Nightwulf|work: hi all
shepherd: Hi, should the RS690M have 3d ?
shepherd: I installed radeonhd and have an RS690m card but do not have hardware acceleration, is there something special I need to do ?
udovdh: I will move from rv630 to rv635, any differences except die size?
libv: udovdh: yeah, whole modesetting engine changed
udovdh: isn't rv630 vs rv635 just a die size change?
udovdh: (gogoled it a few days ago)
udovdh: but the card just has dvi
udovdh: and no cooling
udovdh: so silent for me
udovdh: I can dig up the box, it's an asus card
udovdh: gigabyte only had cooled cards in the store I checked
udovdh: will upgrade to a new mobo maybe this weekend
udovdh: so pciE 2.0 and ditto card
udovdh: (old mobo died a bit so thats the reason)
udovdh: but same xorg.conf should work?
Drewdoo: Hello! I have an Asus EAH3650 Silent Magic 512mb
Drewdoo: Is there a kext for this around?
Drewdoo: I've got it working... but screensaver is very slow, and dashboard "blending" is very slow too - so I think I'm not using the right kext
adamk: Drewdoo, This channel is for the open source radeonhd driver for Xorg. Nothing to do with OS X :-)
cxo: kext? are you using BSD?
cxo: Oh. Broken BSD
Drewdoo: Sorry, didn't realise this was the wrong place :)
cxo: This driver is not going to work on OSX
Drewdoo: Guess I ... better go. hehehehe
Obscene_CNN: shepherd, I'm not sure but I think the rs690 has accel support.
shepherd: Obscene_CNN: yea I read that the rs690 was supported, but glxinfo | grep renderer returns "OpenGL renderer string: Software Rasterizer" does it matter that my model is rs690m ?
Obscene_CNN: I don't think so. You may need to modify your xorg.conf file and possibly change some things in your kernel config.
shepherd: Obscene_CNN: I've not much experience with this.. any idea where I can get more info ? on what might need changing?
Obscene_CNN: look up howto setup DRI
shepherd: Obscene_CNN: Hmm.. it says I have to add 'Option "DRI"' to my xorg.conf to the driver section, does that mean I create a Section "driver" and put it inside?
shepherd: I tried putting it in other areas like section "device" and my x just freezes afterwords
yangman: shepherd: it goes into the Device section
shepherd: yangman: ah.. any idea why my X just freezes :(
yangman: shepherd: running 3D apps or the server doesn't start at all?
shepherd: yangman: as soon as I try and restart X it with option "dri" in the device section, everything freezes
yangman: shepherd: what does the log say?
shepherd: yangman: Where might I find the log ? :p
yangman: X always keeps a backup of the previous session's log
shepherd: alright one sec
shepherd: well I was going to paste it but my firefox froze because it appears to be
shepherd: 100 thousands lines of (EE) RADEONHD(0): RHDDRMCPBuffer: drmDMA returned -9
shepherd: 998k lines *
shepherd: I don't know if that helps, I don't know what to make of it
yangman: shepherd: that's giving a 500 error
shepherd: yangman: What does that mean
yangman: the pastebin you've used is giving a server error
shepherd: haha now it is for me as well..
shepherd: I pasted everything up until it starts with that error
yangman: what kernel version are you running?
yangman: to be honest, I've no idea
yangman: wait until someone more knowledgable speaks up
shepherd: Alright, thanks. I'll stick around.. thanks for the help anyway though
yangman: it looks like you've got the right things installed, so it's probably either a known bug that's fixed in master or there's a quirk with the combination of software and hardware you're running
Jello_Bear: hello I"m a gentoo user, I was wondering if radeonhd supports the 780
Jello_Bear: the 780G
Obscene_CNN: It supports it but no accel yet
Jello_Bear: thank you
revx: I find myself looking into getting mesa running faster
revx: doing software rendering on my cpu
revx: my r700 is just a relativly fast framebuffer ;)
revx: right now I wait for callgrind on ioquake3
revx: (I was having trouble getting mesa built with -pg to give me anything useful)
revx: cxo: I'm a professional student with a four week break starting next wednesday...
cxo: When are we going to get 3d for r700?
revx: my thought: software mesa ought to be able to run quake3 on my current cpu...
revx: cxo: well it's being done on a private tree right now
cxo: it wont, it will be slow
revx: apparently we're waiting for stuff to clear legal at AMD
revx: cxo: no really?
cxo: you wont get more than 20fps at 1280x1024
revx: the goal is to do better than 1.1fps :)
Honk: try 320x280 ;P
revx: cxo: well I'm starting at 1.1fps at 640x480!
Honk: that should give you a nice boost ;P
shepherd: Honk: good times..
revx: it runs really fast at 1x1 :)
Honk: shepherd: yeah, though it wasnt quake3 back then ;P
cxo: 12mb cache, xeon, might give you 20fps
revx: cxo: not with the current mesa tree
cxo: Linux sucks for anything 3d
revx: I disagree.
