EruditeHermit: side effects include death
bridgman: I guess this is one of those cases where "the best documentation is the source code" ;)
damentz: ya i suppose, that would be cool if they could figure out how to use that tweak and prevent compositing corruption
damentz: then 2d performance would be up to par with fglrx and nvidia
damentz: well, nvidia's beta drivers
EruditeHermit: so for OGL 1.3 support we are looking at 6 months or so
bridgman: probably less; my guess is April
bridgman: that assumes a bit of community contribution, schedule will vary depending on who pitches in
EruditeHermit: I assume 4850 is supported by fglrx with 3D
EruditeHermit: OGL 2.0?
bridgman: 2.1 I think but not sure
EruditeHermit: oh cool
damentz: bridgman: wouldn't it just be a good idea to get ogl 2.1 supported soon on the open source drivers?
bridgman: sure, but we need to get 1.3 supported first ;)
damentz: oh wait
damentz: i think i remember someone saying they needed a new memory manager first
EruditeHermit: I am sure they are working on it
EruditeHermit: memory manager is done mostly
bridgman: the memory manager is coming together pretty well (except on spstarr's system ;))
bridgman: there are a lot of very cool things happening in the x/dri world these days
EruditeHermit: fedora 10 uses it I think
spstarr: bridgman: hehehe
EruditeHermit: bridgman: any plans on using gallium for 3D in the open drivers? Would that speed up development?
revx: I heard someone may be working on a gallium driver for radeon?
EruditeHermit: they are about to merge gallium into mesa soon I think
EruditeHermit: revx: haha snap
bridgman: once that happens we'll probably jump over to gallium; my feeling was that going to gallium now would slow down initial 6xx/7xx 3d support, but gallium might make getting to 2.x faster
bridgman: all the code being done for mesa classic should port over fairly easily
damentz: bridgman: good point
damentz: bridgman: but these newer cards tend to be gaming cards anyway
damentz: and fglrx would benefit them more
revx: bridgman: doesn't gallium sufficiently separate video driver and graphics API in such a way that might even make 3.0 implementation relativly easier?
bridgman: yeah, it's hard to say; the available Linux games don't seem to stress the cards as hard
damentz: bridgman: well, there are a lot of smart folks that manage to get games that do ;)
bridgman: revx; assuming that someone else implements 3.0 over gallium, yes; if we have to do it ourselves, no ;)
EruditeHermit: well once you have a gallium driver, the other API support is handled by gallium itself
damentz: warsow with bloom, counter strike source through wine (which i hear will be native soon)
bridgman: damentz; when I see people complaining about 250 fps I start to have doubts
damentz: anything using shaders really
damentz: bridgman: well i'm testing the open source driver on the radeon 7500 mobility and 9000 mobility
bridgman: haven't played warsow, should look at it over the holidays
damentz: they could be a lot faster
EruditeHermit: bridgman: well, fast cards are good for 3D work using blender etc
damentz: bridgman: haha, i love that game, it's so hardcore, packs so much fun for being free
revx: damentz: counter strike source is a crappy benchmark for gpus
damentz: i'm playing it now right now
damentz: revx: yes, it is more cpu constrained
EruditeHermit: which is what I need it for
damentz: terracon: ah, forgot about taht one
damentz: there really isn't anything else i play
damentz: maybe openarena
damentz: but it's not as fun as nexuiz or warsow
bridgman: I don't think any of those will stress a 48x0 unless you run big screen max AA max AF etc..
damentz: bridgman: really, the only thing holding the open source 3d graphics back are the shaders
damentz: either they crash the game
terracon: and there's also suaerbraten/cube 2 is pretty stressful
bridgman: yeah, but shaders usually come along with vbos and fbos, and somewhere in there you need memory management
damentz: or you get 30 frames per _minute_
damentz: terracon: hehe, forgot about that one too
revx: damentz: hah, I was getting 36 frames per minute when profiling ioquake3 :P
bridgman: we definitely need glsl support, and I don't think that needs memory management, but I don't think we can get to the next level of games without mm
EruditeHermit: bridgman: are we taking up your weekend with questions? =p
bridgman: so far, yes
revx: 36 seconds per frame* (~2frames/min)
EruditeHermit: bridgman: lol, thanks for answering them
EruditeHermit: bridgman: we appreciate your efforts
bridgman: no prob
spstarr: bridgman: we'll figure this out
bridgman: this is a fun project
damentz: bridgman: haha, definitely
damentz: every two months i see something new i can do with my old laptops
EruditeHermit: bridgman: my goal for this year is to learn enough to contribute to the radeon driver somehow
EruditeHermit: bridgman: fix some small bug
EruditeHermit: or something small
damentz: bridgman: just an off the wall question, do the radeon developers and the fglrx developers communicate any?
damentz: or is that against a contract of some sort
damentz: like clean room environment
spstarr: bridgman: it was fun, but now i cant use eyecandy ;p
bridgman: it's a good time; I think a year ago we were down to maybe 4 people who understood how the code works, now we're back to 10 or more... makes it a lot easier to get help when you're just getting started
bridgman: they communicate but open source devs can't see closed driver source in general
damentz: bridgman: so, they can't share code, but they can help each other right?
