--- Log opened Вск Авг 06 00:00:07 2006
00:07 -!- KoalaBR [n=KoalaBR@port-83-236-15-12.dynamic.qsc.de] has joined #nouveau
00:16 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has joined #nouveau
00:17 -!- shenki [n=shenki@ppp167-129.lns3.adl4.internode.on.net] has quit [Read error: 110 (Connection timed out)]
00:17 < KoalaBR> Hi
00:18 < KoalaBR> Lumag: I need some help / ideas :)
00:23 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
00:25 < pmdata> ping marcheu
00:26 < KoalaBR> Hi
00:28 < KoalaBR> marcheu: Some things done, some ideas needed
00:31 < KoalaBR> pmdata; Perhaps you could give me some advice?
00:31 < pmdata> koala> what do you need?
00:32 < KoalaBR> Well I am looking at NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS
00:32 < KoalaBR> on a NV43
00:33 < pmdata> then maybe it is the same as for nv10 and nv20
00:33 < KoalaBR> I have added decoding of that command and added a new one : NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMSY as NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS set X dimension and NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMSY set Y dim
00:33 < KoalaBR> so you get now:
00:33 < KoalaBR> 50e   0x00000000   0x0100007c   NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS           = Width  =0x0100  offset =0x007c
00:33 < KoalaBR> 50f   0x00000000   0x00ff0102   NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMSY          = Height =0x00ff  offset =0x0102
00:34 < KoalaBR> Width and Height work correctly, but the offset gives me headaches
00:35 < KoalaBR> 1. Question: Is there interest in the modification? If so, what shall I do?
00:39 < pmdata> if you don't have access to cvs, I can do that
00:39 < pmdata> are you sure about the 'offset' ?
00:39 < KoalaBR> Yes, that leads to next question
00:40 < KoalaBR> The offset depends on the position of the rendering info on the screen
00:40 < pmdata> something like the window position?
00:40 < KoalaBR> Not really, wait I need to paste something
00:41 < pmdata> for nv10 I put offsets as x and y values for viewport (x,y in the color buffer)
00:42 < KoalaBR> Look here: http://rafb.net/paste/results/adZQbi58.html 
00:42 < KoalaBR> I have added the position of the Window for information at the top by hand
00:43 < KoalaBR> See that 2 primitives? They always are at the end of the initsequence before NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS  comes
00:45 < KoalaBR> and you can see the primitive is somewhat dependent on the offset. And the Y-offset is 0x101 + (Y-Pos window % 3)
00:45 < KoalaBR> 8c0 seems to influence X offset and 8c4 influences Y
00:46 < KoalaBR> I have not the faintest idea, what that means
00:48 < KoalaBR> Hm, the mod 3 is wrong
00:49 < KoalaBR> but there is a dependency
00:49 < KoalaBR> will have more tests
00:50 < KoalaBR> pmdata: I told nonsense above: the offset depends on the rendering window on the screen (not the info)
00:52 < marcheu> pmdata: ok, I'll do the texture stuff :)
00:53 < pmdata> koala> then I will put it as x,y position in the color buffer
00:55 < pmdata> koala> try in fullscreen to set if it is set to 0 (change the width,height of screen to your desktop size)
00:57 < KoalaBR> Ok
00:59 < marcheu> hmm renouveau is broken ?
01:00 < KoalaBR> 4ee   0x00000000   0x0fff0010            NV30_TCL_PRIMITIVE_3D      [0x08c0/4] = 0x0fff0010 | UNKNOWN = 0fff0010
01:00 < KoalaBR> 4ef   0x00000000   0x02040000            NV30_TCL_PRIMITIVE_3D      [0x08c4/4] = 0x02040000 | UNKNOWN = 02040000
01:02 < pmdata> marcheu> oops :)
01:02 < pmdata> fixed
01:02 < marcheu> pmdata: if you at least know what it is, that's good :)
01:02 < marcheu> nope, I was thinkg about another one
01:03 < KoalaBR> Well too tired to think about it know. Will come back tomorrow. 
01:04 < marcheu> ah, cool
01:04 < KoalaBR> What about my changes? If I do a patch, this will include quite a bunch of doxygen docs in the source too
01:05 < pmdata> koala> don't know, up to marcheu to decide
01:06  * pmdata in hot potato mode
01:07 < KoalaBR> I don't press for CVS access btw. Don't worry ;)
01:07 < Lumag> koala: you can separate patches :)
01:07 < Lumag> it's always better to provide one patch per feature/fix/etc.