Honk: revx: good luck increasing that by almost 2000% in a month ;P
revx: Honk: well I'm waiting for valgrind right now... taking FOREVER
revx: 1.1fps standalone...
revx: getting 36 SECONDS/frame in valgrind
revx: but that's to be expected when doing software rendering :)
Honk: I know why I prefer to code in java ;P
Honk: profiling is almost free
revx: Honk: well demo four of quake3 has 1260 frames
Honk: ouch ;P
revx: I started this at 10am... so I'm expecting about 12 hours to finish
revx: but the data from this first run should give me lots of clues to the bottlenecks
revx: some I can't avoid, others I can
revx: I'm not trying to come up with anything for the mesa tree upstream... there will be ugly hacks
cxo: why dont you just setup log timestamps for all calls in gdb and watch it go?
revx: but if I end up with anything useful I'll see what those guys think of it
revx: cxo: err there's -pg to do stuff like that
revx: but they both have shortcommings
revx: callgrind in valgrind is going to provide some VERY useful information
cxo: valgrind is too naughty
cxo: its very nosey, it will take days to get the demo done
revx: cxo: ~12 hours
cxo: hence gdb, will be a more practical way of analysis
revx: cxo: yes, past this
revx: but I'm getting an intial overall image of what's going on from callgrind
cxo: unless of course, the 12hour log will give you data for the next year to chew
revx: cxo: it will give me cpu cycles in tons of execution paths and let me compare larger structures -- data that will still apply in many areas after I start changing things
revx: it's not like I need callgrind constantly here -- I'm doing it once right now with 1260 sample frames
cxo: Process Design Patterns. A relatively new topic in software engineering
cxo: Design not by logic, but by runtime efficiency
cxo: You know what would be funny? You wake up one morning and you see in the news that terrorists have kidnapped an ATi dev and will not release him until ATi releases a working Linux driver.
shepherd: That would certainly speed things uo
cxo: I would laugh lots. And hysterically. I might even aid in his kidnapping if I was asked to.
shepherd: That makes 2 of us
shepherd: I feel something coming on here..
cxo: I think we should start an Open Source Liberation Syndicate
cxo: Either that, or we get the talaban to buy ATi equipped Linux computers
cxo: Nvidia has come out with a very sleek new PCB for the GT200 processor (GTX260/280). I think I should try pawn my HD4870 off to a Windows user and get one of those
revx: cxo: and their specs are NEVER comming
cxo: Neither is ATi's
revx: cxo: I disagree
revx: they've already released far more than nVidia will ever do
revx: and I have a feeling r700 docs are comming soon enough
cxo: Life is like a realtime system. The longer you wait for a response, the less important that response becomes
cxo: At a certain, its not worth anything
afief: cxo, I think they are releasing specs faster than they create them, so at some point it'll be realtime
cxo: Nvidia doesnt need to release anything, when they have a very (very) working driver.
revx: cxo: wrong
revx: they have a working non-free driver
cxo: I dont have a beard, nor a multicolored tie dye shirt., so thats meaningless to me.
Guest24294: that driver is crap also
cxo: Everything is relative
Guest24294: hard locked my display at least once a day
revx: cxo: then what the hell are you doing in here/
afief: cxo, tell me when that driver doesn't break everytime the xorg developer decide to deprecate a function
cxo: afief, You are in the wrong shoes. I'm sure I'm safe in saying that the majority of users are NOT developers. and dont give 2 shits about whats what on their system, as long as it works. So put yourself into the shoes of a happy go lucky Linux user that wishes to play games
cxo: revx, funnily, I came in here to ask for fglrx help :)
cxo: I have a feeling i screwed up my toolchain. I was tracking git for a while, and now everything is a little wonky
cxo: By the way, is rotation meant to work with radeonhd?
cxo: (i get an invalid opcode error)
ndim: Hmm. Old reflexes die hard.
revx: cxo: AFAIK r5/6/700 are up on xf86-video-ati -- where rotation does work
ndim: has just typed "cvs up" to get the updates from radeonhd's git repo.
ndim: cxo: rotation works on radeonhd git. When using EXA, the rotate screens are even updated in acceptable speed, but that will conflict with 3D...
revx: wow.. done
revx: mcount is 15% of the cpu cycles
revx: (something it calls 12% of them)
cxo: So there is rotation in radeonhd for r700 ?
revx: -pg is expensive if 27% of cycles went to it :/
ndim: cxo: I have not checked the specific chips this http://cgit.freedesktop.org/xorg/driver/xf86-video-radeonhd/commit/?id=c1827740 and the following patches apply to.
shepherd: anyone know why I get this error: http://fpaste.org/paste/1538 ... it happens when I add option "DRI" to my section "device" part of xorg.conf ..
ndim: rotation should be more modesetting than acceleration related, though, so I'd guess...
cxo: well i'm running the latest git, and it doesnt work...
ndim: cxo: Eh :)
ndim: So I did guess wrong.
ndim: shepherd: Something in your xorg.org is not kosher.
ndim: shepherd: Can you pasted that?
ndim: shepherd: (Into th efpaste, not into the IRC channel)
shepherd: yea.. xD
cxo: DRI is a default setting, i think it will always be on
ndim: cxo: It is not on by default on radeonhd.
ndim: OK, let me rephrase that.
ndim: cxo: DRI is off by default on radeonhd.
shepherd: but like I said, it only happened when adding option "DRI" to my section "device"
ndim: shepherd: Option "DRI"
shepherd: and I need DRI... -.- :(
ndim: shepherd: Option "DRI" "on"
shepherd: Oh ?
cxo: Well you OBVIOUSLY HAVE A Bad file descriptor!!!
ndim: Any of those?
shepherd: I didn't try it with.. "on"
ndim: should not be necessary.
ndim: but still...
ndim: And write that line anew. Perhaps you had some non-standard whitespace or something.
shepherd: ok.. one sec
shepherd: You're right I musta made some kinda typo cause now I am getting a different error ( one I had seen prior though.. unfortunatly )
shepherd: I don't know what to make of it though
cxo: internal server error
shepherd: again.. ok one sec
shepherd: the last line repeats itself about 1kk times
ndim: I have seen this line in other bug reports.
ndim: Or at least something very similar.
shepherd: Yea... :(