDEO_GL_DRIVER");
00:58 < marcheu>                 if ( path == NULL ) {
00:58 < marcheu>                         path = DEFAULT_OPENGL;
00:58 < marcheu>                 }
00:58 < marcheu>         }
00:58 < marcheu> where DEFAULT_OPENGL is the default system location for libGL
00:58 < pmdata> strange, when I add '-lGL', then libGL.so can be loaded, if I don't , loading the libGL fails
00:59 < pmdata> maybe something with glX?
00:59 < marcheu> -lGL is not static linking
00:59 < marcheu> it's dynamically linking
00:59 < pmdata> oops
00:59 < marcheu> I though you were talking about things like -static  -lGL :)
01:00 < marcheu> usually we say dynamically linking and dlopening to differentiate
01:00 < pmdata> ah, bad vocabular on my side then
01:00 < marcheu> yeah
01:01 < pmdata> anyway, it seems texture 3D is done in software on nv1x
01:01 < marcheu> yup
01:01 < marcheu> only nv20 has hw support for 3D textures
01:06 < pmdata> I don't have a dump for texture_3d
01:06 < marcheu> on what card ?
01:09 < Colbert> do the quadros have 3dtexture? voxel data is common with things like MRI and CAT scans, but I cant see someone using a pc for that
01:09 < marcheu> all nv20+ have 3D textures
01:09 < marcheu> voxel visualization is my phd topic btw :)
01:09 < pmdata> for nv20 :) or does someone already found how it is done?
01:10 < marcheu> pmdata: for nv30 darktama found the 3D bit, I extended it to nv20 without further testing (the texture engine seemd the same)
01:11 < pmdata> ok
01:11 < pmdata> hm, vtxprog segfault with 6106
01:15 < pmdata> I should also update my dyn loader for glsl, there are missing functions
01:16 < Duke`> grumf lost the X-axis on my mouse :/
01:17 < pmdata> rotate your screen :)
01:17 < Duke`> seems to be xorg, same problem with another mouse
01:18 < Duke`> will restart it
01:18 -!- Duke` [n=gnu@ANantes-251-1-81-125.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
01:18 < Colbert> nv17-pci-test_blends dump:more to come
01:18 < Colbert> http://rafb.net/paste/results/z82aKP84.html
01:19 -!- Duke` [n=gnu@ANantes-251-1-81-125.w86-203.abo.wanadoo.fr] has joined #nouveau
01:19 < Duke`> it's better o/
01:20 < pmdata> hum, 0x1658/4 = 1, some stuff enabled
01:20 < Colbert> nv17-pci-test_blends2 dump:more to come
01:20 < Colbert> http://rafb.net/paste/results/d45cmc17.html
01:20 < pmdata> 0xd74/4 = 0
01:21 -!- EdB [n=EdB@ARennes-251-1-152-124.w86-214.abo.wanadoo.fr] has quit ["Konversation terminated!"]
01:26 < Colbert> nv17-pci-test_clear dump:more to come
01:26 < Colbert> http://rafb.net/paste/results/w0il4413.html
01:29 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has joined #nouveau
01:32 < Colbert> nv17-pci-test_color_mask dump:more to come
01:32 < Colbert> http://rafb.net/paste/results/EuLwyc19.html
01:34 < pmdata> d68,d6c seems to be a clear command and mask, like on nv20 and nv30
01:34 < Colbert> nv17-pci-test_depth dump:more to come
01:34 < Colbert> http://rafb.net/paste/results/fq3SyZ53.html
01:35 < pmdata> lma zbuffer??
01:35 < marcheu> yup, I think so too
01:35 < pmdata> yep, it is always smaller than zbuffer
01:35 < marcheu> starting nv17, all cards are supposed to have it
01:36 < pmdata> then this is it
01:36 < marcheu> good, please name it accordingly
01:36 < marcheu> so that we can match that to higher level functionality
01:36 < marcheu> and so that we can brag about it, too
01:37 < pmdata> LMA_DEPTH?
01:37 < marcheu> as you like
01:37 < marcheu> LMA_CLEAR probably
01:37 < pmdata> DEPTH_MIPMAP :)?