bridgman: probably 60% of communication is with the hw folks
bridgman: yep, lots of "help each other"
terracon: fglrx is probably mostly windows than they try to port it to linux
bridgman: spstarr; it's always something simple, the trick is finding it
bridgman: terracon; it's actually a bunch of code modules which get wrapped up differently for each OS; more than just windows & linux
damentz: ah i see
damentz: so platform agnostic modules
damentz: and then you have interfaces for different OS's
bridgman: as much as possible, yep
damentz: bridgman: really the only thing that's still annoying me about fglrx is the xorg and kernel support
EruditeHermit: just like gallium
damentz: they seem to always be a step behind
bridgman: yeah, we try to keep os-specific stuff outside the common code
damentz: and being on debian sid, it almost never works
EruditeHermit: damentz: being on debian sid is hard because its a moving target
spstarr: bridgman: that helps, having the video driver in kernel attracts kernel people
EruditeHermit: damentz: changes everyday
damentz: EruditeHermit: yes but, how is nvidia doing it then?
bridgman: given that we make most of our money selling into systems running RHEL and SLED/SLES it's kinda hard to be focused on the bleeding edge; one of the benefits of doing more with open source is that we can "be" the bleeding edge
damentz: the moment one of the kernel hit's rc, their driver already works
EruditeHermit: damentz: I am sure nvidia drivers break too
damentz: or when a new version of xorg gets into beta, they release a new version
bridgman: they started supporting consumer users earlier and have had more time to beat the driver into shape; once you get there it tends to stay there
damentz: EruditeHermit: but never where the software is already stable
revx: bridgman: I'd like to point out that you arn't doing more with open source *yet*
damentz: they usually get support somewhere in beta/rc stage
moses: how do i get my comp to pump out higher quality video?
revx: bridgman: same-day specs are what I'd like to see.
revx: (say when R800 comes out.. specs come out!)
damentz: bridgman: ahhh
bridgman: damentz; sure we are; 15 months ago we weren't doing anything ;)
spstarr: we really should have moved the video driver modes to the kernel - years ago - ;/
bridgman: sorry, revx not damentz
spstarr: but at least, now we are
revx: bridgman: indeed, and I'm sure there's some red flags at AMD related to this
damentz: bridgman: well, that ubuntu stunt made it obvious that maybe you guys are a little too late with xorg and kernel support
bridgman: for future gpus we can do same day specs; before that, my time machine was in the shop
bridgman: which ubuntu stunt ?
terracon: hmm yeah what ubuntu stunt
damentz: an 8.10 hotfix was released for ubuntu 8.10
damentz: but wasn't released on the main site
damentz: to get xorg 7.4/ xserver 1.5 support
bridgman: why should it be ? It was only tested on ubuntu ?
revx: but AMD has now put themselves in a unique position: same day specs (or even support by having specs before that) would be really interesting...
bridgman: that's the plan; not necessarily same day but Alex is already sitting in on future GPU design meetings
revx: BSD & Linux could be on "the bleeding edge" in an area where they've always lagged
bridgman: revx, not sure what you mean re: red flags
revx: bridgman: I've never met a bussiness major that was strongly receptive of any sort of open source model
damentz: revx: haha, especially a business i work for
damentz: they distrust open source which is so strange to me
bridgman: you can find them, but the benefit model is pretty complex
revx: and I can't imagine that there's a lack of bussiness majors working in management at AMD :P
EruditeHermit: AMD has been better than some with open source
bridgman: actually these days a lot of management came up the engineering side
EruditeHermit: probably only because AMD bought out ATI that we have what we have now
revx: bridgman: will any of the UVD or UVD2 stuff be appearing with the 3d docs?
damentz: EruditeHermit: what are you talking about, amd is kicking ass with open source with ati's gpu's
damentz: it's a late start, so it doesn't look effective right now
bridgman: to some extent; if we were only selling GPUs it would be tough to justify supporting open source; it's the tie-in to CPU sales and being able to sell platforms and enterprise-wide where the justification comes in
bridgman: revx; no just 3d engine, no UVD
EruditeHermit: damentz: ATI when only a GPU manufacturer didn't release docs. Only after AMD bought them is ATI stuff working better
revx: bridgman: not sure that really effects me yet as I found before (on r500) I didn't have too many issues decoding four 1080p streams in software to the sides of a desktop cube in compiz
bridgman: we actually used to release more docs than anyone back in the r200 days; we stopped as the cost and risk got higher; if we weren't selling CPUs too I don't think we would necessarily have been able to justify doing more with open source
damentz: bridgman: ohhh, that's why the r100 and r200 had useable support for 3d
bridgman: revx; yeah, as long as you have Xv that takes care of the biggest CPU hogging; if someone felt like writing XvMC that would make another big dent
revx: bridgman: but the XvMC is where UVD would be nice.. I'm sure one could write a universal shader to do it...
revx: but having hardware there for the job and not using it just seems wrong :P
bridgman: yep; part of the reason was that the weather channel wanted to do some spiffy things with linux and our cards, so they either funded or helped with funding for drivers and we provided documentation; I wasn't involved back then so not sure of details
EruditeHermit: bridgman: a lot of animation studios use Linux machines
bridgman: you don't need UVD for XvMC; it's done on the shaders anyways; the 5xx docs have all the info you need and so will 6xx
bridgman: it's just that nobody has bothered to write xvmc
EruditeHermit: bridgman: I am sure Linux support is something that more people would be interested in if there were better drivers
damentz: EruditeHermit: probably proprietary drivers though
terracon: weather channel contracted tungsten graphics to write the driver
revx: bridgman: then what does UVD provide?