01:07 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
01:07 < KoalaBR> well it is a patch to object.c (one more entry and 2 print functions)
01:08 < KoalaBR> I could offer some doxygen docs, I added to the source too
01:08 < KoalaBR> I did that in order when I tried to understand renouveau
01:09 < KoalaBR> I will do a patch without any docs against current renouveau
01:10 < marcheu> pmdata: the nv20 texture filter mode is 1b54
01:10 < marcheu> pmdata: what do you think there is in 1b14 ?
01:11 < pmdata> the same for tx0
01:12 < pmdata> 1b54 is for tx1
01:12 < marcheu> ah right, the test function sets tex1
01:15 < KoalaBR> Well. I have problems, patch doesn't apply correctly. Possibly one of the new changes in renouveau
01:15 < KoalaBR> I will have look at it and offer it tomorrow - too late for today
01:15 < KoalaBR> Bye
01:16 -!- KoalaBR [n=KoalaBR@port-83-236-15-12.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"]
01:20 < marcheu> any idea about 1b24 ?
01:21 < pmdata> give me access to a dump  :)
01:21 < pmdata> the texture_filter one please
01:22 < marcheu> it's always 0
01:22 < marcheu> thus my question :)
01:23  * pmdata need a dump to see what is around
01:24 < marcheu> http://nouveau.sourceforge.net/tests/nv28/card_10de-0281_test_texture_filter.txt
01:27 -!- EdB [n=EdB@ARennes-251-1-61-23.w81-53.abo.wanadoo.fr] has quit ["Konversation terminated!"]
01:27 < pmdata> if it is set to 0, maybe it is enable/disable something for a texture unit
01:32 < pmdata> there is also 1b08+0x40*n
01:35 < pmdata> ok, added tx filter to cvs
01:35 < marcheu> hey, I was doing it !
01:35 < pmdata> now I have nearly same infos as for nv10, except missing stuff in format (rgba, cubemap flag, npot flag)
01:35 < pmdata> I can let that to you
01:36 < marcheu> duplicating effort like that is just a waste of time
01:36 < pmdata> then, don't do dumps for me :)
01:36 < marcheu> (but actually you have the bits wrong)
01:36 < pmdata> I will let you nv20 tcl then
01:36 < marcheu> yeah, maybe that's better
01:37 < marcheu> so that I can test it
01:37  * pmdata need to figure the rest of nv10 tcl (quadro stuff also)
01:38 < pmdata> are you sure about the bits?
01:39 < marcheu> yeah, 0x01012000 means 01 01 2000 which is nearest/nearest/2000
01:39 < pmdata> yes, mag/min filter
01:40 < pmdata> oops, I understand
01:40 < marcheu> anyway, I have to sleep
01:40 < marcheu> I'll let you fix that
01:40 < pmdata> so I must fix that myself?
01:40 < marcheu> but really, this shows we need to test stuff more toroughly
01:41 < pmdata> yeah, time to outsource to india
01:43 < pmdata> marcheu> you should also use '-f' command line to output unknown stuff as float
02:00 -!- pmdata [i=patrice@ANantes-154-1-75-246.w86-199.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
02:12 -!- Duke` [n=gnu@ANantes-251-1-136-224.w86-210.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
02:34 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has joined #nouveau
03:04 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has quit ["Au revoir."]
03:11 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has joined #nouveau
03:43 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has left #nouveau []
03:46 -!- Myrizio [n=Myrizio@host141-96.pool80104.interbusiness.it] has quit ["Leaving"]
03:58 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Remote closed the connection]
05:43 -!- K [i=hazel@tor/session/external/x-461dc742c429bc1d] has quit [Remote closed the connection]
05:55 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has quit ["Au revoir."]
09:15 -!- swany_ [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
09:23 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit [Read error: 60 (Operation timed out)]
11:22 -!- Myrizio [n=Myrizio@host40-101.pool80104.interbusiness.it] has joined #nouveau
12:18 -!- darktama [n=darktama@gentoo/contributor/darktama] has quit [Read error: 110 (Connection timed out)]
12:19 -!- darktama [n=darktama@124-168-238-39.dyn.iinet.net.au] has joined #nouveau
12:46 -!- KoalaBR [n=KoalaBR@port-83-236-15-214.dynamic.qsc.de] has joined #nouveau
12:46 < KoalaBR> Hello
12:47 < KoalaBR> marcheu: Good morning
12:47 < marcheu> hi KoalaBR 
12:48 < KoalaBR> marcheu: 2 things
12:48 < KoalaBR> marcheu: 1. I have a small patch to objects.c to better parse NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS
12:50 < KoalaBR> marcheu: 2. I have found something in the init phase, which corresponds to the viewport height and weight, but it is an unknown command and I have no idea what is does
12:50 < KoalaBR> Shall I paste the patch?
