MostAwesomeDude: agd5f: xf86-video-ati requires xf86miscproto to build, but doesn't actually check for it during configure; should I fix that?
hifi: please do, I also missed that when building
hifi: there were possibly some other things too that were needed but not checked
MostAwesomeDude: agd5f: Would you kindly sign off http://rafb.net/p/ghWkv020.html , patch for xf86-video-ati that adds xf86miscproto to configure.
bridgman: finally back on a real PC... one quick comment from backlog surfing
bridgman: re: selling boards with part of the chip disabled; a few generations back
bridgman: we only had the ability to disable big chunks of the chip, so of a chip
bridgman: failed testing on one quad pipe but not on the other we could sell
bridgman: that chip on a less expensive card and enable only the part which
bridgman: worked to full spec. You try to set the price based on yield so that
bridgman: demand for both full-chip and partial-chip boards match what comes out
bridgman: of the manufacturing process. There are a couple of potential problems,
bridgman: of course; and they start when demand for the cheaper "half-chip" board
bridgman: is higher than you expected, giving you three options. Raise the price,
bridgman: allow your dealers to run out of product, or sell some boards with
bridgman: fully functional chips but with one half turned off anyways. I think
bridgman: we only did this once, but for years afterwards people were turning on
bridgman: anything we turned off in the hope of getting more performance for the $$
bridgman: These days testing is more sophisticated and so you tend to see things
bridgman: like speed-binning rather than turning off parts of chips, although
bridgman: I guess at least one vendor is still turning off shaders (eg. 112 vs 128)
bridgman: On most chips though we all have better ways of managing yield so I think
bridgman: the days of selling "partial chips" are almost behind us
awj9999: next time bridgman comes in, someone ask him what the idea was rebranding cards with names that overlapped with cards from a completely different hw generation
awj9999: e.g. some "X550" cards actually being rebranded X700s
awj9999: though I guess it's not as bad as what nvidia did--the ersatz "X550" actually has *better* shader capabilities than a real X550
awj9999: contrast that with the infamous "GeForce 4 MX" cards...
rx__: ...
rx__: for a low end part does it really matter?
awj9999: it confuses people... including driver developers, it seems
rx__: uhm... what?
awj9999: guy on phoronix forums is having trouble with his X700... it looks like his kernel and X disagree on what GPU is in his graphics card
awj9999: http://phoronix.com/forums/showthread.php?t=10954
rx__: how is that a hardware branding problem?
awj9999: well, it looks to me that the Xorg driver thinks that his PCI ID is an rv370 card (when it's actually an rv410)
awj9999: which is likely because whoever added it to the PCI IDs list was working from a list of retail card names and went "X550, aha, rv370"
rx__: uh
rx__: xf86-video-ati works off of amd/ati'
rx__: s official list
rx__: so if there's a problem with pci-id <-> chipset matching.. i'm sure there would be problems on the windows side as well
rbrett: GeForce 4 MX tricked so many of my friends :P
awj9999: no, i checked an inf file from a windows catalyst
awj9999: and every sub-id of that pciid was listed as an rv410
rx__: yes.. and the driver won't have any problems with detecting that
rx__: { .ATI Radeon X550/X700 Series .,RV410,5657}
rx__: tada
awj9999: hm, the guy on phoronix must have some outdated component with wrong/missing pciids then...
rx__: wait..
rx__: actually going through the source
rx__: this line concerns me
rx__: #define PCI_CHIP_RV370_5657 0x5657
awj9999: also notice that in his lspci his card is listed as an "unknown device"
awj9999: so its pciid must be missing altogether from his kernel...
rx__: ok.. yea.. going through the source it's being ID'd as a rv370 chip
rx__: it's correct in ATI's list though
rx__: one of the devs just needs to run the parse_pci_ids.pl script
rx__: http://developer.amd.com/gpu_assets/ATI_Vendor_ID_May_2008.txt
awj9999: so I was right. someone must've added it by hand and got it wrong
rx__: it's not added by hand
awj9999: the one that's wrong in the driver source must have been... unless you're telling me that at some point ATI's own list was wrong
rx__: or i would hope not at least
rx__: http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=0c26806245381b925b73ff9f3b747e3dcf0ebd6f
rx__: the ati_pciids.csv was updated
rx__: pretty sure it was from ati
bridgman: awj9999: my understanding is that the marketing numbers are supposed to
rx__: http://ati.amd.com/developer/ATI_Device_IDs.txt
rx__: obviously it's been updated
bridgman: represent more of "what a card can do this year" rather than engineering
bridgman: internals. We have another set of numbers (eg RV370) we use internally when
bridgman: talking about the guts of a chip
rx__: developers don't get this wrong.. when in doubt, blame ati :P
bridgman: rx__: the list is typed in by humans, so...
