--- Log opened Чтв Авг 03 00:00:04 2006
00:03 < pmdata> 1400 are first and count-1 for gldrawarrays, time to commit
00:06 < pmdata> yeah, arb_vertex_buffer_object done
00:07 < marcheu> pmdata: try to change the array size, I think different mechanisms will be used
00:07 < marcheu> at least that was the case with display lists
00:17 < pmdata> this is different, a dma object is setup to feed the vbo engine
00:17 < marcheu> ah
00:18 < marcheu> sounds good, I wonder why that wasn't triggered by display lists
00:22 < pmdata> vbo is setup with only the params for gldrawarrays: mode,first,count-1, vertex data is sent via dma
00:23 < pmdata> previously, I did not see that, the driver 'optimized' it by not using these commands
00:23 < pmdata> when I simplified the code, it triggered luckily
00:31  * pmdata would like some dumps for arb_vbo now :)
00:39 < Colbert> OT: with mail progs like kmail. it shouldnt matter if exim is only configure for local, right?
00:40 < Colbert> card_10de-0171_test_vbo.txt
00:40 < Colbert> http://rafb.net/paste/results/4ffaDm13.html
00:45 < pmdata> did you change something in my test function?
00:50 < pmdata> count-1=0x17=23 for you, whereas I got 2
01:32 < Colbert> no change,  from multifile though
01:35 < pmdata> so it is a byte count (for nv17, or pci?) instead of a vertex count (for nv15, or agp?)
01:36 < marcheu> hmm, I give it a nv11 shot
01:36 < pmdata> agp or pci?
01:36 < marcheu> agp
01:36 < Colbert> pci
01:37 < pmdata> colbert> you have 3 pci transfers inited, whereas I only got 1 (agp or inside video ram? I don't know)
01:37 < marcheu> ah, he's on pci
01:38 < marcheu> pci is really a different beast than agp when it comes to system <-> card transfers
01:38 < marcheu> but it's good to understand it nonetheless :) pci-e is compatible with it
01:38 < Colbert> local mem usage?
01:39 < marcheu> yeah, I don't think your pci card is able to pull vertex data from system ram like agp can
01:46 < Colbert> might have to take a look at  mmtrace soon,after my mail is fixed.
01:47 < Colbert> to see where the mem is. or is that info here shoewher
01:47 < Colbert> somewhere
01:48 < pmdata> lspci -vvv
01:49 < Colbert> forgot, yeh 3 reigons.fd000000, f0000000 & fc780000
01:49 -!- tibbs|h is now known as tibbs
01:49 < Colbert> fd000000 (32-bit, non-prefetchable) [size=16M]
01:49 < Colbert>  Region 1: Memory at f0000000 (32-bit, prefetchable) [size=128M]
01:50 < Colbert> which is BS because its only a 64M card
01:50 < marcheu> Colbert: that area is NOT the video ram but the AGP area
01:52 < Colbert> its a real pci card, not agp
01:53 -!- pmdata [i=patrice@ANantes-154-1-2-52.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
01:53 < Colbert> didnt think pci cards maped like agp
02:01 < marcheu> hmm
02:01 < marcheu> yeah, they shoudln't
02:08 -!- atcl [n=Miranda@Lb83f.l.pppool.de] has quit ["Nobody needs a quit message :)"]
02:14 < Colbert> next time i reboot I'll see what lspci looks like before nvidia.ko loads
02:15 < marcheu> just the same probably
02:15 < marcheu> but it might be that you have an AGP chip with a PCI bridge
02:15 < marcheu> and thus it exposes some weird non-functional AGP stuff
02:17 < Colbert> weird is the key word, who else has a nv17 MX 440 to compare
02:20 < Colbert> talked to my friend in charge of my email, as soon as he gets home he's going to fix my acct. (hopefully)
02:21 < marcheu> I've got an AGP nv18 which is close enough. I'm heading to bed though
02:21 < Colbert> goodnight
02:22 < marcheu> I'll try to plug it tomorrow, and we'll see
02:24 -!- qfire is now known as qfire_away
02:33 < Colbert> couldnt connect to email because there was a ! in my password (dont look at me I didnt set it)
02:34 -!- qfire_away is now known as qfire
02:35 < qfire> what system didn't like the !?