moses: whats the setting to pump out a higher video quality?
moses: anyone know?
bridgman: pretty much all proprietary drivers; they like having open source as a fallback in case of some other problem (network or whatever) so nobody complains about the heartbreak of kernel taint
bridgman: uvd provides very low power consumption, lower than cpu or shaders can do on their own
bridgman: bluray on airplanes is the killer app
revx: bridgman: IFLC paying for that? :P
EruditeHermit: so they won't release UVD specs?
bridgman: orly ? iflc ?
revx: IFLC is the component of AIG that leases corprate jets to companies, worth over $50bn
bridgman: Not a question of "won't". We're going to look at whether it's possible without violating other agreements; if so we will release info. For now I am saying "assume no" 'cause I don't know what will happen
bridgman: now that all the airplanes are getting inflight entertainment I'm not sure how much the demand will grow; personally I would prefer legroom over inflight movies but that's just me
revx: bridgman: I havn't flown since March but I saw what Delta was doing
damentz: bridgman: so politics is regressing progression of the open source drivers
revx: they had a flat screen embeded in the seat in front of me.. pay per view movies...
bridgman: not politics, business
bridgman: we sign a bunch of agreements in order to get access to the largest markets, and some of those agreements get in the way of doing some things in open source
terracon: cough .. Microsoft
revx: speaking of.. it would be really interesting if a gallium driver for Radeon appeared
bridgman: yeah, but it really goes further than Microsoft. MS and Intel tried to come up with standards that would be sufficient to convince the content providers (cough, Hollywood) to allow playback of first DVDs, then HD content, on PCs
revx: and suddently DirectX 10 interface for gallium...
bridgman: I think Jerome had a basic Gallium driver running on r300 early in 2008; didn't do much but I think it drew triangles
revx: bridgman: well the important step being DX10->Gallium :P
revx: I have no doubt radeon driver for gallium is going to come together quickly rather soon here
revx: (after gallium is merged that is)
revx: if the Wine guys targeted gallium I think there would be a very powerful effect..
bridgman: that would be interesting; probably easier to make it work well than translating to OpenGL, but a lot more work
bridgman: they would basically have to write an entire DX driver
revx: bridgman: they're writing DX10 now
revx: AFAIK anyways
bridgman: for wine ? are they actually writing a full driver or just dx10 to opengl ?
revx: dx10 o gl
revx: to gl*
bridgman: opengl is kind of at the same level so you don't have to write as much
revx: just like dx9 etc that they have
bridgman: haven't played with Wine much; I'm one of those wierd people who plays Windows games on Windows ;)
RTFM_FTW: doing D3D10 is pretty easy if you have support for EXT_texture_integer, EXT_gpu_shader4, ... through your GL runtime
bridgman: sigh... so many extensions ;)
RTFM_FTW: mmm those two are quite fun to implement (heh)
bridgman: yeah, I guess it would be no fun if they were easy
revx: bridgman: Goga777's on top of it -- next week we'll have them!
RTFM_FTW: heh you need a GLSL runtime first
RTFM_FTW: and a shader compiler :D
Goga777: revx, what do you mean ?
revx: Goga777: suanle... I was kidding.
EruditeHermit: bridgman: so is there a possibility that AMD will pull open source support in the future?
EruditeHermit: bridgman: or are they firmly in the mindset that they will support this going forrward
revx: EruditeHermit: OGD-1 will overtake by then :P
bridgman: I think it will continue, but this isn't a "mindset" thing; every year we look at risk/benefit/cost and decide what to do
EruditeHermit: I live with trepidation that I will buy a card and AMD will drop support
EruditeHermit: I've been waiting and waiting to see what happens
EruditeHermit: but I need a new box soon
EruditeHermit: can't wait much longer
Goga777: bridgman, is there any news about UVD support - in Catalist ? or opening of specs for uvd ?
bridgman: well, we can't go out and take back what we have already published, so just go on the basis of what is available today
bridgman: or tomorrow
EruditeHermit: Ideally I want to wait till the specs for r600/700 are published
EruditeHermit: but I am probably going to buy one tomorrow
EruditeHermit: out of necessity
EruditeHermit: my box is 7 years old
EruditeHermit: and my laptop 4.5
bridgman: Goga777; I can't comment on fglrx futures; for open source I just won't know until we spend some time on it; top priority is 6xx/7xx 3d followed by some basic power management; will look at video decode after that
revx: EruditeHermit: there's some really intersting techniques being sued this day and age when a fibre optic line is cut using neural nets. The same type of thing might be useable for partially automated reverse engineering of future closed cards :)
bridgman: 7 years is old; I think 6xx/7xx is a pretty safe bet
bridgman: my group is funded for 2009 if that helps
bridgman: just don't assume UVD, OK ? I really don't know what will happen there
EruditeHermit: yeah, I am probably going to get 4850 with the hope that r600/700 docs come soon
EruditeHermit: well, as long as video playback is done decently with something else
bridgman: that would be my choice; you can do a lot of cool things with 800 ALUs
revx: bridgman: I picked one up for that very reason
revx: bridgman: now waiting for docs on it :P
EruditeHermit: yeah its quite powerful for a decent price
bridgman: keep fingers crossed but we're getting close
RTFM_FTW: start using shaders... you have massive amounts of compute power to play around with
RTFM_FTW: at a minimum you should get quite a boost just by offloading color-space conversion (i.e. YUV 4:2:0, YUV 4:2:2, YUV 4:4:4 -> RGB) which is fairly trivial to do WRT implementation
EruditeHermit: I was talking to someone from another company, and they said that they asked for $1million for other companies to see their source
bridgman: the worst case for video would be XvMC on shaders and the rest on CPU; just go with a socket that can take a quad core even if you start with a cheap dual-core today
EruditeHermit: bridgman: probably going to get X58 with core i7
bridgman: yeah, I think that would be fast enough ;)
EruditeHermit: I don't like to buy often
bridgman: AGP, right ?