awj9999: nonono
awj9999: not ati's fault
rx__: bridgman; in that case.. we should get rid of humans.. they're error prone
awj9999: looks like "mirko" (FDO bugreporter) mistakenly thought his card was an RV370 based on the retail name, and Alex didn't bother to check before adding it as such to the driver
bridgman: agreed, but we have to replace them with software written by humans
bridgman: and so far it's not clear that is always an improvement ;)
awj9999: http://bugs.freedesktop.org/show_bug.cgi?id=14289
rx__: sighs
bridgman: what can you do ? Monkeys were even worse ;)
rx__: how's the weather up there?
bridgman: perfect IMO - cool in the evenings (50s F) and warm during the day (70-80f)
bridgman: no A/C, light a small fire in the evening kind of weather
rx__: lucky
bridgman: yep; trying to enjoy it while it lasts ;)
rx__: i see no 48xx pci ids here
bridgman: I don't actually know how the list is generated but will find out
bridgman: we probably need to update more frequently so devs don't have to use INF files
bridgman: from drivers as the "master list"
rx__: generated meaning.. how IDs are put into the list on amd's site?
glisse: bridgman: we used to take windows driver inf files
glisse: but now i think it's generated from some others sources files
rx__: the two sites i listed above have the pci ids/chipsets/names
rx__: but they changed the formatting
bridgman: rx__: we have a big honkin' online database with all the info on each chip
bridgman: variant, including IDs, marketing names etc..; various extracts are made
rx__: available to the public? ;)
bridgman: for different purposes and used around the company
awj9999: bridgman: I only resorted to an inf file because I didn't know about that online list. I'm not a dev, just an enduser with JUST enough tech. knowledge to be able to ask the stupidest and most annoying questions imaginable ;)
bridgman: sure, we just love releasing details about the next four years of development
bridgman: oh wait, no we don't
bridgman: awj9999: don't worry, I didn't know about the online list until today either ;)
glisse: bridgman: so what is the code name in 4year, r900 ? ;)
bridgman: glisse: I think we run out of numbers again by then, so up to marketing
glisse: me makes fun of stupid marketing people but ing are not that good at finding name :)
bridgman: that's where some of the human element comes in; we tend to add the marketing
bridgman: names near the end
awj9999: is there doc on the API-visible improvements of r400 as opposed to r300?
glisse: awj9999: i think old card review have this information
awj9999: in IRC log a couple days ago, MostAwesomeDude said something about GLSL on r400
glisse: iirc it's mostly more unit better clock and bug fixes
awj9999: no I mean doc on how to program them :)
awj9999: how to program r400 to do the new things that weren't in r300
bridgman: awj9999: from a driver-visible point of view the r5xx acceleration doc covers
bridgman: everything except the display block (which is totally different in 5xx) and memory
bridgman: controller / bus interfaces (which we initialize in VBIOS and suggest you leave along ;))
bridgman: hw implementation of the 3d engine is totally different in 5xx but programming is
awj9999: but r400 fragment shader is somehow intermediate between r300 and r500, isn't it?
bridgman: pretty similar; main difference is shader model 3 vs 2.whatever in 4xx
bridgman: yeah, I think we have some docs oriented to application developers as well
bridgman: AFAIK main diff between 3xx and 4xx is things like max # instructions in a shader
awj9999: yeah. r300 is 2.0 in D3D terms, r400 is 2.0b, r500 is 3.0
bridgman: etc, while going to sm3 is a much bigger jump since you get more flow control options
awj9999: I understand that
awj9999: so I guess the question is how do you program the longer shaders in r400
bridgman: just write longer shaders in the application program
awj9999: I mean from driver developer's pov
bridgman: for opengl that would typically be using arb_fragment_program
glisse: awj9999: there is just more room
glisse: awj9999: from driver pov this doesn't change anythings
awj9999: oh, it's that simple?
bridgman: from driver developers pov you wouldn't do much with longer shaders
bridgman: what he said
awj9999: gotta look at the register map again
bridgman: yep; but the longer shaders are nice for app developers
awj9999: I thought there was something else right after the "shader instructions" block
bridgman: pages 7 and 8 in the r5xx acceleration manual summarize the differences betwee
bridgman: 3xx, 4xx and 5xx
awj9999: n/m, glisse pretty much answered my question
awj9999: I thought there would be something tricky to it since r400 shaders can be sooo much longer than r300
rx__: decides to crash
awj9999: like the register region where you poke shader instructions was banked or something
bridgman: I don't remember seeing mention of banking but didn't look at 4xx regs much
bridgman: going from 4xx to 5xx everything changed from 1 register per instruction word
bridgman: to loading everything indirectly via index/data, and on 6xx it changes
bridgman: again since we store the shader instructions in video memory
KeefR: hi, im currently trying to get my onboard x1250 ati working with an nvidia peg-card. I'm currently running a minimal xorg.conf, but only get '(EE) Screen 2 deleted because of no matching config section.', tried with current ati and radeonhd-git, it works with only the ati-card active. I think i missed something in my xorg.conf or is there a fundamental prob? (xorg.conf: http://phpfi.com/325863)
adamk: KeefR, If it works with only the ati card, this sounds like something you need to take up with nvidia.