02:52 -!- Duke` [n=gnu@ANantes-251-1-101-145.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
03:09 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Remote closed the connection]
03:35 < Colbert> qfire>not sure yet, only a guess still, accessing it through a web page right now
03:42 < Colbert> actually i think its getting lost at his server or being rejected by his firewall.
03:42 < Colbert> ! is used in uucp still, i think
03:43 < Colbert> so it might thnk it is part of an address
03:44 < Colbert> changing my nick to my sf.net name
03:44 -!- Colbert is now known as colberts
03:45 < colberts> colbert was already used by someone else
03:45  * colberts is back
03:45  * colberts is away: colberts
03:46 -!- th0ma7 [n=megalout@modemcable040.100-70-69.mc.videotron.ca] has joined #nouveau
04:07 < ajmitch> hi
04:09 < th0ma7> hi
04:09 < th0ma7> nibody around?
04:13  * colberts is back
04:13 < colberts> hi
04:13 < th0ma7> hi
04:14 < th0ma7> I have'nt done a lot of coding since a few years (mostly sysadmins stuff instead...) although I'd like to help out...
04:15 < th0ma7> I got access to a bunch of nvidia products... (we run about 70 linux workstations most of them uses nvidia cards in dual or quad head configs)..
04:16 < colberts> I one of the noobs here, marcheu pmdata lumag and darktama are the guru's, any of your cards quadros?
04:16 < th0ma7> a few ... yup.
04:17 < th0ma7> we have at least 3-5 different quaddros cards... 
04:17 < colberts> sweet, trying to figure out the differences between cards.
04:18 < colberts> a good place to start helping is by dl renouveou, thats our re tool
04:18 < th0ma7> I also got acces to: geforce go fx 5xxx, go2, geforce fx 6800, 6600, geforce2 pro, 4 ti 4200, etc...
04:19 < th0ma7> ok... http://nouveau.cvs.sourceforge.net/nouveau/renouveau/ ?
04:19 < colberts> the tool is userspace only dosent need root access
04:19 < colberts> yep, thats it
04:20 < th0ma7> can i simply download the daily package?
04:21 < colberts> i think so, not too many changes today, or anonymous cvs
04:21 < th0ma7> is there a wiki on how to run it? and where to send the dumps?
04:22 < ajmitch> it's more changing a few things in the tests & watching what changes in the dump, to work out the commands & data sent
04:22 < colberts> not yet, that i know about,
04:23 < th0ma7> and simply redirecting stuff in a file?
04:24 < colberts> there is a new compile time option to dump to files or redirection works 
04:24 < colberts> brb
04:24  * colberts is back
04:25 < th0ma7> where should I send the output? .. or to who?
04:26  * colberts is back
04:26 < th0ma7> Because I could send an output from a geforce fx 6800 and a geforce go 5 or 6xxx right away!
04:27 < colberts> not real sure right now, pmdata has been doing a large ammount of the analysis lately.
04:28 < colberts> I be interested in the output of the geforce go
04:29 < colberts> http://rafb.net/paste/
04:30 < colberts> paste the output there.(wish my test were included and workin right nowLOL)
04:32 < colberts> in a real bad thunderstorm right now need to sign off, be back soon.
04:32 -!- colberts [n=colby@adsl-69-211-9-63.dsl.chcgil.ameritech.net] has quit [Remote closed the connection]
04:33 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau
04:33 < th0ma7> wget ftp://th0ma7.no-ip.org/pub/MSI-GeForceFX6800-test1.txt.gz
04:34 < th0ma7> I'll try to add up a GeForce Go in a few mins.
04:41 < th0ma7> wget ftp://th0ma7.no-ip.org/pub/DellLatitudeD800-GeForceFXGO5650-test1.txt.gz
04:41 < th0ma7> I can keep thoses available for tonight.
04:49 -!- colberts [n=colby@adsl-69-211-29-118.dsl.chcgil.ameritech.net] has joined #nouveau
04:49 < th0ma7> hi again colberts.