01:38 < pmdata> go for LMA_
01:42 < Colbert> nv17-pci-test_depth2 dump:more to come
01:42 < Colbert> http://rafb.net/paste/results/K6kijT42.html
01:45 < Colbert> nv17-pci-test_dither dump:more to come
01:45 < Colbert> http://rafb.net/paste/results/Z2PLUy53.html
01:47 < Colbert> nv17-pci-test_enable dump:more to come
01:47 < Colbert> http://rafb.net/paste/results/1c0h8S84.html
01:48 < pmdata> hum, /me think 0x1658 = lma_depth_enable ?
01:48 -!- Duke` [n=gnu@ANantes-251-1-81-125.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
01:49 < pmdata> yes
01:50 < Colbert> nv17-pci-test_lights dump:more to come
01:50 < Colbert> http://rafb.net/paste/results/eSqg4Y30.html
01:50 -!- tibbs|h is now known as tibbs
01:54 < pmdata> 3f8,3fc: lma depth range?
01:54 < pmdata> no
01:54 < Colbert> nv17-pci-test_lights_two_sides dump:more to come
01:54 < Colbert> http://rafb.net/paste/results/jWDbzp61.html
01:55 < Colbert> which test? (re: 3f8,3fc: lma depth range?)
01:55 < pmdata> there is no test, I just try to find what 3f8,3fc can be
01:55 < pmdata> always set when polygon offset or depth func is changed
01:55 < marcheu> zflush ?
01:55 < marcheu> zcache flush
01:56 < marcheu> set to what ?
01:56 < pmdata> always set to 0
01:56 < marcheu> yeah, maybe some kind of flush
01:56 < marcheu> or maybe
01:56 < marcheu> zpass counter
01:56 < marcheu> used for occlusion queries
01:57 < pmdata> does nv17 has any occlusion_query extension?
01:57 < marcheu> I don't think so
01:58 < marcheu> no it doesn't
01:58 < marcheu> according to delphi3d.net that requires a nv20
02:02 < pmdata> fsaa setting?
02:02 < pmdata> for multisampling
02:05 < pmdata> colbert> could you set __GL_FSAA_MODE to 1, then 2, in single test?
02:05 < Colbert> ok
02:06 < Colbert> which test do you want?
02:06 < pmdata> the draw_pixel on
02:07 < Colbert> one moment
02:07 < pmdata> hum, wait, add a tri() in it, before dump_after()
02:14 < pmdata> colbert> I have to go, you can post the link there, will have a look tomorrow
02:14 -!- pmdata [i=patrice@ANantes-154-1-44-18.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
02:14 < Colbert> could i set__GL_FSAA_MODE with SDL_GL_SetAttribute?
02:15 < marcheu> nope
02:15 < marcheu> it's an env variable
02:15 < Colbert> darn
02:15 < marcheu> export __GL_FSAA_MODE=1
02:15 < marcheu> then ./renouveau
02:15 < Colbert> easier, ok
02:23 < Colbert> no difference with fsaa 1 or 2, big change with  __GL_FSAA_MODE=0
02:26 < Colbert> could it be Allow Flipping instead?
02:29 < Colbert> nope. no change
02:37 < Colbert> glPolygonMode?
02:37 < Colbert> writing test
02:37 < marcheu> I think we have a test for that
02:37 < Colbert> cool
02:37 < marcheu> test_polygon_mode
02:37 < marcheu> I'll commit the "separate file output" stuff in one minute
02:37 < marcheu> when I get it to work :)
02:51 < ddl> hey
02:52 < marcheu> hey ddl what's new under the sun ?
02:54 < ddl> been working for a couple of hours every night, think im closing in on something... :)
03:16 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Remote closed the connection]
03:29 < marcheu> ok, the option works now
03:29 < marcheu> just uncomment OUTPUT_MULTIPLE_FILES at the top of re.c
03:39 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
03:39 < Colbert> Things flow  here so. I was thinking of making a change similar to that, but I would have taken MUCH longer to get it done
03:40 < marcheu> well, as long as pmdata is happy... :)
03:41 < marcheu> I'll try to save a run for him before I go to sleep
03:44 < marcheu> hmm I'll make the files world-readable by default
03:44 < marcheu> so that I can scp to the webserver directly
03:45 -!- qfire is now known as qfire_away
03:49 < Colbert> i misunderstood your last change, I think. How do I use it?