KeefR: adamk: probably, but the error message is strange, because there is a valid config section
KeefR: and the ati card gets detected properly
bridgman: which driver are you running for the nvidia hw ?
adamk: 'nvidia'
KeefR: the proprietary one
KeefR: yes i know, that's something that isn't supported ;-)
KeefR: but i'd like to know wether i made a stupid mistake in my xorg.conf or wether it's a fundamental problem
awj9999: btw bridgman, congratulations to everyone at ATI on R700, the nvidia fanboys on 4chan are having conniptions :)
KeefR: and i thought the people here would notice an obvious mistake in the xorg.conf
adamk: KeefR, Well your xorg.conf file looks fine here.
glisse: awj9999: there is bench available ?
KeefR: adamk: ok, that's what i wanted to hear, so i probably have to look somewhere else for the problem
awj9999: windows benches
KeefR: if only there would be an alternative to proprietary nvidia-driver like there is for radeon...
awj9999: but 4850 seems to be holding up to nvidia's shiny new toy incredibly considering it costs half as much
awj9999: looks like nv only engineered their new chip for synthetic benchmarks
awj9999: it honestly looks like they have a new geforce fx on their hands
KeefR: or any advice for a cheap dual-dvi ati-card so i could throw out the nvidia-card?
bridgman: KeefR: can you paste your logs as well ?
KeefR: bridgman: just a second
bridgman: awj9999: we're pretty happy with the 770; it's always tough because
bridgman: you have to guess two years in advance where the sweet spot will be, but it looks
bridgman: like this time we guessed right ;)
awj9999: yeah
awj9999: total 180 flip from last gen
KeefR: bridgman: http://routerhell.de/Xorg.0.log
glisse: KeefR: what is not working ?
KeefR: hm, that might be the problem: (--) Assigning device section with no busID to primary device
KeefR: glisse: trying to get onboard ati working together with peg-nvidia
glisse: KeefR: i know that but what is the problem ? no picture ? no 3d ?
glisse: the log pretty much say everythings is fine
bridgman: KeefR: either add a bus ID to device 2 or swap the devices
KeefR: bridgman: swapping isn't possible with bios and onboard, but I'm currently trying with the bus-id
bridgman: KeefR: not sure, but I think X is swapping for you anyways ;)
bridgman: where do the boot messages pop up - screen attached to 1250 or to the nvidia card ?
KeefR: the boot messages go to the onboard x1250
KeefR: what do i have to enter for the busid? "PCI:1:5:0" didn't work
KeefR: lspci says: 01:05.0 VGA compatible controller: ATI Technologies Inc RS690 [Radeon X1200 Series]
bridgman: KeefR: so why do you think you can't swap devices ?
KeefR: bridgman: i can't swap them physically and there's no way to configure in bios to say nvidia is primary
KeefR: probably we're talking about different ways to swap ;-)
KeefR: you think of trying ati as first device in xorg.conf?
bridgman: right
KeefR: ok, i'll test that
bridgman: I don't *think* it should make a difference, but then again it should
bridgman: never rain on weekends eithers
rindolf: Hi all.
rindolf: I bought an ATI HD2600 card, and would love to use the FOSS drivers with it.
rindolf: Previously I had an old Nvidia GeForce 4 MX.
rindolf: Whose ventilator broke down.
adamk: rindolf, FYI, the FOSS drivers do not support 3D acceleration on that card at the moment.
adamk: rindolf, I'm not sure if they even support more than modesetting in terms of 2D.
rindolf: adamk: ah.
rindolf: adamk: no problem.
KeefR: ok, changing primary did change something: now x complains about no config section for screen 1+2 (the nvidia ones)
rindolf: adamk: I'm not really a gamer.
rindolf: adamk: I'm using the nv driver now.
rindolf: And previously used nouveau.
KeefR: and now everything is messed up, had to go to the notebook and am now rebooting the system
bridgman: rindolf: either radeonhd or radeon will work fine with your HD2600 card other than
bridgman: acceleration, and acceleration is in process now
rindolf: bridgman: OK, thanks.
rindolf: bridgman: I'd also like to help a little with the development.
rindolf: Or a lot if time permits.