EruditeHermit: I have 3 computers right now
EruditeHermit: all AGP
EruditeHermit: Centrino 1 laptop
bridgman: yeah, definitely time for an upgrade
EruditeHermit: there have been 3 generations of intel CPUs since then
revx: bridgman: do you happen to know anything about the expected lifetime of socket am2+?
bridgman: my main office machine is a PIII-500 AGP; that's next on the upgrade list
EruditeHermit: I honestly wouldn't need to upgrade but for 1) flash player is really slow and sucks 2) video drivers that fully used my card would make it just about enough to run blender 3) my battery is dead
bridgman: linux box has a Phenom and a 4870; something is backward there
revx: EruditeHermit: NiCd?
bridgman: the little lithium thingy that keeps the CMOS RAM going
bridgman: I think most PCs are replaced when the battery goes
EruditeHermit: bridgman: oh I am talking about the laptop
EruditeHermit: battery doesn't charge
revx: bridgman: you can get replacements for those at walgreens :)
EruditeHermit: the time chip battery is easily replaced =)
revx: EruditeHermit: is it lithium ion, nicd, or what?
EruditeHermit: I don't know
EruditeHermit: Inspiron 8600 laptop
EruditeHermit: getting a new laptop soon too
revx: if something might explode on my lap I want to know what it is before I set it there :P but that's me
revx: bridgman: intentionaly so -- I have a few nasty habits
EruditeHermit: revx: it hasn't exploded in 4 yearsr
bridgman: looks like lithium-ion
revx: bridgman: I purchased a phenom 9500 with the intention of upgrading to Fusion if it targeted AM2+, but if that's not likely I'll get a phenom II
EruditeHermit: waiting for Dell to refresh their laptops
bridgman: I run lithium-ion batteries in a lot of stuff around the house including most flashlights, but the rule is "if it starts doing anything funny throw it far away"
revx: bridgman: hence me asking about the expected lifetime of am2+ :)
bridgman: I'm not sure but my guess is that there aren't enough unused pins on AM-anything to handle all the extra stuff that comes with fusion
bridgman: that is not an official AMD statement, just my guess
bridgman: phenom II should be nice assuming it gets the same big cache as the server SKUs
revx: ... please tell me they're not getting rid of socket F for Fusion
revx: it has over 1200 fscking pins!
revx: how many more can you need!
bridgman: don't know; socket F has a lot more pins AFAIK; the problem I see is that existing motherboards won't have anything hooked up to them (like video connectors)
bridgman: should talk to the Fusion folks more...
revx: bridgman: there's been a strange lack of any information regarding R800 as well
bridgman: yeah, we're getting better at keeping secrets
revx: bridgman: is the CPU side of Fusion based on previous work or is it more or less of a complete redesign?
EruditeHermit: bridgman: is r800 coming anytime soon?
bridgman: I think the idea is "based on" but optimized
EruditeHermit: bridgman: should I wait for it?
revx: EruditeHermit: I hear june '09
EruditeHermit: ahh too long
bridgman: I wouldn't wait for it
EruditeHermit: hard drives haven't really changed in the last 5 years
EruditeHermit: just got bigger
revx: you know that whole SSD comming to maturity thing doesn't matter
bridgman: yeah, it's good; every time my hard drive dies at work the new one is 5 times as big and costs half as much
EruditeHermit: hope SSDs don't fail as often
EruditeHermit: HDs fail every 2-3 years for me
EruditeHermit: laptop HDs
EruditeHermit: too much suspending and resuming
bridgman: AFAIK SSDs still wear out, but clever filesystem design can put that off for a while
spstarr: my laptop HD disk is backed up to a 500GiB disk :)
revx: and when they do wear out they're still readable!
revx: just not writable
spstarr: so if this dies can just scp the content over from a recovery CD
spstarr: revx: thats good and bad :)
spstarr: revx: especially if $HOME and /tmp are wanting to be written to
bridgman: interesting; I hear conflicting views about what happens after failure; if you can still read then that's not a big deal
spstarr: bridgman: if you can read you can just dd it to another SSD :)
bridgman: just keep a pocketful ;)
EruditeHermit: what happens if you are MI5
EruditeHermit: or CIA
EruditeHermit: thats a nightmare
revx: EruditeHermit: no
EruditeHermit: SSDs that you can't erase
EruditeHermit: have to melt them
spstarr: EruditeHermit: so you melt it
bridgman: ice crusher
revx: EruditeHermit: indeed.. a little bit of thermite goes a long way ;)
EruditeHermit: yeah, but people are lazy
EruditeHermit: I am sure governments won't do it
bridgman: but thermite is cool
EruditeHermit: McCain is selling blackberry's with his contacts list still on them
EruditeHermit: for $20
EruditeHermit: you can read all his text messages etc
bridgman: you're kidding; the BB is not wiped ?