03:49 < marcheu> you edit re.c
03:50 < marcheu> and change 
03:50 < marcheu>  // #define OUTPUT_MULTIPLE_FILES
03:50 < marcheu> into 
03:50 < marcheu> #define OUTPUT_MULTIPLE_FILES
03:50 < Colbert> did that, 
03:50 < marcheu> so
03:50 < marcheu> renouveau has output card_*_*.txt files in the current dir
03:50 < marcheu> one for each test
03:51 < Colbert> thats what I thought, not working on this end for some reason
03:51 < marcheu> you have no files ?
03:51 < Colbert> nope
03:51 < marcheu> you did a full recompile ?
03:52 < Colbert> yep
03:52 -!- qfire_away is now known as qfire
03:52 -!- qfire is now known as qfire_away
03:54 < Colbert> Sorry, Im just an Idiot, found the file.
03:54 -!- qfire_away is now known as qfire
03:54 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has quit ["leaving"]
03:55 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has joined #nouveau
03:55 < marcheu> ah, good
03:55 < marcheu> I'll upload my tests and go to sleep then
03:58 < Colbert> So, I can uncomment all the tests in test.c and it will make a file for each. I like it!
03:58 < marcheu> yup
04:00 < Colbert> I was thinking before about  adding command line switches for all the tests,
04:00 < Colbert> but thats just added work for quick tests
04:04 < marcheu> ok, this time is really bedtiem
04:05 < marcheu> http://icps.u-strasbg.fr/~marchesin/nvdri/tests/ holds the tests I did
04:26 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
04:30 < darktama> marcheu: pong.  but, you're probably in bed already :)
04:31 < ddl> hey :)
04:31 < darktama> hey ddl
04:51  * Colbert is away: Colbert
05:21 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau
05:53 -!- JKnife [n=jknife@unaffiliated/jknife] has joined #nouveau
06:10 -!- toresbe [i=tsb@developer.skolelinux.no] has joined #nouveau
06:10 < toresbe> philv: howdy :)
06:10 < toresbe> wow! This is a very neato project!
06:10 < toresbe> How do you guys reverse-engineer this? Do you look at what travels along the bus?
06:12 < toresbe> Is there a donation possibility for this? I can afford a small donation, I think.
06:12 < JKnife> toresbe: i have seen you around on freenode before... was it Arch?
06:12 < toresbe> JKnife: Hmm, howdy, I recognize your nick as well, but certainly not arch
06:13 < JKnife> Xubuntu?
06:13 < toresbe> nopes
06:13 < toresbe> OSdev? debian-edu/skolelinux? debian?
06:14 < JKnife> prob debian...
06:15 < toresbe> yeah. I haven't been there for a good while, which is why your nick was so distant.
06:15 < JKnife> ahh ok
06:16 < toresbe> anyway - neato, very neato.
06:18 < toresbe> nm my questions, btw.
06:22 < JKnife> beats me i JUST found out about this
06:40 -!- JKnife [n=jknife@unaffiliated/jknife] has quit ["leaving"]
08:39 -!- tibbs is now known as tibbs|h
11:38 < marcheu> darktama, lumag_offline: we need to talk a bit about our development plans I think. lumag had some neat idea looking at nvosdk to handle fifo context switch
11:39 < marcheu> toresbe: we use some tools to look at what's going on from the driver to the card. not directly on the bus though
11:40 < marcheu> toresbe: there http://nouveau.cvs.sourceforge.net/nouveau/ ; renouveau/ looks at the card fifo contents, and mmtrace/ traps register writes
11:40 < marcheu> toresbe: about donations, I think the simplest way would be to pick some nvidia card on ebay and get it delivered to a developer of your choice
13:13 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
13:25 -!- Duke` [n=gnu@ANantes-251-1-81-125.w86-203.abo.wanadoo.fr] has joined #nouveau
13:49 -!- ajmitch [n=ajmitch@ubuntu/member/ajmitch] has joined #nouveau
13:53 < ajmitch> evening
14:04 < Colbert> toresbe> we have a program (renouveau) that dumps the registers and fifo's of the card
14:04 < Colbert> andalso sends different openGL commands to it,
14:05 -!- EdB [n=EdB@ARennes-251-1-4-92.w83-195.abo.wanadoo.fr] has joined #nouveau
14:05 < Colbert> basicly, the driver is a black box, we change 1 thing with the opengl commands
14:06 < Colbert> then try to figure out what opcode changed.