Alpo\: Is the DRI and 3D acceleration working with Radeon X1950 ?
rindolf: Thought I'm not really a hardware dev.
adamk: Alpo\, Yes.
Alpo\: I tried the driver about a month ago and DRI didn't work
rindolf: I mean I graduated from EE from the Technion, but it would be a stretch to call me an electrical engineer.
adamk: Alpo\, I have an x1900 and it works fine. It should work on a x1950, too.
bridgman: rindolf: there are all kinds of ways to help, either by coding or (possibly even
bridgman: more useful) helping to find out where & how things broke, eg by bisecting between
KeefR: ok, back to normal again, i think the best way is buying a card from the same vendor
adamk: Alpo\, DRI for the card was added to mesa probably around a month ago, maybe a little less.
bridgman: good and bad builds and reporting the results back to the devs
rindolf: bridgman: ah, OK.
rindolf: bridgman: do you have a version control repos?
KeefR: nvidia + ati don't seem to work well together
rindolf: KeefR: pardon?
bridgman: yes, you can see all the code at gitweb.freedesktop.org
rindolf: I'm so glad I'm dumping hang-vidia.
Alpo\: In which version was it added?
rindolf: bridgman: ah, OK.
bridgman: the source control system is called "git" - very powerful but a bit of a learning
bridgman: curve to fully understand
KeefR: rindolf: i'm talking about my probs getting to run nvidia and ati together
KeefR: and the easiest way to solve this seems to dump the nvidia card ;-)
rindolf: KeefR: ah.
rindolf: KeefR: two cards on the same machine?
glisse: KeefR: try with the nv driver
glisse: or nouveau driver
glisse: closed source driver prety much replace everythings
KeefR: glisse: nv doesn't support my card with xv
bridgman: KeefR: I think that is why there is nouveau ;)
KeefR: glisse: but i bought that nvidia card used for 2 beers, so no money lost, only time :-D
rindolf: KeefR: :-)
KeefR: bridgman: the last time i tried, noveau wouldn't do on my setup ( i think 2 month ago)
bridgman: Alpo\: you pretty much need latest of everything from git, or Fedora 9
bridgman: with everything from updates-testing (or whatever it is called)
bridgman: KeefR: yeah, I'm not sure how NV30/40/50 etc.. translates into products
Alpo\: So I'll need to compile it
KeefR: but before I buy ati, normally 2 videocards from ati should work?
bridgman: yes, but let one of the devs give you the full answer; I'm a bit rusty
bridgman: on the "what works today" details
bridgman: particularly where Xinerama is involved
KeefR: ok, I don't expect a definitive answer: normally it works with radeon or radeonhd is good enough for me
KeefR: well, my setup is even stranger....3 monitors + 1 tv set with hdmi...
Alpo\: bridgman: Do I have to compile only the xf86-video-ati or also something else?
bridgman: yeah, that's where I start to get a bit fuzzy about what works ;)
rindolf: I'm on Mandriva Cooker, BTW.
bridgman: Alpo\: you need -ati, plus mesa/drm, plus mesa/mesa
bridgman: there are some good online instructions about exact build sequence, it does
KeefR: ok, 50 euro for a x2600pro with dual-dvi
bridgman: seem to matter and has also changed recently
bridgman: KeefR: seems like a good choice; you'll get to share the joys as the devs
bridgman: bring up R6xx acceleration support over the next month or two
KeefR: bridgman: hm, x2600 not supported very well atm?
KeefR: i don't need 3d, just xv (ok, I know, that's closely related on ati...)
bridgman: acceleration code is just starting now but shouldn't take too long
bridgman: xv needs 3d these days; pre-5xx we had video processing hw in the overlay
bridgman: but these days it's all done with textured video using the shaders
bridgman: 3d engine is used for pretty much everything; if you look at a die shot
KeefR: hm, the problem is getting an older card with dual-dvi output, especially if you prefer buying stuff local and not ordering online
bridgman: of a modern gpu it's basically a big honkin' 3d engine with some other stuff
bridgman: around the edges
KeefR: well, i have a dual-core athlon x2 5200, scaling in cpu should be possible for 720x576 videos
KeefR: and there's hope, that i'll work some day ;-)
KeefR: ati is probably a better choice for the future looking at open source support on linux
bridgman: I don't think 6xx accel will be long; we have documentation now and devs cleared
bridgman: to work under NDA; just working hard to get the docs ready for public release
KeefR: yes, i read that online
KeefR: i can wait some more month ;-)
KeefR: atm i have a setopbox running linux with intel graphicscard connected to my tv, so I don't have to stop watching tv because of driver issues, but I'd like to get ati with hdmi working to get live hdtv on my tv-set
bridgman: btw here's the best die shot I have seen so far (rv770) :
bridgman: http://s1.directupload.net/images/080620/npcr22yj.jpg
bridgman: KeefR: assume no HDMI on your 690-based motherboards ?
