--- Log opened Срд Июл 26 00:00:58 2006
00:02 < pmdata> so the difference between gf2/quadro2 seems to be 2sided lighting, clip planes and hardware accelerated AA lines
00:02 < pmdata> I guess I found the 2 first ones :)
00:02 -!- sturmflut [n=sturmflu@2001:6f8:992:1337:213:d4ff:feaa:8a9f] has joined #nouveau
00:16 < pmdata> hi sturm, I made some commits using your nv28 dumps
00:20 < sturmflut> oh cool
00:22 < sturmflut> At least that output can be useful to you, I didn't understand anything
00:32 < pmdata> I did not have time to look at nv34 dumps atm
00:34 < pmdata> sturm> do you want an example I explain to you from your nv28 dumps?
00:35 < sturmflut> sure
00:35 < pmdata> then grab the nv28_test_color_mask dump
00:36 < pmdata> the test_color_mask changes 4 values, which can be GL_FALSE, or GL_TRUE, hence 16 different combinations
00:37 < pmdata> so you must find something in your dump that changes 16 times (maybe 17 with the default value when creating the opengl context)
00:37 < pmdata> the rest should remain constant throughout the entire dump
00:38 < sturmflut> okay
00:38 < sturmflut> I'll have a look
00:38 < pmdata> this is something still marked NV20_TCL_PRIMITIVE_3D, because the color mask was not found at this time
00:38 < pmdata> if you want to cheat, look at renouveau cvs commits :)
00:38 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has joined #nouveau
00:38 < pmdata> hi lumag
00:39 < Lumag> hi all!
00:40 < sturmflut> So all those NV20_TCL_PRIMITIVE_3D entries mean "unknown"?
00:40 -!- Anne_ is now known as EdB
00:40 < pmdata> yes
00:40 < pmdata> except for arrays, where sometimes only the first is marked
00:41 < pmdata> for example, 4*4 matrix have 16 values, only the first is named
00:42 < sturmflut> Now it makes much more sense, I tought the NVIDIA guys are simply insane
00:44 < pmdata> once you get it, it is easy to read
00:48 -!- jkolb [n=jkolb@wsi-128-16.wsi.com] has quit ["Leaving"]
00:52 < sturmflut> ah
00:53 < sturmflut> I think I may have got the point
00:54 < sturmflut> Is there an unknown object I can search for in my dumps?
00:57 < pmdata> objects are more tricky stuff, they are 'class' of functions in the gpu
00:57 < pmdata> anyway, did you find the color mask?
00:58 < pmdata> http://nouveau.freedesktop.org/wiki/NvObjectTypes for known object types
01:00 < Lumag> or semi-known :)
01:02 < pmdata> yep
01:02 < pmdata> at least we get a better picture of the chip
01:04 < pmdata> the hard part: every command/function we found in objects will have to be tested
01:07 < sturmflut> I cheated a little and found the 18 entries marked with [0x0358/4] and the different values after them. So 0x358 probably means glColorMask() and the values represent the different combinations of the four parameter values
01:12 < sturmflut> Hm what is a "channel"? It seems to be mostly set to 0x1 but I also see 0x0 and 0x5 sometimes
01:12 < pmdata> \o/
01:13 < pmdata> what channel are you refering about?
01:13 < sturmflut> In the fifo "decode" (du you call that "decode"?)
01:13 < sturmflut> 9da   0x00000000   0x0000bee0             {size: 0x0   channel: 0x5   obj: beef5f01 opcode: METHOD }
01:16 < pmdata> I can't answer you, I don't know
01:17 < sturmflut> channel 5 seems to execute NOP commands all the time
01:17 < sturmflut> Funny
01:17 < pmdata> I leave this to people who made renouveau
01:17 < sturmflut> Okay
01:19 < pmdata> so now, you can try to find more stuff!
01:20 < marcheu> there are 8 channels
01:20 < marcheu> each channel can have a different object bound to it
01:20 < pmdata> marcheu> the more I look at the lighting stuff on my nv15, the more I see 2sided lighting
01:21 < marcheu> pmdata: yeah, that's possible. I think it's something nvidia had for a long time
01:24 < sturmflut> pmdata: Okay, where to start?
01:25 < pmdata> marcheu> will be hard to find without a real quadro2 (or yours softquadro2-ed?)
01:26 < pmdata> sturm> grab renouveau cvs, make some dumps, try to find missing commands
01:26 < pmdata> marcheu> try to enable 2sided lighting, and change the color material (set it different for front and back)
01:29 < pmdata> I have another idea, I can make a test function, and try on nv20 or higher to find same values in commands
01:32 < pmdata> marcheu> what is your current hw?
01:32 < marcheu> nv28
01:33 < pmdata> ok
01:39 < pmdata> marcheu> can you grab renouveau cvs, and make me a dump for test_light_two_sides?
01:40 < sturmflut> How do I know that e.g. glColorMask() is NV20_TCL_PRIMITIVE_3D_COLOR_MASK
01:40 < marcheu> http://icps.u-strasbg.fr/~marchesin/nvdri/nv28_light2side.txt
01:41 < pmdata> sturm> (1) it is the only thing that is changed when calling this function? (2) obiwan kenobi?