bridgman: is he giving the password ?
bridgman: I think they wipe automatically after enough password fails...
revx: bridgman: a long time ago I learned data recovery means NOT running the OS on said storage device :P
bridgman: wierd; I was looking for leaked Deneb die shots and found a June 2008 review of an X1650 AGP, talking like it was state of the art
spstarr: bridgman: it would have been nice if the mobile radeons were replaceable
EruditeHermit: bridgman: http://gizmodo.com/5109038/blackberry-20-mccain+palins-contacts-priceless
spstarr: some sort of daughter card or something
spstarr: instead of being onboard
spstarr: bridgman: there's no bios to flash on these is there?
revx: bridgman: I'm cracking up reading that review
revx: "This graphics card is capable of displaying over a billion colors more than any comparable graphics card."
bridgman: on a laptop the video BIOS is usually stored in the same chip as system BIOS, so you need utilities from the laptop vendor to flash it
bridgman: goes back to the review... sounds good
bridgman: laptops did have daughter cards for a while, not sure what happened to them
spstarr: oh so only ibm bios stuff (i have the newest last bios)
bridgman: I don't know if IBM lets you flash the vbios and sbios separately, or one image with both, or just sbios only
spstarr: not sure, i never saw a vbios update
bridgman: wow, that's a really wierd review
bridgman: getting late, time to go
spstarr: bridgeman: night, see if you can find some rv350 errata info when you get some time this week :)
EruditeHermit: spstarr: want to try to crash your system? on an rv350 card, go to gizmodo on firerfox 3.1 nightly and reload a few times. Usually causes system lock after loading the 2nd or 3rd time.
spstarr: in kms mode or no?
EruditeHermit: non kms
spstarr: what distro
spstarr: it wont for me
spstarr: airlied fixed most of the lockup with EXA on rv350 non-kms
EruditeHermit: has to be ff3.1 nightly
spstarr: Ubuntu == old
EruditeHermit: spstarr: using latest master driver
spstarr: you need patches
EruditeHermit: from where?
spstarr: you may also need a new X server with EXA fixes
logari81: could anyone tell me why in this case: http://pastebin.com/mafd77bd the radeon driver couldn't be loaded?
rhodan: i'm using kde 4.2 b1 on gentoo linux with git-packages of all xorg-related stuff.
rhodan: i searched the IRC logs for qt4-small-pixmap corruption and it seems like this bug has been present for a few months. why isn't it fixed yet?
rhodan: < onestone> MrCooper: software fallback on radeon with accelDFS = ugly icons in qt4 apps
rhodan: (from Fri Jul 11)
rhodan: and it's still present :(
MrCooper: AccelDFS is known broken in some setups
MrCooper: but there hasn't been any progress on it in a while
MrCooper: maybe some PCIe setups are just as broken as AGP :}
rhodan: i have a X1400 on a thinkpad t60
rhodan: is there anything i can do?
hifi: you could always try the blob from ati...
rhodan: fglrx has bad 2d performance... at least for me
rhodan: i thought the r500 specs were open? why is it so hard to fix this issue?
MrCooper: why do you think it should be easy?
MrCooper: until someone finds the bug, maybe we should just disable AccelDFS by default (and possibly enable it on white-listed setups)
rhodan: the bug is not yet found?
rhodan: kde 4 without acceldfs is incredibly slow
mikkoc: rhodan: i also have kde 4.2 + gentoo + x1400, what corruption are you talking about?
rhodan: mikkoc: corruption with small pixmaps
MrCooper: rhodan: if we knew what the problem was and/or how to fix it, we would have fixed it...
felipec: I have the same issue as rhodan
felipec: the square boxes on Gmail are all garbled
felipec: this is an R520
mikkoc: lucky me then
felipec: I think it only happens when I use EXA and kernel modesetting is disabled
mikkoc: maybe it's because i use Exa?
rhodan: felipec: so kernel modesetting fixes it?
rhodan: why didn't you say?
rhodan: how do i get kms btw?
felipec: rhodan: kernel modesetting is not upstream, it's mostly on Fedora afaik
rhodan: e.g. what packages do i need?
felipec: and it makes Xv slow and weird in my laptop
mikkoc: i don't have kms
mikkoc: rhodan: you said you have xorg from git...
mikkoc: have you tried 1.5.3?
rhodan: mikkoc: i tried everything ;)
mikkoc: I have 1.5.3 from the x11-overlay
MrCooper: maybe AccelDFS just isn't used with kms
felipec: rhodan: are you using a T60 laptop?
rhodan: felipec: yes
felipec: mikkoc: and you?
mikkoc: dell inspiron 6400 + x1400
mikkoc: amd64 here
felipec: well, there's a difference right there: X1300 to X1400
mikkoc: he has x1400
mikkoc: how do i know if im using accelidfs?
felipec: oh, weird
mikkoc: oh btw, I have composite and aiglx off
rhodan: do i need special branches for kms?
rhodan: a special kernel?