14:08 < marcheu> ajmitch: morning :)
14:08 < Colbert> I'm pretty new here myself , marcheu, lumag, darktama, pmdata, are the gurus
14:08 < Colbert> morning
14:08 < ajmitch> seems interesting
14:08  * ajmitch only has a couple of cards to play with
14:09 < Colbert> need coffee brb
14:09 < marcheu> ajmitch: can you write C code ?
14:10 < ajmitch> certainly
14:10 < marcheu> ok, then you know enough
14:10 < ajmitch> heh
14:11 < marcheu> what card(s) do you have ?
14:11 < ajmitch> gf2 mx
14:11 < ajmitch> geforce 6600
14:11 < marcheu> cool
14:11 < marcheu> so to get started, I usually advise to grab renouveau from our cvs
14:11 < marcheu> that tool tracks changes in the 3D fifo
14:11 < ajmitch> the geforce2 is in a box that may or may not work - I have to revive it
14:12 < marcheu> and try to figure out some 3D commands with it
14:12 < ajmitch> yes, fetched that, eventually managed to build it
14:12 < marcheu> heh, if you had issues, you can report them here
14:12 < ajmitch> development branch of ubuntu seems to have changed some linux headers
14:12 < marcheu> like ?
14:12 < marcheu> I think it builds on dapper x86_64, at least last time I tried
14:12 < ajmitch> like linux/unistd.h - ran into problems with no _syscall3
14:12 < ajmitch> yes, I built it in a dapper chroot
14:13 < ajmitch> got the renouveau output from the 6600
14:14 < marcheu> so, raw output is not interesting in itself, what we need is figuring out some functionality out of it
14:14 < marcheu> that's where knowing C is useful
14:15 < marcheu> and a little OpenGL as well
14:15 < ajmitch> yes, I've used OpenGL a little
14:15 < marcheu> write a test for a given functionality, make small changes to your test, deduce things like "that is the clear color", go back to step 1
14:16 < ajmitch> & document everything
14:16 < marcheu> yes
14:16 < marcheu> since you have a nv43, you'll probably be interested in the NV30_TCL_PRIMITIVE_3D stuff
14:16 < marcheu> things we know are in nv_objects.h
14:17 < marcheu> NV30_TCL_PRIMITIVE_3D applies on NV40 & G70 btw, so don't be afraid
14:18 < ajmitch> alright
14:19 < ajmitch> I see a lot also in objects.c
14:19 < marcheu> yes, commands from nv_objects.h are grouped into objects
14:19 < ajmitch> that's what is being worked out?
14:20 < marcheu> an object represents some higher level functionality, like "draw a triangle" or "copy a surface"
14:20 < marcheu> that's nvidia's way of exposing hardware functionality
14:22 < ajmitch> seems to be a reasonable amount worked out so far
14:22 < marcheu> maybe you could look at clears on nv40, I don't think we have that yet
14:22 < ajmitch> I'll see what I can do
14:22 < marcheu> IIRC there's a couple of unknown commands with clears
14:23 < marcheu> anyway, if you've got a better idea...
14:23 < ajmitch> mostly adding tests to tests.c?
14:24 < marcheu> I don't think we'll need your gf2mx to much btw, pmdata has been hammering his own nv10 for way too long
14:24 < marcheu> yeah, if you've got interesting tests you can add them
14:24 < ajmitch> good, it may be a challenge to get that box working again
14:24 < EdB> marcheu, did you make some progress on the bios front ?
14:25 < marcheu> EdB: nope, I'm porting a memory manager to the drm right now. I let ddl pick up the ball :)
14:25 < EdB> :o)
14:25 < marcheu> well, he's the one from whom the bios code originates, so I think he knows better anyway
14:31 < Colbert> brb
14:31 -!- Colbert [n=colby@adsl-69-211-9-63.dsl.chcgil.ameritech.net] has quit ["using sirc version 2.211+KSIRC/1.3.12"]
14:52 -!- Colbert [n=colby@adsl-69-211-9-63.dsl.chcgil.ameritech.net] has joined #nouveau
14:59 -!- EdB [n=EdB@ARennes-251-1-4-92.w83-195.abo.wanadoo.fr] has quit ["Konversation terminated!"]