04:49 < th0ma7> i've posted the output on your URL
04:49 < th0ma7> wget ftp://th0ma7.no-ip.org/pub/MSI-GeForceFX6800-test1.txt.gz
04:50 < th0ma7> wget ftp://th0ma7.no-ip.org/pub/DellLatitudeD800-GeForceFXGO5650-test1.txt.gz
04:50 < th0ma7> and here they are for manual download
04:50 < colberts> wow, some storm! lost a tree(missed my power and phone)
04:50 < th0ma7> where are you living?
04:50 < th0ma7> koz we got quite a storm yesterday and today in here also.
04:50 < colberts> chicago illinois US
04:50 < th0ma7> quebec, canada
04:52 < th0ma7> if you need me to remake thoses output using a few switches... it's a good time to ask... 
04:52 < th0ma7> I'll write down the info and I'll try thoses on our workstations at my job.... on thoses quaddro dual and quad headed.
04:53 < th0ma7> i'll be back later.. see ya.
04:53 < colberts> hey thanks
04:57 < colberts> I'll pass these along to pmdata, and marcheu.
05:08 < qfire> hmm... CVS renouveau doesn't seem to compile for me, fails on regl.h since 'GLchar' is not defined. 
05:09 < qfire> defined it manually in that file and it appears to compile fine.
05:20 < qfire> wget http://qfire.net/nv/dumps/ASUS-GeForce7900GT-Top.txt
05:27 < qfire> there were a few 'PLEASE REPORT' in that dump.
05:55 < darktama> qfire: yup, I *think* we know what that bit is now
05:56 < darktama> it seems to indicate that the object isn't in the "normal" instance ram
05:56  * ajmitch is just going through the fun of working out some values on this geforce 6600 :)
05:57 < darktama> ajmitch: what in particular are you looking at?
05:57 < ajmitch> test_clear()
05:58 < ajmitch> currently just playing with different colour values, watching the output change
05:58 < darktama> ah, cool :)
05:59 < darktama> I need to get an NV30 at some stage
06:01  * ajmitch only has 2 cards - NV40 & NV11
06:02 < darktama> yup, I have NV40 and NV05.. need an NV30 to figure out the vertex program instruction set
06:02 < darktama> it seems different to NV40.. fragment programs are mostly the same though
06:03 < ajmitch> how similar is it?
06:03 < th0ma7> I do have one .. a geforcefx go 5650
06:03 < th0ma7> wget ftp://th0ma7.no-ip.org/pub/DellLatitudeD800-GeForceFXGO5650-test1.txt.gz
06:03 < darktama> well, vertex progs seem completely different
06:04 < darktama> they're uploaded the same, but the opcodes differ a lot
06:04 < th0ma7> i can run in a few more tests if you'd like.
06:05 < darktama> th0ma7: that'd be great.. though, you'll need to run a *lot* of tests to figure out the entire layout of the instructions
06:06 < th0ma7> let me boot-up my laptop... got a bit of time on my hand tonight.
06:07 -!- colberts [n=colby@adsl-69-211-29-118.dsl.chcgil.ameritech.net] has quit [Read error: 110 (Connection timed out)]
06:08 < th0ma7> (--) PCI:*(1:0:0) nVidia Corporation NV31M [GeForce FX Go5650] rev 161, Mem @ 0xfc000000/24, 0xd0000000/28, BIOS @ 0x80000000/17
06:09 < th0ma7> on a Dell LAtitude D800
06:09 < th0ma7> What tests would you like me to run?
06:09 < darktama> do you know ARB_vertex_program at all?
06:09 < th0ma7> btw, at my job we run about 50-70 workstations almost all with nvidia cards.. mostly quaddro... if this might help?
06:10 < th0ma7> not at all.. just got in while reading lwn
06:10 < ddl> hey!
06:11 < ajmitch> ah, didn't realise this was on lwn
06:11 < ajmitch> hello ddl 
06:11 < darktama> well, I suggest learning ARB_v_p a bit.. because you'll need to write a lot of vertex progs, changing only small things.. and comparing the resulting instructions
06:11 < darktama> ddl: hey!
06:11 < th0ma7> I wonder if this was not from a comment posted on last weeks front page of lwn.net
06:12 < th0ma7> hurray...! ... hey, I have done a few opengl fractals on win95 in c++ about 10 years ago... 