bridgman: motherboard should be singular, my "s" key sometimes types by itself ;)
KeefR: bridgman: hdmi is on my motherboard and it's working (tested with radeonhd some weeks ago)
KeefR: so x1250 is working fine with open source drivers, but not together with fucking nvidia
bridgman: ahh...
bridgman: it is tough coexisting with the binary nvidia driver on linux; as glisse said, it
KeefR: but testing with you helped me verify, that it's an issue with the nvidia driver and not some stupid mistake in my xorg.conf (like with my new monitor running 1920x1200 and didn't get it to run because i had a line with 'virtual 1280x1024'...)
bridgman: replaces pretty much everything
RTFM_FTW: bridgman yeah that really shows off the shader core :D
KeefR: yap, even without trying to get opengl running
KeefR: but time is money and i don't want to invest more time/money in trying to get that nvidia stuff working, it's probably a better investment to buy an ati card and try running radeon or radeonhd
KeefR: and ati support did become a lot better in the last few month, i'm really impressed about that
KeefR: ati did as promised and the open source devs did a great job
KeefR: thanks for your support, I think I'll come back again, if the ordered ati card arrived and doesn't work together with the onboard ati in a basic setup ;-)
KeefR: but you saved me a lot of time, i worked on that setup the last 3 month and didn't get it working and always thought it's probably a configuration error on my side
glisse: KeefR: making two gpu working at the same time in a computer is not basic setup :)
KeefR: glisse: yeah, I know that, but i did it before (some years ago). with basic setup i mean no acceleration, no 3d, no xvideo
KeefR: and I don't even dare to think about h.264 acceleration ;-)
bridgman: it's actually going to get a bit harder before it gets easier; world is moving to
bridgman: randr, but randr doesn't properly support multiple gpus today
bridgman: not sure if having gpus from different vendors makes it better or worse, but
KeefR: randr is nice, I use it on my eeepc
bridgman: i can't imagine it making things better ;)
bridgman: seems to be agreement that randr1.3 will include a GPU object so everything should
bridgman: get easier again once randr1.3 hits
KeefR: and it worked with the onboard intel quite well, so I tested it with nvidia and ati...
KeefR: and that wasn't nice ;-)
bridgman: yeah, randr works nice with ONE gpu today, kinda not so nice after that
KeefR: but it's really encouraging, that ati opens the documentation
KeefR: so there's hope, that the support will be better
KeefR: i wonder when nvidia will realize, that it looses its advantage on linux because of it's proprietary driver
KeefR: some years ago matrox was the way to go with their g400 dual-head cards, then it was nvidia but now it seems to shift to ati
bridgman: it's going to be a tough call for nvidia; there are obvious benefits but there
Alpo\: Now i have compiled the xf86-video-ati and mesa sources.. How can I install them in Ubuntu without causing problems?
bridgman: are big costs and risks as well; if we hadn't also been a cpu company I'm not
bridgman: 100% sure that going open source would have worked out for us either
adamk: Alpo\, It's not going to be easy... You'll need to install the DRM from git and, if you want AIGLX support, a newer version of the Xserver, too.
KeefR: well, there was no reason one year ago to go open source
bridgman: agreed
Alpo\: adamk: I have Xorg 7.3
KeefR: 2 vendors, both with only proprietary drivers
adamk: Alpo\, I still think you need something newer.
Alpo\: and I have downloaded mesa/mesa from git
Alpo\: and mesa/drm
adamk: Shoot... BRB.
KeefR: and one year ago I told all my friends to buy nvidia cards for linux, because of the better driver support, now it has changed, atm I say: wait and see ;-)
bridgman: Alpo\: I think adamk is right, sorry I had forgotten about the xserver
bridgman: I'm just doing a quick search for current build / install instructions, wait one
bridgman: this seems pretty current : http://ubuntuforums.org/showthread.php?p=5146504
Alpo\: bridgeman: do i have to change the line ./autogen.sh --prefix=/usr --with-dri-drivers=radeon,r300 ?
Alpo\: i have RV5xx card
Alpo\: nevermind i just foundout that option r500 doesnt exist
nha: --with-dri-drivers=r300 is enough
nha: radeon is r100 only
nha: (I assume this is for building Mesa)
Alpo\: installed,booted and no DRI
Alpo\: and the /dev/dri/ is empty
osiris_: nha: I added new bug (#16452) to bugzilla about texture problems I was talking about few days ago
bridgman: Alpo\: anything interesting in the logs ?