12:50 < marcheu> sure
12:50 < marcheu> what did you see about 2. ?
12:52 < KoalaBR> http://rafb.net/paste/results/UeiCs721.html   (it is my first patch ever to any project, please tell if I did something wrong)
12:53 < KoalaBR> marcheu: regarding 2: I will need to paste a summary I did, please wait a few seconds
12:53 -!- Duke` [n=gnu@ANantes-251-1-100-252.w86-203.abo.wanadoo.fr] has joined #nouveau
12:56 < KoalaBR> marcheu: Have a look here: http://rafb.net/paste/results/1mAmPm38.html
12:57 < KoalaBR> marcheu: You will the the 8c0 / 8c4 command exactly two times within the initsequence
12:59 < KoalaBR> the first value 8c0: Highest 16 bit are always 0x0fff. The lowest 16 bit describe the offset of the rendering window. Please note, that it is always 3 higher than the value I set the SDL-window.
13:00 < marcheu> ok, 2 things about the patch :
13:00 < KoalaBR> this is the size of the window frame in pixels
13:00 < marcheu> - just remove commented out code
13:00 < KoalaBR> Ok
13:00 < marcheu> - don't hardcode values like {0x00000a04, please define them in nv_objects.h instead
13:01 < KoalaBR> Ok
13:01 < KoalaBR> Shall I redo the patch?
13:01 < marcheu> no, get a sf.net id instead
13:02 < KoalaBR> thanks for accepting :)
13:03 < KoalaBR> The X offset however is offset % 128 the highest value you'll see is 7f
13:04 < marcheu> yeah, I'm not sure we fully understand that window pos stuff yet
13:04 < KoalaBR> Now for 8c4
13:04 < marcheu> it looks like it's full of corner cases
13:04 < KoalaBR> Yes
13:05 < KoalaBR> for 8c4 only the highest 16 bit change
13:09 < KoalaBR> They represent the exactly the height of the viewport. If height viewport < Height SDL-Window then there is a value added (here 4 or 5)
13:10 < KoalaBR> I don't know why, my Windowdecoration is 27 Pixel high, would have expected that value
13:11 < KoalaBR> However, the values which you will at 8c0 (lowest 16 bit) and 8c4 (highest 16 bit) will reappear in the offset values of  NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS(Y)
13:11 < KoalaBR> That's all I can currently say
13:12 < KoalaBR> s/you wil/ you willl see/
13:13 < KoalaBR> marcheu: Do you have any idea regarding this?
13:15 < marcheu> I was thinking that maybe there is some kind of alignment
13:23 < KoalaBR> marcheu: If I remember correctly, my gfx memory is accessible via a 128 bit wide bus....
13:25 -!- EdB [n=EdB@ARennes-251-1-31-195.w81-250.abo.wanadoo.fr] has joined #nouveau
13:25 < KoalaBR> marcheu: That could explain the offset... 
13:29 < KoalaBR> Will be back later
13:29 < KoalaBR> bye
13:29 -!- KoalaBR [n=KoalaBR@port-83-236-15-214.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"]
13:53 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
14:53 -!- K [i=hazel@tor/session/external/x-cb0748e1d65ab4f0] has joined #nouveau
15:28 -!- atcl [n=Universe@La501.l.pppool.de] has joined #nouveau
15:32 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Read error: 104 (Connection reset by peer)]
15:33 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
15:34 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Read error: 104 (Connection reset by peer)]
15:35 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
15:41 -!- johill [n=johannes@p5487C4DD.dip.t-dialin.net] has joined #nouveau
15:49 -!- atcl [n=Universe@La501.l.pppool.de] has quit [Read error: 104 (Connection reset by peer)]
16:26 -!- johill [n=johannes@p5487C4DD.dip.t-dialin.net] has quit [Read error: 113 (No route to host)]
16:44 -!- atcl [n=Universe@La501.l.pppool.de] has joined #nouveau
16:47 -!- _stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
17:02 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Read error: 110 (Connection timed out)]
17:56 -!- KoalaBR_ [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has joined #nouveau
17:56 < KoalaBR_> Hi
17:58 < KoalaBR_> marcheu: I think, I found out the way the offset for NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS(Y) and 8c0 / 8c4 are calculated
18:21 < KoalaBR_> Anyone here?