06:13 < th0ma7> now I'm mostly a sysadmin... sadly not a low-level developper... snip!
06:13 < th0ma7> but suggest me some readings and I can surely learn.
06:14 < darktama> well, for ARB_v_p there's the current tests in renouveau that could be used as a base.. and http://oss.sgi.com/projects/ogl-sample/registry/ARB/vertex_program.txt is the spec for it
06:20 < darktama> airlied: btw, if you're looking at modesetting - haiku has a much better implementation than the Xorg driver does
06:23 < ajmitch> darktama: how do you get names like NV30_TCL_PRIMITIVE_3D_CLEAR_VALUES? from existing documentation, or just what they appear to do?
06:23 < darktama> we name them based on what they appear to do
06:23 < ajmitch> ok
06:25 < th0ma7> except running "renouveau" binary directly what tests exactly can I run using the renouveau package?
06:27 < th0ma7> btw, that is pretty heavy reading... 
06:29 < darktama> hehe, yeah.. it took me quite a while to get used to it when I started work on r300 fragprogs
06:29  * ajmitch is starting with the easy stuff
06:30 < darktama> ajmitch: well, on >=NV40 vertex and fragprogs are needed to even draw just a solid white triangle!
06:30 < ajmitch> ouch
06:31 < ajmitch> that sounds overly complex :)
06:31 < th0ma7> humm... not quite there myself... ;)
06:31 < darktama> nah, it actually simplifies things quite a bit imo.. once you figure out the instruction sets you can do lighting/texture environments etc quite easily
06:34 < ajmitch> ah, how I hate cvs :)
06:35 -!- eLShaman [n=elshaman@pac33-1-82-235-249-217.fbx.proxad.net] has quit ["Au revoir."]
06:36 < qfire> so in >=NV40 they implemented all of the OpenGL static pipeline as vertex and fragprogs? that's kind of cool.
06:36 < darktama> yeah, r300 and higher do the same
08:07 -!- colberts [n=colby@adsl-69-210-252-213.dsl.chcgil.ameritech.net] has joined #nouveau
08:10 < qfire> wb colberts
08:21 < colberts> should have left this off. storms still going. dsl flakey. be back tomorrow
08:21 -!- colberts [n=colby@adsl-69-210-252-213.dsl.chcgil.ameritech.net] has quit [Remote closed the connection]
09:01 -!- Myrizio [n=Myrizio@host135-101.pool80104.interbusiness.it] has joined #nouveau
11:51 -!- EdB [n=EdB@ARennes-251-1-87-238.w86-199.abo.wanadoo.fr] has joined #nouveau
12:27 -!- Duke` [n=gnu@ANantes-251-1-101-145.w86-203.abo.wanadoo.fr] has joined #nouveau
12:30 < marcheu> th0ma7: if you've got access to quadro cards, especially recent ones, we're very interested in the dumps
13:15 < ajmitch> hi marcheu 
13:15 -!- EdB [n=EdB@ARennes-251-1-87-238.w86-199.abo.wanadoo.fr] has quit ["Konversation terminated!"]
13:40 -!- `Duke` [n=gnu@ANantes-251-1-134-122.w86-210.abo.wanadoo.fr] has joined #nouveau
13:54 -!- Duke` [n=gnu@ANantes-251-1-101-145.w86-203.abo.wanadoo.fr] has quit [Read error: 110 (Connection timed out)]
14:05 -!- `Duke` is now known as Duke`
15:20 < th0ma7> marcheu:  I do have access... I'm on holyday this week but next week I'll try to gather some quaddro dumps.
15:54 < airlied> darktama: are  haiku sources sosmewhere obvious I'm missing..
15:59 < airlied> ah I think I have it in svn..
16:16 < darktama> airlied: http://svn.berlios.de/viewcvs/haiku/haiku/trunk/src/add-ons/accelerants/nvidia/ (if you haven't found it yet)
16:17 < darktama> there are a couple of things missing in haiku.  1. the ability to detect and program flatpanels that the BIOS hasn't.  2. the ability to swap output assignments
16:18 < airlied> cool, I just got it .. I'm just having a quick look over it to see what info I can get from it ..