15:31 -!- taavi_r [n=k0htlane@84-50-161-35-dsl.rgu.estpak.ee] has joined #nouveau
15:45 -!- Duke` [n=gnu@ANantes-251-1-81-125.w86-203.abo.wanadoo.fr] has quit [Read error: 60 (Operation timed out)]
15:54 -!- Duke` [n=gnu@ANantes-251-1-101-145.w86-203.abo.wanadoo.fr] has joined #nouveau
16:24 < Colbert> hmm, the tests I wrote wont work now with the regl stuff, need to look at it.
16:25 < Colbert> why change to all sdl calls? is ther an advantage?
16:38 -!- taavi_r [n=k0htlane@84-50-161-35-dsl.rgu.estpak.ee] has quit ["Leaving"]
16:39 -!- EdB|w [n=EdB@212.234.68.206] has joined #nouveau
16:46 < Colbert> Need to review  structures. Well I'm doing alright for a guy who couldnt even speak 3 years ago.lol
16:47 < ajmitch> heh :)
17:05 < darktama> marcheu: what was Lumag's idea about handling the context switch?
17:06 < darktama> I'm fairly certain now that the hardware can do the context switching automatically, but I have no details on it yet.. but, until we figure it out we could do it manually
17:09 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has quit ["Leaving"]
17:14 -!- shaAway [n=Sha@stern.ulb.ac.be] has quit [Read error: 145 (Connection timed out)]
17:16 -!- ajmitch [n=ajmitch@ubuntu/member/ajmitch] has quit [Read error: 104 (Connection reset by peer)]
17:29 -!- shavengerAway is now known as __sha__
17:39 -!- ajmitch [n=ajmitch@ubuntu/member/ajmitch] has joined #nouveau
17:54 -!- __sha__ is now known as shavengerAway
19:25 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
19:51 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has quit ["Au revoir."]
20:03 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
20:21 < Colbert> pmdata> when you read this. did you try softquadro with the 39xx or 41 xx series drivers?
20:21 < Colbert> i think nvidia didnt add extra checks and only checked the pciid back then
20:26 < Colbert> I would be interested in dumps from left right front and back  on softquadro, (need to finish rewriting my test first)
20:28 < Colbert> Also, how should i add GLU stuff back in, I wasactually using the error function from libGLU
20:52 -!- pmdata [i=patrice@ANantes-154-1-2-52.w81-53.abo.wanadoo.fr] has joined #nouveau
20:53 < Colbert> hey
20:53 < pmdata> hello
20:54 < Colbert> did you try softquadro when you had the 3xxx driver?
20:59 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has quit [Read error: 110 (Connection timed out)]
21:04 < Colbert> pmdata> i could start pasting dumps to if you want them
21:05  * Colbert is away: Colbert brb
21:07 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
21:11  * Colbert is back
21:20 < Colbert> ARRRRG!!! I still cannot get a hold of my friend my email is through. and his voicemail cuts me off mid sentance.
21:21 < pmdata> no I did not try with 3xxx, I tried with 4xxx without positive result
21:21 < Colbert> darn
21:21 < Colbert> was hoping to see stereo buffers
21:21 < pmdata> I noticed the pci id is not changed
21:21 < pmdata> should it?
21:22 < Colbert> yes, it did with mine at least
21:22 < Colbert> are you using the nvclock from linuxhardware?
21:24 < Colbert> usage is nvclock -Q n where n= 1 -16 (it dosnt understand hex)
21:25 < pmdata> I got mine from nouveau cvs
21:27 < pmdata> ow, khronos updated opengl spec to 2.1, just after having taken control of it
21:30 < Colbert> http://nvclock.cvs.sourceforge.net/nvclock/
21:30 < Colbert> thunderbird added and modified our changes
21:31 < Colbert> works a little different though, so dont overwrite the nouveau nvclock
21:35 < pmdata> I just commited some changes in renouveau
21:36 < Colbert> I had a little trouble getting the autoconf junk to work, dont use the script, and call the highest installed aclocal autoheader and autoconf
21:36 < Colbert> at least on debian unstable
21:37 < pmdata> I read nv11 dump with the multi-file output, this is strange that most of them are empty
21:40 < Colbert> hmmm, didnt notice that some of mine are empty until now.