18:22 -!- pmdata [n=patrice@ANantes-154-1-42-222.w81-53.abo.wanadoo.fr] has joined #nouveau
18:22 < KoalaBR_> Hi
18:33 < pmdata> hello
18:33 < KoalaBR_> Hi
18:33 < pmdata> hello
18:33 < darktama> ...
18:33 < darktama> :)
18:33 < pmdata> nothing interesting to say?
18:34 < darktama> just commenting on the.. "hi,hello,hi,hello"
18:34 < KoalaBR_> Well, I hope I figured out how the offsets for NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS(Y) and those 8c0 / 8c4 commands are calculated
18:34 < pmdata> darktama> :)
18:35 < KoalaBR_> Have a look here: http://rafb.net/paste/results/D8plEL69.html  some pseudo code
18:39 < KoalaBR_> I guess, the 8c4 /8c0 commands are related to buffer setup, as they only appear if I use :  	SDL_GL_SetAttribute(SDL_GL_DOUBLEBUFFER,0); if I omit this line, 8c0 / 8c4 are missing too
18:40 < KoalaBR_> does this qualify as "interesting"?  ;o)
18:42 -!- _stringfellow is now known as stringfellow
18:44 -!- EdB [n=EdB@ARennes-251-1-31-195.w81-250.abo.wanadoo.fr] has quit [Remote closed the connection]
18:44 < darktama> KoalaBR_: well, anything we don't currently know is interesting imo :P
18:44 -!- EdB [n=EdB@ARennes-251-1-31-195.w81-250.abo.wanadoo.fr] has joined #nouveau
18:45 < KoalaBR_> Darktama: How current is the matrix on the website? Too much yellow for my taste 
18:46 < darktama> for NV40 it seems fairly correct.. buffer clears we could do (with a flat 3D primitive), but I believe that refers to any "special" clearing methods that are available
18:47 -!- johill [n=johannes@p5487C4DD.dip.t-dialin.net] has joined #nouveau
18:47 < darktama> DMA, we know how to use MEMORY_TO_MEMORY_FORMAT.. but should probably learn how to use the other object nvidia uses to upload textures
18:48 < darktama> nvidia actually use both to upload textures depending on <insert reason here>
18:48 < KoalaBR_> Well, I can only offer help with the NV4x problems
18:49 < KoalaBR_> So, what do we do with this offset calculation?
18:49 < darktama> same, luckily most of it applies to NV30 and G70 aswell
18:49 < pmdata> I see m2mf used for compressed textures
18:50 -!- EdB [n=EdB@ARennes-251-1-31-195.w81-250.abo.wanadoo.fr] has quit [Remote closed the connection]
18:51 < darktama> I see either m2mf or the 0x72/0x89/0x9e combo.. usually the latter is done for 2D textures, but sometimes m2mf.. I haven't looked into it much though
18:51 -!- EdB [n=EdB@ARennes-251-1-31-195.w81-250.abo.wanadoo.fr] has joined #nouveau
18:52 < darktama> pmdata: I've seen you do a few NV30 additions, do you have NV30 hardware?
18:57 -!- atcl [n=Universe@La501.l.pppool.de] has quit [Read error: 110 (Connection timed out)]
18:58 < KoalaBR_> So, what do you suggest that I do? I have a small patch to objects.c and this information on how the offsets are calculated...
18:59 < darktama> did you end up getting a sf.net account?
18:59 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
18:59 < KoalaBR_> Ok, will do that
19:05 < KoalaBR_> Got it
19:06 < darktama> now, you'll need to see marcheu about getting cvs access.. if you get it, you can commit your changes
19:07 < KoalaBR_> Ok, will do that too
19:07 < KoalaBR_> But where do you put this calculation info? Into doc/ ?
19:08 < darktama> sure :) it hasn't seen much use lately
19:11 < KoalaBR_> So, how much work for reverse engineering is left, what do you estimate?