16:18 < airlied> I'm trying to write-up a 2D register info file..
16:18 < marcheu> hmm
16:18 < marcheu> a 2D register file...
16:18 < marcheu> airlied: you don't have nv10reg.h ?
16:19 < darktama> airlied: http://rivatv.sourceforge.net/stuff/rules.tar.gz
16:21 < darktama> marcheu: that "modesetting object" that Lumag seen, do you recall what engine it was using?
16:21 < darktama> ie. software or graphics
16:21 < airlied> ah yes that is quite a good file...
16:21 < darktama> if it was software, it's possible that it's not a real object.. and that the kernel module gets called to handle it
16:21 < marcheu> darktama: nope, I don't think he mentioned it even
16:22 < marcheu> darktama: yes, that's what we were suspecting
16:22 < marcheu> I'm not sure what that object would buy us compared to the legacy VGA regs
16:23 < airlied> the legacy VGA regs aren't all that legacy..
16:23 < airlied> they have two sets of them which are mapped at different spots..
16:23 < darktama> if it's not a real object, then it'd buy us nothing at all :)
16:23 < darktama> yup
16:23 < airlied> so really they are enhanced...
16:23 < marcheu> darktama: even if it was
16:24 < darktama> yeah, true
16:24 < airlied> where did rules.xml come from?
16:24 < marcheu> heh
16:25 < marcheu> no one knows
16:25 < marcheu> except possibly the guy who leaked it :)
16:25 < airlied> cool I'm hoping when Keith P drops his new randr stuff I can do an nv versison..
16:26 < darktama> I made an attempt at mergedfb a few months back, but never got it to work.. I might have the code around somewhere still
16:27 < darktama> most of the work was basically cleaning up the code so it could save/restore modes to both heads, and porting the radeon mergedfb code over
16:29 < airlied> darktama: yeah I was going to start doing mergedfb but I'd like to see what keithp does..
16:29 < airlied> I can hopefully rip off the Intel driver when they get it done ass a better starting point than radeon..
16:29 < darktama> indeed, I saw the post on the xorg ml.. looks interesting :)
16:31 < darktama> I suspect the issue I had getting it to work is the same reason why I have to force the vanilla nv driver to use CRTC 0
16:58 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has quit ["Leaving"]
17:44 -!- shenki [n=shenki@ppp167-129.lns3.adl4.internode.on.net] has quit ["Leaving"]
19:48 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
20:07 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
20:14 -!- atcl [n=Universe@Lb837.l.pppool.de] has joined #nouveau
20:31 -!- th0ma7 [n=megalout@modemcable040.100-70-69.mc.videotron.ca] has quit ["using sirc version 2.211+KSIRC/1.3.12"]
20:37 -!- KoalaBR_ [n=KoalaBR@port-83-236-15-70.dynamic.qsc.de] has joined #nouveau
20:41 < KoalaBR_> Anybody here?
20:42 < marcheu> yeah
20:42 < KoalaBR_> Hi, I would like to offer some help. I already downloaded renouveau from CVS, including doc
20:43 < KoalaBR_> I have an PCI-Express GForce 6600 GT card
20:43 < marcheu> does it work for you ?
20:43 < KoalaBR_> C-Experience and a very basic knowledge of GL
20:43 < marcheu> or what's your question
20:43 < KoalaBR_> Wanted to know, what kind of dumps do you need for that card?
20:44 < marcheu> I think we need people figuring out features more than raw dumps
20:44 < KoalaBR_> I have read most of the IRC-logs
20:44 < KoalaBR_> Well, I'm just starting. Where should I look first then?
20:45 < marcheu> did you read the README ?
20:45 < KoalaBR_> Yes
20:45 < KoalaBR_> I know the basic routine from that and the logs 
20:46 < marcheu> hmm, do you understand the dumps already ?
20:47 < KoalaBR_> I would guess, that I would need practice, If you give me _now_ a dump to analyse, I certainly would be lost
20:47 < KoalaBR_> Compiling without any changes (x86) bails out: 
20:48 < KoalaBR_> regl.h:635: error: syntax error before '*' token
20:48 < marcheu> is it current CVS ?