21:47 < Colbert> havnt been paying attention to the news, is the khronos group expected to be OSS and Free software friendly?
21:48 < pmdata> I think the driver 'optimize' rendering by not changing the hw state when not needed
21:48 < pmdata> this is why you must output a single triangle after each state change
21:49 < pmdata> about khronos, at least no more or less that arb from my pov
22:06 < Colbert> pmdata> i dont think the cvs version of nvclock has the softquadro files.
22:07 < pmdata> I remember marcheu sent me a patch
22:09 < pmdata> there may be a bug with the redirection to multifiles, nothing is outputted
22:10 < Colbert> was looking through the log to find marcheu's page for it. (nvclock)
22:10 < Colbert> http://icps.u-strasbg.fr/~marchesin/nvdri/nvclock-softquadro.tgz
22:11 -!- qfire [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has joined #nouveau
22:11 < Colbert> in the main branch of nvclock one can set the pciid explicitly
22:12 < pmdata> I think there is no output in some cases because some test function don't disable stuff they test
22:17 < Colbert> yea, somthings not right here
22:17 < Colbert> didnt look closely enough (i.e. at all)
22:21 -!- K [n=hazel@212.145.81.31] has joined #nouveau
22:22 < K> i cant compile renouveau
22:22 < pmdata> using vertex program crash 6106 :)
22:22 < K> gcc 4.0.4
22:23 < pmdata> what error?
22:23 < EdB|w> K give the log
22:23 < K> wait..
22:23 < K> re.c:505: warning: format '%lx' expects type 'long unsigned int', but argument 3 has type 'unsigned int *'
22:23 < K> twice
22:23 < K> its a printf
22:23 < K> but there are more errores
22:23 < K> errors
22:23 < K> sorry for my spanglish ;)
22:24 < pmdata> this is a warning, not an error
22:24 < K> oh
22:24 < K> regl.h:474: error: syntax error before 'GLprogramcallbackMESA'
22:24 < K> sorry, this is the error
22:24 < pmdata> ah, better
22:25 < EdB|w> K, it's renouveau ? or card driver ?
22:25 < K> renouveau
22:26 < Colbert> what does it print before "regl.h:474: error: syntax error before 'GLprogramcallbackMESA'"
22:27 < EdB|w> K, can you paste all the low on a web site like pastebin or rafb ?
22:27 < pmdata> fixed, checkout cvs
22:27 < EdB|w> so ...
22:27 < Colbert> nevermind
22:27 < K> hehe
22:27 < pmdata> I was using GL/gl.h from mesa, where this is defined
22:27 < K> ok
22:27 < pmdata> colbert> can you checkout, and remake some dumps?
22:28 < Colbert> sure
22:28 < Colbert> hold on need to remove my code,(dosnt compile yet)
22:29 < K> error continues..
22:30 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has joined #nouveau
22:31 < K> now fixed
22:31 < pmdata> I never noticed the 'object creation:' line at start also has the object type, time to check for unknown ones
22:31  * K feels on the edge
22:33 < EdB|w> K you are no alone
22:33 < Colbert> lol
22:34 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
22:35 < Colbert> ok pmdata what do you need run?  multifile?
22:36 < pmdata> yes, try that
22:37 < Colbert> whats test_default? that's new.
22:38 < pmdata> I added it to check how the default state of opengl context is done when simply rendering a triangle
22:38 < Colbert> should I uncomment all the test cases still
22:38 < K> how can i see the Gl funtions that are reveverse enginered?
22:38 < pmdata> colbert> you can try all
22:38 < pmdata> K> the ones still listed as [0xnumber/4], not by a name
22:39 < pmdata> what is your hw?
22:39 < K> mine?
22:39 < pmdata> yes
22:39 < K> 01:00.0 VGA compatible controller: nVidia Corporation NV4 [RIVA TNT] (rev 04)
22:39 < K> 10de:0020 (rev 04)
22:40 < pmdata> do you have some unknown command listed when trying a test?
22:40 < K> i dont know any command ;)
22:41 < K> but im trying to learn something
22:43 < pmdata> can you give a link to a dump?
22:44 < K> what commands would you like see in the dump?
22:45 < K> (this is because of my hw?)