MrCooper: mikkoc: grep Download /var/log/Xorg.0.log
mikkoc: (==) RADEON(0): Using accelerated EXA DownloadFromScreen hook
mikkoc: (II) DownloadFromScreen
rhodan: why did you disable composite and aiglx?
mikkoc: it's not AccelDFS?
mikkoc: because i hate those effects?
rhodan: but you can disable them in systemsettings
rhodan: why did you disable them in xorg.conf?
rhodan: does that affect performance?
mikkoc: well, i figured, the less crap loaded, the less resources taken
MrCooper: 'accelerated EXA DownloadFromScreen hook' == AccelDFS
rhodan: do you others agree to this?
mikkoc: MrCooper: ok
MrCooper: it works fine on most PCIe setups and even some AGP setups
rhodan: but not on mine...
rhodan: 16: 1173119 8 IO-APIC-fasteoi uhci_hcd:usb1, yenta, radeon@pci:0000:01:00.0
felipec: rhodan: do you also have this flickering on the right side of the screen? (very subtle)
mikkoc: have you tried disabling composite?
rhodan: i will try, brb
felipec: mikkoc: can I see your config?
MrCooper: if disabling AccelDFS works around the problem, disabling RENDER acceleration isn't directly related
mikkoc: MrCooper: do i need this? :D
mikkoc: SubSection "extmod"
mikkoc: Option "omit xfree86-dga"
MrCooper: just Load "extmod"
mikkoc: even if "(II) "extmod" will be loaded. This was enabled by default and also specified in the config file."
MrCooper: just drop it then, whatever
mikkoc: ok :) thx
MrCooper: it has no bearing on the problem
felipec: mikkoc: DRI and EXA are enabled by default I think
MrCooper: not EXA, at least not upstream
mikkoc: on radeonhd DRI isn't
mikkoc: at least last time i tried
MrCooper: this isn't a radeonhd channel though
felipec: mikkoc: no, it's not, but radeonhd sucks
mikkoc: i know... im just answering him
mikkoc: that's why im using radeon felipec ....
rhodan: disabling auglx and composite crashed kde4
mikkoc: disable the desktop effects before?
mikkoc: try a new profile
mikkoc: i've been using kde4 without composite for more than a year now
rhodan: i disabled acceldfs now
rhodan: probably forever
rhodan: kde4 is so damn slow without it...
felipec: disabling compositing and aiglx doesn't help here
rhodan: i now hate my card...
rhodan: xaa is even slower
mikkoc: yea xaa sucks with kde4
felipec: xaa works fine in GNOME
felipec: rhodan: does it gets rid of the glitches?
rhodan: felipec: no
rhodan: i won't switch to gnome because of a bug in xf86-video-ati...
felipec: rhodan: what bug?
rhodan: felipec: because of the pixmap bug...
felipec: rhodan: small pixmap bug?
rhodan: felipec: yes, that's the bug i'm talking about all the time....
felipec: yes, me too
felipec: you don't get this issue in kde4?
felipec: that's wird
rhodan: felipec: i _do_ get this issue in kde4
felipec: ah, I re-read what you wrote, I understand you now
felipec: I thought you where saying that a bug was preventing you from switching to GNOME
felipec: is there a bug report for this small pixmap issue?
MrCooper: I think there are several reports about AccelDFS breakage
felipec: MrCooper: can I disable it?
MrCooper: sure, per the radeon manpage: Option "AccelDFS" "off"
felipec: MrCooper: awesome! it works
felipec: this channel is great :)
felipec: uh, wait, now I'm getting more annoying issues
felipec: minimizing windows leaves some artifacts for a couple of seconds
reinoud: it is a known bug that displaying trough DVI on ATI IGP 9100 goes wrong? the BIOS and say `console' works fine but when i start X the signal disapears and the monitor switches to VGA
reinoud: (--) Chipset ATI Radeon 9100 IGP (A5) 5834 found
scsiraider: reinoud: does your display work in gdm/kdm
scsiraider: or does it just disappear when get into gnome/kde
reinoud: if i just start X the (console) screen disapears and the monitor switches to Analog input
reinoud: when i query the running X server all seems to indicate that for the X server all is OK internally
porcodildo: is tv-out support in the main package?
reinoud: i.e. correct resolution, monitor etc only i dont see output and my monitor detects no output
reinoud: or at least switches directly to analog
adamk_: Is X
adamk_: Is X trying to run at a resolution your monitor doesn't support, perhaps?
reinoud: my monitor accepts 1600x1200 and it does that fine in analog
rhodan: hi again
reinoud: under windows it also works at 1600x1200 only then digital too
rhodan: i thought about the acceldfs-qt4-pixmap problem. is it possible that it will be fixed automatically with dri2?
adamk_: But are you sure that X is trying to run at 1600x1200 when the digital is hooked up?
reinoud: screen #0:
reinoud: dimensions: 1600x1200 pixels (408x306 millimeters)
reinoud: resolution: 100x100 dots per inch
reinoud: depths (7): 24, 1, 4, 8, 15, 16, 32
reinoud: root window id: 0x6e
adamk_: Well that would definitely confirm it.