20:48 < KoalaBR_> Yes
20:48 < KoalaBR_> error: syntax error before "GLchar"
20:48 < KoalaBR_> I will have a look
20:53 < KoalaBR_> This type should be defined from SDL_opengl.h? 
20:54 < marcheu> nope, it comes from the OpenGL extension headers
20:54 < marcheu> that means you don't have a header that defines it, that's all
20:54 < marcheu> (because SDL_opengl.h include the standards opengl headers)
20:54 < KoalaBR_> Ok
20:55 < KoalaBR_> So what now?
20:55 < KoalaBR_> I'm on Gentoo btw
20:55 < marcheu> you should upgrade your glext.h
20:55 < marcheu> hot sure what's the process on gentoo for that
20:55 < marcheu> alternatively, we should remove the offending prototypes from the header
20:56 < KoalaBR_> is it included in nvidia drivers or mesa?
20:56 < KoalaBR_> void (*BindAttribLocation)(GLuint, GLuint, const GLchar *); would be it
20:57 < KoalaBR_> there would be ~10 prototypes which are using this type
20:57 < marcheu> yup
20:58 < marcheu> but can't you just update your glext.h file ?
20:58 < marcheu> http://oss.sgi.com/projects/ogl-sample/ABI/glext.h should be OK
20:59 < KoalaBR_> I did look for glext.h I found quite a few:
20:59 < marcheu> the reference one is SGI's
20:59 < KoalaBR_>  /usr/include/GL/glext.h
20:59 < KoalaBR_>   /usr/lib/opengl/global/include/glext.h
20:59 < KoalaBR_>   /usr/lib/opengl/xorg-x11/include/glext.h
20:59 < KoalaBR_>  /usr/lib/opengl/nvidia/include/glext.h
20:59 < marcheu> and the standard location is /usr/include/GL/glext.h
21:00 < marcheu> but gentoo has an opengl-update IIRC which plays with that
21:00 < KoalaBR_> I will tweak the makefile (as I have seen a GLchar typedef in that file /usr/include/GL/glext.h)
21:02 < marcheu> does it work if you add :
21:02 < marcheu> #define GLchar char
21:03 < marcheu> on top of regl.h ?
21:03 < KoalaBR_> Yes
21:04 < marcheu> alright, I'll commit that
21:05 < KoalaBR_> Well, my card is recognized as PCI-E
21:05 < marcheu> which it is ?
21:05 < marcheu> you said it was
21:06 < KoalaBR_> Want lspci output? A 128MB 6600 GT PCI-E
21:08 < marcheu> we don't need it, renouveau outputs the pci id anyway
21:08 < marcheu> we might need it later if we find some bug in renouveau :)
21:08 < marcheu> but not for now
21:08 < KoalaBR_> fine
21:09 < KoalaBR_> I did the default test and I get a bunch of NV30_TCL_PRIMITIVE_3D
21:09 < KoalaBR_> This are still unknown objects / command, aren't they?
21:10 < marcheu> yeah, when there's a command number next to it
21:10 -!- Myrizio [n=Myrizio@host135-101.pool80104.interbusiness.it] has quit [Read error: 104 (Connection reset by peer)]
21:11 < KoalaBR_> Those data in brackets [ ] ?
21:11 < marcheu> yes, [0xSOMEWHERE/4]
21:12 < KoalaBR_>  NV30_TCL_PRIMITIVE_3D      [0x0280/4] = 0x00001400 | UNKNOWN = 00001400
21:12 < marcheu> yep, 280 is an unknown
21:12 < KoalaBR_> Ok, fine. Understood so far. What can I do to help?
21:13 < marcheu> figure out what 280 is ? :)
21:13 < marcheu> more generally, find some opengl commands that trigger unknown commands, and try to make sense out of what you see
21:14 < KoalaBR_> Ok, will try
21:14 < marcheu> once you think you understand a command, say 280 is the secondary material color, add it to nv_objects.h/objects.c
21:16 < KoalaBR_> Most certainly I will pester you and the others with questions before I reach that point :)
21:44 -!- pmdata [i=patrice@ANantes-154-1-59-3.w81-53.abo.wanadoo.fr] has joined #nouveau
21:44 < pmdata> hello
21:45 < KoalaBR_> Hi
21:46 < KoalaBR_> I'm trying to offer help (GF 6600 GT, 128MB). If you need anything (simple) from me, just ask
21:47 < pmdata> hum, is it nv30?