Alpo\: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
Alpo\: and before it: /dev/dri/card0 no such device
Alpo\: I modprobed radeon and drm but no success
glisse: Alpo\: drm is not loaded
glisse: Alpo\: maybe too old drm
Alpo\: i just compiled and installed it from git
glisse: Alpo\: there should be somethings in your syslog about them loading
glisse: and if they fail why
Alpo\: nothing about radeon but drm is loaded
glisse: Alpo\: than radeon don't know about your device id
Alpo\: How can i set it?
Alpo\: i did 'rmmod radeon drm' and 'modprobe radeon drm', still only drm shows up in syslog
glisse: Alpo\: edit device*txt in shared-core
glisse: and add your device id their
glisse: lspci -n shoult help you to find it out
Alpo\: where is that file?
glisse: dmr/shared-core
glisse: you have to recompile & reinstall after
Alpo\: there is no .txt files
Alpo\: should i create one?
glisse: drm_pciids.txt
nha: osiris_: seen it, posted to the bug report right now
nha: osiris_: come to think of it, crossbar is probably not a good test, because it uses a uniformly coloured texture
nha: osiris_: can't think of a good pre-built test right now, though
nha: osiris_: the cubemap test in piglit could be interesting (using the 'd' key to cycle through different cubemap sizes)
egone: zwischenstand: http://www.egore911.de/vype/rl/storage_tent.jpg
egone: doh! wrong channel :-)
nha: nice model though :)
egone: nha: thanks :-) it's for open anno
ut2004player: Hey All
spstarr: airlied: congrads!
Alpo\: I have problem with DRI with X1950PRO. I have compiled and installed mesa,drm and xf86-video-ati from git but there are nothing in /dev/dri/
nha: Alpo\: check whether the drm was loaded (should appear in dmesg / kernel log), and check your Xorg.0.log for DRI/DRM-related messages
Alpo\: drm was loaded, and in xorg.0.log were messages about /dev/dri/card0 not found, dri disabled
nha: interesting
nha: I don't know who/what exactly is responsible for creating that file, but as long as there was no error during loading of the radeon.ko kernel module, the file should just be there
Alpo\: hmm
Alpo\: I reinstalled and modprobed drm and radeon
Alpo\: and now the file is there
Alpo\: maybe i just missed something in compiling&installing
Alpo\: but now i have problem with AIGLX
Alpo\: in Xorg.0.log it says (EE) AIGLX error: dlsym for __driCreateNewScreen_20050727 failed (/usr/lib/dri/r300_dri.so: undefined symbol: __driCreateNewScreen_20050727)
Alpo\: (EE) AIGLX: reverting to software rendering
adamk: Did you compile xserver from git like we said earlier? :-)
Borgvall: aiglx only works for xserver>1.4.99
Borgvall: however installing this is non-trivial
Borgvall: maybe you should wait until your distro packaged it
ut2004player: I was wondering does ati have a CUDA alternative/equivalent?
ut2004player: CUDA can do some amazing things.
adamk: ut2004player, CTM (Close To Metal).
adamk: ut2004player, http://en.wikipedia.org/wiki/Close_to_Metal
ut2004player: Yes, i have heard of that, but its discontinued and no longer supported i believe.
ut2004player: also
ut2004player: do ati cards support folding@home?
RTFM_FTW: yes
nha: ut2004player: as that wikipedia article tells you, they're calling it FireStream now
RTFM_FTW: and AMD / ATI et al has CTM (mentioned above -- specific to the R5xx et al series) as well as CAL (the replacement to CTM supported on R6xx et al)
nha: well, that's the product name anyway
RTFM_FTW: CAL et al is supported through high-level languages / environments like Brook
RTFM_FTW: also note the following AMD press release -- http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543~126593,00.html
ut2004player: "The ace that NVIDIA has is PhysX support starting right now. ATI of course has support with HAVOK, but where are the drivers?"
dorino: GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!
dorino: GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!! GERBILSOFT I LOVE YOU!!!!
GerbilSoft: dorino: this channel is for the open-source radeon driver, please take your spamming elsewhere
dorino: OPEN SOURCE??? WHAT'S IT?
GerbilSoft: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-ati.git
dorino: WHO CARES.