01:43 < sturmflut> hehe no that's not what I meant
01:44 < sturmflut> Let
01:45 < sturmflut> I mean how do I know if a dump for a given test on a given card has already been analyzed? Is there a list of missing commands?
01:46 < marcheu> if it's recognized in renouveau, it's done
01:46 < marcheu> otherwise, it's not done
01:46 < marcheu> pretty easy, huh ?
01:51 < pmdata> marcheu> with your dump I found same values on my gf2, but the rendering is done in 2 passes, one for front, one for back
01:52 < pmdata> it would be nice to test this function on a quadro2 or your nv15 modded
01:57 < sturmflut> I think 0x00001818 is NV20_TCL_PRIMITIVE_3D_VERTEX_DATA, not 0x40001818. The command type is only 12 bits long so I think 0x40001818 can't be right
01:59 < pmdata> yep, I fixed this today for nv10, time to do it for nv20 and nv30
02:00 < sturmflut> It's time for me to go to sleep
02:00 < sturmflut> cu
02:00 -!- sturmflut [n=sturmflu@2001:6f8:992:1337:213:d4ff:feaa:8a9f] has quit [Remote closed the connection]
02:01  * pmdata also dodo
02:01 -!- pmdata [i=patrice@ANantes-154-1-21-66.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
02:18 -!- Duke` [n=gnu@ANantes-251-1-93-215.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
02:22 -!- EdB [n=EdB@ARennes-251-1-33-96.w81-250.abo.wanadoo.fr] has quit [Read error: 110 (Connection timed out)]
02:33 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Remote closed the connection]
03:32 -!- etzel [n=thisnuke@69.160.140.26] has quit [Remote closed the connection]
03:55 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has left #nouveau []
05:17 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau
05:33 -!- tibbs [n=tibbs@fedora/tibbs] has joined #nouveau
08:42 -!- tibbs is now known as tibbs|h
13:09 -!- Duke` [n=gnu@ANantes-251-1-103-178.w86-203.abo.wanadoo.fr] has joined #nouveau
14:00 -!- EdB [n=EdB@ARennes-251-1-152-142.w86-214.abo.wanadoo.fr] has joined #nouveau
14:04 -!- ChanServ [ChanServ@services.] has quit [Shutting Down]
14:04 -!- EdB [n=EdB@ARennes-251-1-152-142.w86-214.abo.wanadoo.fr] has quit [Client Quit]
14:11 -!- ChanServ [ChanServ@services.] has joined #nouveau
14:11 -!- ServerMode/#nouveau [+o ChanServ] by irc.freenode.net
15:08 -!- shenki [n=shenki@ppp153-94.lns3.adl2.internode.on.net] has joined #nouveau
16:01 -!- EdB|w [n=EdB@212.234.68.206] has joined #nouveau
17:23 -!- shenki [n=shenki@ppp153-94.lns3.adl2.internode.on.net] has quit ["Leaving"]
17:29 -!- jkolb [n=jkolb@wsi-128-184.wsi.com] has joined #nouveau
18:36 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has quit ["Leaving"]
19:47 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
19:51 -!- jkolb [n=jkolb@wsi-128-184.wsi.com] has left #nouveau ["Leaving"]
20:30 -!- Duke` [n=gnu@ANantes-251-1-103-178.w86-203.abo.wanadoo.fr] has quit [Read error: 110 (Connection timed out)]
20:31 -!- Duke` [n=gnu@ANantes-251-1-155-39.w86-203.abo.wanadoo.fr] has joined #nouveau
21:14 -!- `Duke` [n=gnu@ANantes-251-1-154-62.w86-203.abo.wanadoo.fr] has joined #nouveau
21:15 -!- qfire_away [n=qfire@dsl093-010-213.cle1.dsl.speakeasy.net] has quit [Read error: 104 (Connection reset by peer)]
21:18 -!- johill [n=johannes@p5487DE20.dip.t-dialin.net] has joined #nouveau
21:24 -!- Duke` [n=gnu@ANantes-251-1-155-39.w86-203.abo.wanadoo.fr] has quit [Success]
21:50 -!- pmdata [i=patrice@ANantes-154-1-27-43.w81-53.abo.wanadoo.fr] has joined #nouveau
22:18 < pmdata> hello
22:20 < pmdata> marcheu> if you plug your nv15 in softquadro mode, I need the test_color_material and test_light_two_side dumps
22:36 -!- philv is now known as ralf0
22:37 -!- ralf0 is now known as philv
22:38 < pmdata> schizophrene?
22:39 < philv> Naw, a joke between Ralf Baechle and I.
23:04 -!- EdB|w [n=EdB@212.234.68.206] has quit ["Parti"]
23:36 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has joined #nouveau
23:37 < Lumag> hi
23:38 < pmdata> hi lumag
--- Log closed Чтв Июл 27 00:00:59 2006