22:45 < pmdata> some that are unknown
22:45 < K> ok
22:45 < K> wait
22:51 < Colbert> pmdata> do you want whole dumps? texture_cube_map seems odd btw
22:52 < marcheu> pmdata: did you see my dumps ?
22:52 < marcheu> (and the brand new one-file-per-dump functionality :)
22:53 -!- atcl [n=Miranda@Lb83f.l.pppool.de] has joined #nouveau
22:53 < K> pmdata: sorry, i must to go (helloween concert!)
22:53 < K> but
22:54 < K> hurry
22:54 -!- EdB|w [n=EdB@212.234.68.206] has quit ["Parti"]
22:54 < marcheu> wow, there's a lot of people in there, what happened ?
22:54 < K> pmdata: how can i send it?
22:54 < K> DCC
22:54 < K> email
22:54 < K> hurry up!
22:54 < marcheu> K: what do you need ?
22:55 < K> send a file to pmdata
22:55 < K> NOW
22:55 < marcheu> send it to me, I'll put it online for him
22:55 < K> ok
22:55 < K> email
22:55 < K> DCC
22:55 < marcheu> marchesin@icps.u-strasbg.fr
22:55 < K> ok
22:55 < marcheu> or DCC if you like
22:56 < Colbert> marchieu> hit the limit in here earier
22:56 < marcheu> K: I don't think I can dcc your file
22:57 < marcheu> K: email is ok
22:57 < marcheu> Colbert: limit ?
22:57 < Colbert> Extended nick list info turned off for #nouveau, too many people in channel
22:57 < Colbert> lol
22:58 < marcheu> hehe
22:58 < K> marcheu: done
22:58 < K> bye
22:58 -!- K [n=hazel@212.145.81.31] has quit [Remote closed the connection]
22:59 < marcheu> pmdata: http://icps.u-strasbg.fr/~marchesin/nvdri/tests/10de.0020
22:59 < Colbert> card_10de-0171_test_alpha.txt
23:00 < Colbert> http://rafb.net/paste/results/CHa8uz82.html
23:03 < Colbert> card_10de-0171_test_arb_texture_cube_map.txt
23:03 < Colbert> http://rafb.net/paste/results/04z4IJ64.html
23:04 < marcheu> Colbert: maybe you could do a tarball of all your files, mail it to me and I can put it online with the others
23:04 < marcheu> instead of pasting 1000 different tests
23:04 < marcheu> which is not really convenient
23:04 < marcheu> or maybe we can use the freedesktop.org or sf.net webspace for that
23:04 < Colbert> card_10de-0171_test_array_client.txt
23:04 < Colbert> http://rafb.net/paste/results/8M3zeh90.html
23:06 < Colbert> marchieu> tried to get my mail fixed at local lunchtime. still broken.(its through a friends webpage)
23:06 < marcheu> Colbert: do you have a sf.net account ?
23:06 < marcheu> Colbert: we could add you, and you'd upload card dumps on the sf.net page
23:06 < Colbert> i was going to try to get ahold of him this evening
23:07 < Colbert> no i dont yet.
23:07 < marcheu> everyone has a sf.net account, so we could put up all dumps at the same place, which would be convenient
23:08 < Colbert> yes, would be easier.
23:11 < marcheu> darktama: the nvosdk has code in the interrupt handler which seems to switch the fifo
23:12 < marcheu> Colbert: so, ping me when you have a sf.net account
23:13 < Colbert> ok marchieu 
23:13 < marcheu> btw there's no "i" in my nick :)
23:14 < Colbert> sorry
23:14 < marcheu> np, it's just that this won't light my client if you do that
23:18 < Colbert> probably should have used a email other than the one i cant get at right now,
23:18 < Colbert> any suggestions on good free email?
23:19 < marcheu> gmail, but you need another working email to get an account
23:21 < pmdata> strange, 0x308 (cull_face_enable on nv10), still set to 1 after a disable, something to investigate
23:23 < pmdata> hum, 308 may be something else?
23:26 < pmdata> na it's ok, the function does Enable() whereas the printf() say disable
23:27 < marcheu> hehe :)
23:46 < pmdata> aah, 13fc,1400 used for vbo, still unknown!
23:53 < pmdata> 13fc is begin_end_data (type of object to draw)
--- Log closed Чтв Авг 03 00:00:04 2006