BigBrain: dorino: Take a look at the wikipedia article on that, but don't spam please
BigBrain: we do
dorino: BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAIN BIGBRAIN THE LITTLE BRAI
BigBrain: tzz, I'm off
dorino: I love you. I want to marry you BigBrain
dorino: LIBV YOURE GAY
dorino: BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG BY DORINO OMG
cokane: omg leave
cokane: or gag yourself
dorino: COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE COKANE = COCAINE
spstarr: dorino: stop being a script kiddie
spstarr: you're not getting ops
spstarr: and you sure aren't taking this channel over :-)
dorino: NO I'm GETTING OPS
dorino: AND I'M TAKING THIS CHANNEL OVER THANKS TO MY NORTH KOREAN FRIENDS
GerbilSoft: spstarr: he's a script kiddy who doesn't even know what OS he's using, and thinks that /whois is a special hacking tool.
spstarr: hahahahah
spstarr: dorino: uh huh... and im the present of the United States
spstarr: president
dorino: WHO CARES.
dorino: I CAN HACK YOU ALL
spstarr: dorino: please, take your games elsewhere, im bored
dorino: ONE OF MY FRIENDS IS KIM JONG IL'S SONS
spstarr: dorino: i dare you, you won't get far on my box
dorino: FEAR ME
cokane: yeah you and your n. korean friends
GerbilSoft: dorino: saying that isn't especially smart, especially in a channel for a linux driver :)
dorino: FEAR NORTH KOREA
dorino: What's linux?
dorino: a command prompt
spstarr: dorino: c'mon hack into my boxes...DO IT
dorino: ?
GerbilSoft: an operating system.
silverpower: noob.
spstarr: im in the mood for some entertainment/fun
spstarr: DO IT
spstarr: some me what your made of
spstarr: show
dorino: SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN SPSTARR THE GAYMAN
cokane: dorino: "dorito" has a "t" in it
spstarr: ugh im typing wrong words today
spstarr: dorino: so?
cokane: heh
spstarr: done
cokane: I guess he went to confer with his "n korean" friends
spstarr: see, just play into their game and they go away
dutchfish: hi, can someone help me with radeon HD 2400 XT on debian linux lenny, with kernel 2.6.25-7. I am trying to get VLC working without stutter and so high CPU load. Its a r600 chipset. Tried xserver-xorg-video-ati as well radeonhd. problem still exists
dutchfish: i have no working dri (yet)
dorino: ___THE DORINO___
dorino: ___AGAIN___
dorino: LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV THE FUCKER THAT SUCKS THE COCK'S CUM LIBV
dorino: OH YOU NOW
dorino: GIMME USERLIST OF #supermariowar
t: dorino: hi.
dorino: PEOPLE JOIN #supermariowar
dorino: FIGHT THE STUPIDS
t: dorino: could you please stop spamming freenode channels? :)
dorino: YOU AN IRCOP? RIGHT?
t: I am.
dorino: FUCK YOU. HAHAHAHAHAHA
dorino: DORINO BEAST YOU
dorino: THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY THE IRC OP HAS TALKED BUT HE S GAY
spstarr: t: haha :)
spstarr: danke
t: np :)
dmb: ot
ddoorriinnoo: THE IRCOPS DIDN'T DEFEAT THE DORINO
ddoorriinnoo: GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE GAY GERBILSOFT THE
ddoorriinnoo: THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCKED THE IRC OPS CAN BE FUCK
GerbilSoft: *sigh*
ddoorriinnoo: THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENODE THE DORINO RULES FREENO
D0r1n_the_Ruler: DORINO DORINO
D0r1n_the_Ruler: MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS MIGAS
D0r1n_the_Ruler: IRC OPS CAN BE FUCKED THIS IS MY TERRITORY NOT YOURS
D0r1n_the_Ruler: FUCK OFF
D0r1n_the_Ruler: I RULE FREENODE NOT YOU
D0r1n_the_Ruler: THE IRC OPS CAN BE FUCKED
D0r1n_the_Ruler: THE DORINO RULES EVERYTHING
spstarr: you're boring
spstarr: yawns at D0r1n_the_Ruler
peterz: damn kids
spstarr: yep
lupine_85: back when I was a lad, irc idiots were worth looking at
lupine_85: things ain't like what they used to be
lupine_85: "zomg I know how to use cgi:irc"
General_Guy: AH AHA HHAHAH THE DORINO!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
spstarr: zomg lolz wtf i r waz teh kool thing
spstarr: Dorino.. Dorito chips shoulds better :-)
spstarr: they make good tortilla chips
GerbilSoft: mmm doritos
GerbilSoft: has a bag of the zesty taco / chipotle ranch chips
GerbilSoft: :D
spstarr: yup
mikkoc_: someone should seriously take a look at this leak i'm having: http://zlin.dk/p/?NjRlZTE4.txt
mikkoc_: it's pretty big i think
mikkoc_: im off to bed for now
chori: not sure If I should ask this here or in #xbmc but here goes. Trying to get xbmc and pulse to play togheter. (Pulse seems to work with everything else, but migh be configured wrongly anyway). xbmc complainas about trying to create a passthrough and failing. At the same time a Pulse message appear saying: *** PULSEAUDIO: Unable to create stream. Any ideas on this?
rx__: heh..
rx__: spammer was on the loose?
spstarr: rx__: you like doritos too no? ;-)
GerbilSoft: he's actually still in here under two names
rx__: sighs
D0r1n_the_Ruler: DORINHO!!!!!!!!!!!!!