19:12 < darktama> there's quite a bit of work left to do before we can have a GL driver imo.. not so much on the FIFO command side, more the engine init and context handling stuff
19:14 < KoalaBR_> Well, I will go back and try something new, clipping planes perhaps...
19:16 < KoalaBR_> But perhaps you could give me short explanation?
19:17 < darktama> of?
19:18 < KoalaBR_> Well, I know how renouveau finds the mappings. but how do we find the object, the commands and the channels / subchannels and what use are they (the channels)
19:18 < KoalaBR_> Just a short overview
19:18 < KoalaBR_> Don't write a novell :)
19:19 < darktama> ok.. nvidia hardware has multiple FIFOs (channel).. each FIFO has a 8 "slots" (subchannels) to which you can assign objects
19:20 < pmdata> darktama> no, I just read some nv30 dumps, and compare against nv10 and nv20 ones
19:20 < darktama> a command in a DMA buffer says what subchannel the command is for, and the combination of the offset in the command and the object bound to the subchannel determines what the gpu deos
19:20 < darktama> does
19:21 < KoalaBR_> Ok, thanks :)
19:21 < darktama> pmdata: ah damn :) was hoping you were going to tackle NV30 vtx progs!
19:22 < KoalaBR_> See you later then ...
19:22 < pmdata> I was thinking about it, I read some nv30 dumps about it
19:22 < darktama> yeah, I should sleep soon :) not yet though
19:22 -!- KoalaBR_ [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"]
19:25 < darktama> I've been meaning to either get an NV30, or to take marcheu up on his offer of access to a machine with an NV30.. just haven't gotten around to either yet
19:25 < pmdata> I would like to finish nv10 tcl 
19:25 < darktama> how much is left?
19:25 < pmdata> mostly the lighting stuff, and quadro
19:26 < darktama> ah
19:26 < pmdata> for example, I would like to know how the 6 values used for shininess are calculated
19:26 < darktama> hmm, I'm going to have to make renouveau map the regs before it creates a GL context :(
19:28 < pmdata> couldn't we try to draw a triangle now?
19:29 < darktama> I've been trying on NV40.. theres something missing though
19:29 < darktama> I've basically implemented EXA with the 3D engine.. :)
19:30 < darktama> shame it doesn't work inside X itself..
20:25 -!- johill [n=johannes@p5487C4DD.dip.t-dialin.net] has quit ["Client exiting"]
21:38 -!- Myrizio [n=Myrizio@host40-101.pool80104.interbusiness.it] has quit [Read error: 110 (Connection timed out)]
21:40 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Remote closed the connection]
21:47 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
22:12 -!- KoalaBR_ [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has joined #nouveau
22:15 -!- Myrizio [n=Myrizio@host40-101.pool80104.interbusiness.it] has joined #nouveau
22:23 -!- pmdata [n=patrice@ANantes-154-1-42-222.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
22:24 < marcheu> KoalaBR_: what's your sf.net id ?
22:25 < KoalaBR_> koala_br
22:27 < KoalaBR_> Hi, I think I found out how the offsets for this unknown comand set (8c0 / 8c4) and the NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS are calculated
22:28 < KoalaBR_> marcheu: Have a look here http://rafb.net/paste/results/D8plEL69.html
22:33 < marcheu> quite complex. what's your card ?
22:34 < KoalaBR_> A NV43 / GForce 6600 GT
22:35 < KoalaBR_> The 8c0 & 8c4 only appear if you use  	SDL_GL_SetAttribute(SDL_GL_DOUBLEBUFFER,0);   Delete the line and both occurances of 8c0 / 8c4 are missing from the dump
22:35 < KoalaBR_> Perhaps has something to do with the buffers?
22:37 < KoalaBR_> BTW: The values on the right side are only some sample values for me. It helped me to verify the code
22:37 < marcheu> darktama: "make renouveau map the regs before it creates a GL context" why do you need that ? you'd rather use mmtrace then
22:38 < marcheu> KoalaBR_: yeah, I can imagine :)
22:38 < marcheu> KoalaBR_: did you try at another bpp ?
22:38 < marcheu> KoalaBR_: is is still %128 or does that turn into %64 ?