reinoud: Screen 0: minimum 320 x 200, current 1600 x 1200, maximum 1600 x 1200
reinoud: VGA-0 disconnected (normal left inverted right x axis y axis)
reinoud: DVI-0 connected 1600x1200+0+0 (normal left inverted right x axis y axis) 408mm x 306mm
reinoud: 1600x1200 60.0*+ 59.9
reinoud: S-video disconnected (normal left inverted right x axis y axis)
adamk_: I think one of the developers will need to help you out here.
adamk_: Be patient and hopefully one of them will chime in shortly.
reinoud: i hope so :)
rnoland_: hrm, found the bug in server 220.127.116.11
adamk_: reinoud: Please use a site like http://pastebin.ca/ so that you can show them your /var/log/Xorg.0.log file.
reinoud: i'll paste it there
reinoud: X -verbose 1000 -logverbose 1000 enough?
adamk_: reinoud: I think so.
agd5f: felipec: try Option "DisplayProirity" "HIGH" if you have an rv515/rv550 based card. they may help the flickering
reinoud: adamk_: who should i contact/look out for?
agd5f: reinoud: so far we haven't been able to get external DVI working properly on IGP chips
rnoland_: huh, i hadn't heard that...
rnoland_: noone has reported that to me...
reinoud: rnoland_: well i am one :) see http://pastebin.com/m2d5da8f0
agd5f: rnoland_: it's hit or miss
rnoland_: reinoud: i'm freebsd... slightly diff case.
reinoud: i'm using NetBSD
rnoland_: bjs is your man...
rnoland_: though if agd5f says it's not reliable, i doubt that bjs could do anything for you...
reinoud: well bjs (Blair) is doing the NetBSD ATI dri for me
reinoud: but he has not much time lately
rnoland_: yeah, i haven't talked to him for a few weeks anyway....
reinoud: err the ati drm
rnoland_: yep, i'm the "freebsd drm guy"
reinoud: i am the "netbsd UDF guy" ;)
agd5f: rnoland_: external dri on pre-atom chips is problematic in general. the i2c transations never seem to work on most setups. probably some missing gpio magic
rnoland_: agd5f: hrm...
rnoland_: agd5f: i really need hardware to play with...
reinoud: i already thought it might be due to some missing i2c or gpio; but then why does it revert back correctly again?
agd5f: if the bios init's the external dvi chip, then it usually works ok
reinoud: does the bios reinit the dvi chip then on X exit?
reinoud: agd5f: have you peeked the xorg log and xorg config i pasted?
agd5f: reinoud: no, X saves/restores the registers from the console state
reinoud: so register saving/restoring works fine... but why is the signal not there then? the dvi signal is not present aparently
agd5f: reinoud: yeah I saw it. in this case the log isn't really useful. we need to sort out the magic settings needed to make dvi work
agd5f: reinoud: if I knew, it would be fixed :)
agd5f: reinoud: if you can dump the regs I can take a look. try running vesa at 1600x1200 and then use radeontool to dump the regs
reinoud: hmmm a PCI/AGP snooper ... or are the new docs not specifying it?
agd5f: reinoud: it's available here: http://cgit.freedesktop.org/~airlied/radeontool/
agd5f: as root: ./radeontool regmatch '*'
agd5f: then run X with radeon and do the same thing
agd5f: ./radeontool regmatch '*' > regs.dump
agd5f: reinoud: the problem is external dvi setup varies by oem
reinoud: is it OS independent?
kkuno: agd5f, is tv-out in the main package?
reinoud: or does it connect to the X server?
kkuno: or do I have to use a different branch
agd5f: reinoud: it's oem specific for external dvi
agd5f: kkuno: it's supported in the main package, but only for r1xx-r3xx chips
agd5f: reinoud: each oem can choose an external dvi transmitter and the encoder setup is specific to that transmitter, etc.
agd5f: reinoud: if you dump the regs I can compare them, and probably get it working for you
reinoud: building git now to check out the sources
reinoud: blerk... git needs tcp/tk
agd5f: reinoud: gitk does, but git-core shouldn't
agd5f: although I suppose it depends how your vendor packages it
reinoud: there is also a -core i see
kkuno: agd5f, so if I compile the drivers from git (http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/) will I get tv-out support for r500?
agd5f: kkuno: no. tv-out only works on r1xx-r3xx at the moment
kkuno: oh ok
kkuno: I founded this posts on phoronix:
agd5f: kkuno: yeah, we ended up disabling it since it never really worked that well. it needs more investigation
reinoud: when i do `radeontool regmatch '*' > dump' when running with `vesa' it promptly switches off dvi
reinoud: WTF! after running with `vesa' and restarting X with `radeon' it works!
rnoland_: vesa set state that wasn't cleared most likely...
rnoland_: reboot to get the register diffs.
reinoud: now lets see if i can get a dump of the `non working' radeon
reinoud: agd5f: http://pastebin.com/m1daa768a
reinoud: it doesn't allways seem to do this trick; as in it didn't work the 2nd time
reinoud: radeon-not-working v.s. vesa: http://pastebin.com/m2ec71faa
reinoud: radeon not working v.s. radeon working: http://pastebin.com/m1daa768a
agd5f: reinoud: cool
agd5f: reinoud: unforunately, there don't seem to be any relevant changes. I think the bios probably does some proper init of the external dvi chip
agd5f: reinoud: can you send me a copy of your vbois?