21:47 < KoalaBR_> PCI-e Card
21:47 < marcheu> 6600=nv44
21:47 < KoalaBR_> renouveau reports it as  NV40 family
21:47 < marcheu> KoalaBR_: btw if you plan on helpin in the long run, please add yourself to the wiki
21:47 < marcheu> you & your hw, of course
21:47 < marcheu> nv43 actually
21:48 -!- Myrizio [n=Myrizio@host194-101.pool80104.interbusiness.it] has joined #nouveau
21:48 < toresbe> Is the aid of someone with an nVidia Riva TNT2 desired?
21:48 < marcheu> just like us : http://nouveau.freedesktop.org/wiki/CategoryHomepage
21:48 < KoalaBR_> I will try at least for a few weeks. If I realize that I'm to stupid to understand this, I will stop as I'd be no help
21:49 < KoalaBR_> But I would really like to see this project succeed
21:49 < marcheu> toresbe: I think we've got plenty of TNT2 alread. but if you can code, that can be helpful
21:49 -!- K [n=hazel@212.145.81.31] has joined #nouveau
21:49 < toresbe> marcheu: I'm a mediocre C programmer, but I have very little experience coding drivers.
21:50 < K> pmdata: did you recive my file?
21:50 < pmdata> which one?
21:50 < marcheu> the one that went through me I suppose
21:50 < K> emm
21:50 < K> 10de.0020
21:52 < pmdata> ah, yes I looked at it, lumag did most tnt2 stuff
21:52 < K> ok
21:53 < pmdata> it seems everything has been found (at least the commands listed in your dump)
21:53 < pmdata> you should ask lumag if there is anything that should be checked
21:53 < KoalaBR_> Running renouveau I get Unknown bits in context: 00010000 PLEASE REPORT!
21:53 < K> ok
21:56 < marcheu> KoalaBR_: I think darktama has that mostly sorted out now
21:56 < KoalaBR_> Ok
21:58 < pmdata> I saw in the irc logs someone today which got quadro cards, I hope he has some nv1x ones
21:58 < marcheu> pmdata: hey, I've got 2 softquadroed nv1x, don't they fit ?
21:59 < marcheu> and I can even send you separate dumps now :)
21:59  * pmdata just wanted to compare
21:59 < marcheu> well, they run in quadro mode I guess, because the vertex submission is different
21:59 < pmdata> but yeah, post the full nv1xgl dumps!
21:59 < marcheu> well, I've got these here : http://icps.u-strasbg.fr/~marchesin/nvdri/tests/
21:59 < marcheu> I can fire the nv15 if you like
22:00 < marcheu> I'll do a short nv15 session and then play with the nv28
22:05 < pmdata> did you notice most of your nv11 dumps are empty?
22:06 < marcheu> yeah
22:06 < marcheu> no idea why
22:06 < marcheu> it's not the case with the nv15
22:06 < pmdata> did you check the object creation list at stat for unknown ones?
22:07 < marcheu> hmm, could that lead to empty dumps ?
22:07 < pmdata> I don't know
22:07 < marcheu> I think they would just show up as empty objects
22:08 < marcheu> what I think rather happens, is that the fifo might be bigger/at a different place than we though
22:08 < pmdata> but the tcl engine is used in each test for a different function, so it should not be empty
22:08 < marcheu> and we end up missing commands
22:11 < marcheu> I'll have to investigate that
22:12 < KoalaBR_> NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3X (and _3Y _3Z) are only used for colorsetting? Or define other attributes as well ?
22:20 < pmdata> you should check vertex attrib doc
22:23 < K> i execute a glBegin() but i cant see a dump...
22:24 < marcheu> pmdata: http://icps.u-strasbg.fr/~marchesin/nvdri/tests/
22:25 < pmdata> marcheu> do you have some stats about people visiting the nouveau.freedesktop.org site?
22:25 < marcheu> pmdata: no, why ?