22:38 < marcheu> like, when switching to 16bpp instread
22:39 < marcheu> instead
22:39 < KoalaBR_> Need to check. Didn't think of that
22:39 < marcheu> I suspect 128x8 is a graphics tile, and 128 might be in bytes. or might not
22:40 < KoalaBR_> Ok, will need to look, at which bpp I run, sec
22:42 < KoalaBR_> marcheu: according to xdpyinfo   depth of root window:    24 planes
22:42 < marcheu> yeah, so you're running 32bpp
22:42 < marcheu> please try with 16
22:43 < marcheu> now that I think of it, running renouveau tests at 16bpp and figuring out the differences might be a good first project for starters
22:44 < KoalaBR_> Wait a second, I will restart my X Server with 16 bpp
22:45 < KoalaBR_> (my xorg says for current setup: DefaultDepth = 24
22:45 < KoalaBR_> Be right back
22:45 < marcheu> yeah, 24 bpp of color + 8 of alpha
22:45 < marcheu> makes 32
22:45 -!- KoalaBR_ [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has quit [Remote closed the connection]
22:47 -!- KoalaBR [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has joined #nouveau
22:48 < KoalaBR> Back, but I can change my bpp: NVidia fails to find configs for 16bpp or 32bpp (no usable Screens found)
22:49 < marcheu> hmm, look at your list of visuals
22:49 < marcheu> you probably don't have visuals with 16bit depth or 8 stencil when at 16bpp
22:49 < marcheu> but for most tests that doesn't matter
22:52 < KoalaBR> you are correct xdpyinfo shows only visuals for 24 bit - strange
22:53 < marcheu> hmm ? I'm talking about visuals from glxinfo
22:54 < KoalaBR> Oh sorry *blush*
22:55 < KoalaBR> Says the same: only 24bit visuals available
22:55 < marcheu> so you're still at 24bpp
22:56 < KoalaBR> Yes, I know. Every other option resulted in no screens with usable configuration
22:57 < marcheu> "DefaultColorDepth 24" ?
22:57 < marcheu> hmm
22:57 < marcheu> "DefaultColorDepth 16" ?
22:57 < marcheu> :)
22:57 < KoalaBR> I tried to change that
22:57 < KoalaBR> to 16 and 32 (two tries)
22:57 < marcheu> then you need a depth 16 subsection
22:57 < marcheu> with the modes
22:57 < KoalaBR> no success
22:58 < marcheu> paste your /etc/X11/xorg.conf somewhere
22:59 < KoalaBR> Sec
23:01 < KoalaBR> http://rafb.net/paste/results/ONIj0U16.html (note only Screen Section and subsection for 32 bits was missing when I tried...)
23:04 < marcheu> and when you replace 24 by 16 it doesn't work ? weird
23:05 < KoalaBR> Yes, that's how it went
23:06 < marcheu> you should put up your /var/log/Xorg.0.log then
23:06 < marcheu> the one where it fails
23:07 < KoalaBR> Sec
23:10 < KoalaBR> http://rafb.net/paste/results/ar9sf917.html   but I'm not sure whether I tried 16 bit or 32 here (I can retry any time)
23:10 -!- pmdata [i=patrice@ANantes-154-1-42-222.w81-53.abo.wanadoo.fr] has joined #nouveau
23:10 < marcheu> (EE) NVIDIA(0): Given color depth (32) is not supported
23:10 < marcheu> as far as Xorg is concerned, 32bpp doesn't exist
23:11 < marcheu> 16 should be however
23:11 < KoalaBR> Hmmm
23:11 < pmdata> marcheu> does 8 bits would work with nv driver?
23:11 < marcheu> pmdata: it used to, alghouth there's no 3D IIRC
23:11 < marcheu> although
23:11  * marcheu swaps letters
23:12 < KoalaBR> I will retry 16 bit
23:12 < pmdata> so no possibility to test opengl color index mode
23:12 < KoalaBR> Be right back
23:12 < marcheu> nope, not that it's important anyway
23:12 -!- KoalaBR [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has quit [Remote closed the connection]
23:12 < marcheu> color index mode is a remnant from the iris gl days, and was introduced to ease portability
23:12 < marcheu> no real application uses it today :)
23:15 < pmdata> ok
23:15 -!- KoalaBR [n=KoalaBR@port-83-236-12-209.dynamic.qsc.de] has joined #nouveau
23:15 < pmdata> I also see that there was a nv_evaluators extension, maybe there is some hw accel support in gpu?