D0r1n_the_Ruler: GERBILSOFT TELL #supermariowar THAT IM GOING TO RAID THEM
spstarr: go do that then
spstarr: type /join #supermariowar and /part #radeon
GerbilSoft: what type of RAID
GerbilSoft: RAID-0, RAID-1, RAID-5, RAID-6?
spstarr: hag
spstarr: hah
chori: well I'm no spammer but I actually solve my problem myself. There's a bug in libasound2 for hardy :)
spstarr: at least our little friend has calmed down :-)
damentz: hey airlied, i got exa to perform wicked fast on my r100 card
damentz: FBTexPercent -> 0
damentz: AccelDFS
damentz: works great now
damentz: then i set agpsize to 64 so that opengl apps run the same
damentz: i guess you can put that in the manual
damentz: users of exa should set fbtexpercent lower
damentz: if they have low videoram
damentz: in my case, 32mb
airlied: yeah its a tradeoff.
damentz: not really actually
spstarr: airlied: congrads on baby coming!
damentz: i tried playing warsow w/ it set to 0
damentz: same speed
airlied: damentz: wierd I wonderwhere the textures go
damentz: haha ya me too
airlied: damentz: it sounds like it'll be sw rendered.
damentz: i have AGPSize set to "64"
airlied: as there is no way to put the textures into card memory.
airlied: spstarr: thanks.
airlied: damentz: oh r100
damentz: yes
airlied: damentz: I forgot r100 does texture from GART
airlied: none of the other drivers do that
spstarr: airlied: some interesting performance improvements.. does MigrationHeuristic = greedy and EnablePageFlip make that much a difference?
damentz: hehe
damentz: oh really?
spstarr: airlied: if so, composite is working _WONDERFULLY_ fast
airlied: spstarr: greedy just means use CPU mostly
damentz: airlied: i don't have that enabled
airlied: spstarr: I didn't think pageflip works.
damentz: i have pageflip enabled...
spstarr: it seems to be(?)
damentz: should i turn that off for exa?
airlied: spstarr: as in it may do nothing or it may not.
damentz: especially 3d apps?
airlied: damentz: shouldn't affect exa
damentz: ok
spstarr: airlied: i am noticing faster redrawing of windows when they are offscreen
spstarr: more smooth composite
airlied: spstarr: wierd is cpu usage much higher?
spstarr: even full screen windows not much
damentz: airlied: someone told me that you guys might allow exa to use gart for texture storage
airlied: as greedy just uses the CPU
airlied: damentz: the plan is to get a memory manaer that nca take care of it all
airlied: pixmaps will go wherever they need to .
damentz: ohhh
damentz: thats cool
spstarr: X is using like 20% or less
damentz: airlied: maybe it's all sw rendered without that migration heuristic
damentz: kind of like unaccelerated graphics
spstarr: airlied: this is fine as r300 gets more gaps filled in
damentz: i'm starting to think that ati's cards will have the best open source drivers
damentz: my r100 card is heads over tails faster than intel's cards even using mesa these days
damentz: you guys are doing an awesome job!
spstarr: with greedy then composite is usable, i haven't seen kwin cause a DRI lock recently.. but im not actively trying to encourage it to :)
spstarr: airlied: moving xchat window around under other windows X shows < 15% cpu use
damentz: spstarr: hmm, my cpu usage is pretty high
spstarr: i am using the latest builds from koji
spstarr: but a F9 kernel since the ones in rawhide are debug enabled and dog slow
spstarr: this is fine if im connected with AC, but for OLS i'll turn off composite (that reminds me.. I need a new laptop battery)
damentz: airlied: http://paste.debian.net/7130/plain/7130
damentz: thats my xorg.conf that i use for exa on my radeon 7500 mobility 32mb
damentz: works pretty well on 24bit!
thoraxe: hi... so question -- using xrandr i can get output going to the tv but its... squished... as if either the pixel size is off or the refresh rate is off or something
joneal: hi
damentz: thoraxe: probably the wrong resolution
damentz: try using something like grandr
damentz: set the resolution to 640x480, that should be compatible
thoraxe: damentz_: grandr??
thoraxe: it definitely did not like 640x480
thoraxe: lots of errors