22:26 < marcheu> I also wonder where all those people suddenly come from, if that's the underlying question :)
22:27 < pmdata> I want to know how many people are interested by the nouveau project
22:27 < KoalaBR_> Well, me I have read something about you on LWN (in the comments) and some blogs from Kernelplanet, I think it was
22:29 < pmdata> nv17 [0x2bc] is set to 1 when alpha stuff is changed
22:29 < marcheu> pmdata: no, really. I suppose I could bug the fd.o admins, but it's not really important, or is it ?
22:29 < pmdata> no, it would keep us on working
22:29 < pmdata> "oh my god, we made xxxx visits this month!"
22:29 < marcheu> :)
22:30 < pmdata> "then add some p0rn and banner ads"
22:30 < marcheu> and a link to an ebay-wishlist for poor developers with only a nv15 ;)
22:30 < K> when i execute glBegin() i only obtain this: http://pastebin.ca/113685
22:31 < K> i cant see a dump
22:31 < marcheu> K: glbegin is not enough to trigger anything
22:31 < marcheu> that's why we have a tri() function
22:32 < K> in the doc it says: 
22:32 < K> So here, the packet carries 1 dword (or 4 bytes) of data, is sent to channel 1,
22:32 < K>   and is of type 0x1808. It turns out that 0x1808 almost means glBegin()
22:32 < marcheu> yeah, it will issue a glbegin once it finds a matching glend
22:32 < K> ok
22:33 < marcheu> now, can that nv18 be softquadroer
22:33 < marcheu> softquadroed
22:33 < pmdata> marcheu> if I did not have a nv15, I would not have done the nv10 tcl object :)
22:33 < marcheu> pmdata: well, I think you need something else now :)
22:34 < pmdata> there are some missing commands (not much I agree)
22:34 < marcheu> so does lumag, really
22:37 < pmdata> I think nv10[0x318] unknown is some point parameter stuff, always set with point_size and point_smooth_enable
22:39 < pmdata> nv17[3f8,3fc] seems to be nv20[384,388] both unknown
22:39 < pmdata> nv20 also has [0x318]
22:39 < marcheu> 318 is point size for nv20 ?
22:39 < marcheu> point size is one difference between NV20 & NV25
22:43 < pmdata> no, point size is 43c on nv20
22:43 < marcheu> there are 2 point size, one for nv20, one for nv25+
22:44 < marcheu> but real nv20 (geforce 3) are pretty rare
22:45 < pmdata> hum, time to write a test_point_size then
22:45 < pmdata> on nv10, point size is 0x3ec
22:50 < atcl> hi there
22:50 -!- pmdata is now known as pmdata_pause
22:59 < marcheu> hi atcl 
23:00 < KoalaBR_> Well, stopping for today. Bye
23:01 -!- KoalaBR_ [n=KoalaBR@port-83-236-15-70.dynamic.qsc.de] has left #nouveau []
23:01 < atcl> hmm, i'd like to help
23:02 < atcl> i know assembler quite well
23:02 < marcheu> atcl: ok, do you have any specific question ? did you read what I discussed with KoalaBR_ ?
23:02 < atcl> well discuss, just had a flash in mind, i need to code for a second..., sorry :)
23:10 < marcheu> pmdata_pause: so, I think softquadro could probably work for you too
23:10 < marcheu> pmdata_pause: it works on my nv11, nv15 & nv18
23:11 < marcheu> pmdata_pause: but there's one thing I noticed. you have to softquadro the card before ever starting X or it doesn't work
23:11 < marcheu> so next time, try it on a fresh boot
23:43 < Myrizio> hi, is there any tool to track the read/writes from and to the card done by the true driver?
23:44 < Myrizio> s/true/nvidia binary blob
23:45 -!- pmdata_pause is now known as pmdata
23:45 < pmdata> marcheu> is the newest nvclock 0.8b2 ok?
23:49  * pmdata noticed something in nv18 color_mask dump
23:49 < pmdata> 0x02bc/4 changes -> alpha component?
23:49 < pmdata> I now that at least on mine, alpha flag is never enabled
23:53 < pmdata> well, it is always enabled in fact
--- Log closed Птн Авг 04 00:00:05 2006