23:15 < KoalaBR> Found the culprit: I had RandR enabled :(
23:16 < KoalaBR> running in 16 bit
23:16 < KoalaBR> Will retry the tests
23:16 < marcheu> KoalaBR: good, you're now officially responsible of figuring out differences at 16bpp :)
23:16 < marcheu> pmdata: I don't know, I don't think so
23:16  * pmdata will add a simple test for the texture format
23:17 < pmdata> I saw a document on nvidia site about the formats supported by each gpu
23:17 < marcheu> oh ?
23:20  * pmdata try to get back the url
23:21 < marcheu> http://icps.u-strasbg.fr/~marchesin/perso/extensions/NV/evaluators.html
23:21 < marcheu> looks like vertex attributes from the other nvidia extensions
23:22 < marcheu> I wonder if there's real hw support behind
23:22 < marcheu> and I wonder what interesting could be done with it
23:22 < marcheu> interesting things
23:24 < marcheu> I would say no hw support, given that the TNT has it
23:25 < KoalaBR> Well, regarding the offsets: Nothing seems to change for the Y offset calculation, but the x offset is now calculated with % 256
23:26 < marcheu> right, so that's probably a tile : it's memory size remains constant while you need more pixels to fill it
23:27 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Remote closed the connection]
23:28 < KoalaBR> So does this give us a hint for this 8c0 command?
23:29 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
23:31 < KoalaBR> marcheu: Need anything else? Otherwise I wanted to have a look at the clip planes (Just tell me, what you need)
23:31 < marcheu> commit what you know about viewport stuff first
23:31 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Remote closed the connection]
23:32 < marcheu> we have a doc/ module where you can be talkative
23:32 < KoalaBR> I will add a text to the doc module NV40_8c0_and_viewport_offsets.txt ok=
23:32 < KoalaBR> ?
23:32 < marcheu> yeah
23:33 < marcheu> maybe something like nv40_viewport if you want to follow our naming conventions
23:33 < marcheu> about clip planes, do you have 2D or 3D clip planes in mind ?
23:34 < marcheu> because we probably need a quadro (or softquadro) to figure out 2D clip planes
23:36 < marcheu> you could also play with 16bpp/24bpp depth to see if that changes anything
23:36 < KoalaBR> Well I had a look at the test_clip_plane() function
23:36 < marcheu> ok, that one is the 3D clip planes
23:37 < marcheu> but maybe it's done with a fragment program on nv30+ ...
23:37 < marcheu> that would explain their desastrous speed
23:38 < KoalaBR> Well, all I can do is to look at the output. If I get an idea, I will continue, if not I will have a look at some other open issue....
23:38 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
23:39 < pmdata> grr, I don't remember where I fount this doc
23:39 < marcheu> darktama, pmdata: so, would that be useful if I put my nv28 "online" ?
23:39 < marcheu> that involves quite a bit of work (notably installing X on the machine) so I'm asking first :)
23:40 < pmdata> I don't know, just updating the dumps regularly is sufficient for me
23:41 < pmdata> but vertex programs surely will need more than that
23:41 -!- K is now known as Unbeliever
23:41 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Remote closed the connection]
23:45 < pmdata> unbeliever> only have nv15 (see wiki for hw that people have)
23:46 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
23:47 < Unbeliever> where this page?
23:47 < Unbeliever> where is*
23:47 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Remote closed the connection]
23:48 < pmdata> http://nouveau.freedesktop.org/wiki/PatriceMandin this is mine
23:48 < pmdata> http://nouveau.freedesktop.org/wiki/RecentChanges you'll find other people hw
23:51 < Unbeliever> ok
23:52 < Unbeliever> how can i know what model is NV15?
23:52 < Unbeliever> for example
23:52 < pmdata> lspci -n
23:53 < pmdata> search for vendor id 0x10de (nvidia)
23:53 < Unbeliever> but i dont have a NV15
23:53 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
23:53 < pmdata> ah sorry, nv15 = geforce 2
23:53 < Unbeliever> why?
23:54 < pmdata> because nvidia decided: http://nouveau.freedesktop.org/wiki/CardFeatures
23:54 < Unbeliever> oh, ok
23:55 < KoalaBR> Sorry, but I need help with login in at cvs correctly (First project at sourceforge). Do i use the same user name and password for cvs login as for SF access / login?
23:55 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has quit [Remote closed the connection]
23:56 < KoalaBR> Don't bother, found it (SSH access only)
23:58 -!- etzel [n=thisnuke@69-160-140-26.ontrca.adelphia.net] has joined #nouveau
--- Log closed Пнд Авг 07 00:00:08 2006