agd5f: cd /sys/bus/pci/devices/
agd5f: as root
reinoud: hmmm.... how to do that on NetBSD
reinoud: got a 1Mb file with the vbios i think
reinoud: file vbios
reinoud: vbios: BIOS (ia32) ROM Ext. IBM comp. Video (112*512)
reinoud: looks OK i'd say
reinoud: only is it 512 Kb? and not 1Mb ?
mattst88: reinoud, yeah, that's probably OK
reinoud: agd5f: email address?
agd5f: reinoud: it's usually ~64k
agd5f: alexdeucher at gmail dot com
mattst88: hmm, 512Kb = 64KB
reinoud: 512 kilobyte sounds more logical to me since its an IGP and thus has no seperate rom
reinoud: prepare for a mailbomb!
reinoud: and transmitted to gmail
agd5f: reinoud: got it. thanks!
reinoud: hope you can find some things in it....
reinoud: keyboard problenns
dronn: i'm using archlinux with radeon oss driver (xorg 7.4) and have smaller glinches recently (since upgrading to xorg 7.4) i've got dri, but blender is very slow, but only in solid draw mode in shaded mode it operates normal.
dronn: I'm here right to ask such a question??
adamk: dronn, This is the right channel, yes.
adamk: dronn, You'll have to wait patiently for someone who has either had this problem, or can shed some light into what might be happening, though.
dronn: good, my card is a x800 gto. ok. :-)
adamk: dronn, Did blender get updated, too, or just Xorg?
dronn: just xorg
adamk: And you're sure your drivers are installed properly?
dronn: wings3d is working fine, crystalspace ok
dronn: i'm quiet sure, but perhaps i've overlooked some new options??
adamk: If you run blender from a terminal, do you get any errors or warnings from the driver?
dronn: no, ther used to be a r300 message, but its gone since the update (i think)
dronn: in wings there is a message:
dronn: File r300_render.c function r300Fallback line 371
dronn: Software fallback:ctx->RenderMode != GL_RENDER
adamk: You may want to toggle the driconf option concerning software fallbacks, just in case.
dronn: sorry to ask, but how and where?
adamk: Does archlinux have a 'driconf' package? If so, install it and then start it up. The disable software fallbacks option should be easy to see once you have it running.
adamk: I can't guarantee that it will fix this problem, but it's worth a shot.
dronn: i've changed some driconf settings concering softwarefallback and now get some error messages.. performace is still the same
dronn: IRQ's not enabled, falling back to busy waits: 0 18
dronn: File r300_render.c function r300Fallback line 387
dronn: Software fallback:ctx->Line.SmoothFlag
dronn: File r300_render.c function r300Fallback line 386
dronn: Software fallback:ctx->Line.StippleFlag
dronn: I'll try a git-version of the radeon driver
dronn: thanks so far
dronn: hi, back again, still the same :-(
Arunas`: hello i need help about 2900XT c,before 4 month i played all games , like cod4, but now even i cant play cod4
mentor: Arunas`: and that would be why...?
felipec: agd5f: the "DisplayProirity" option didn't help for the flickering
moses: hey, so is there a better radeon driver for windows?
moses: or do i have to use the amd posted ones?
moses: they are horrid
adamk: moses, This channel is for discussion of the open source radeon drivers for Xorg, which has nothing to do with windows :-)
moses: ok can i ask about hardware
moses: did they make a 1600 for pcie?
adamk: You can. You might not get an answer, but you can :-)
adamk: Yes, the x1600 is PCIe.
moses: why is there no pcie driver?
Circuitsoft: Debian lenny, with onboard HD3200
Circuitsoft: "(EE) fglrx(0): vm86() syscall generated signal 11."
Circuitsoft: I've been seeing that on both x86 and x64 with multiple distros.
mjg59: But fglrx is probably a topic for #ati
Circuitsoft: Is that really a problem?
Circuitsoft: radeon and radeonhd I think both also gave me that message.
Circuitsoft: Neither of them actually got me any video output.
Circuitsoft: If I could use radeonhd in preference to fglrx, I would.
Circuitsoft: Unfortunately, I don't get any output at all from it.
Circuitsoft: Motherboard is MSI Media Live DIVA
airlied: spstarr: you'll be glad to know agp r350 on my 865 motherboard looks screwed up.
spstarr: same corruption?
airlied: pretty close
airlied: ah well something to play with later then.
spstarr: so then if you get that fixed, there's a high chance it will fix me
airlied: well its a good place to start :)
adamk: Are you guys talking about KMS + Xorg corruption on AGP cards?
airlied: adamk: on certaqin chips.
airlied: I'm seeing minor corruption on my rv280 on AGP now that I'm sure is somehting we introduced receently.
airlied: I shold try the rv280 in the 865 as well.
adamk: Yeah, my corruption happens wiuth an X1300 on an 875.
soreau: So what is KMS supposed to accomplish? It seems to be more problematic than anything so far
soreau: Does it have to do with dri2 ?
adamk: And an R350 on an 865.
luke-jr: my radeon locked up on me a minute ago
luke-jr: anyone care?
airlied: soreau: funnily re-writing the graphics drivers can cause some instability
airlied: esp with radeons and AGP which was always hit and miss.
luke-jr: Ran out of GART memory (for 1048576)!
luke-jr: mean anything?
soreau: airlied: Not to be rude, but I'm not sure if that answers my question :|
airlied: soreau: well you said kms was problematic, I was explaning why.
airlied: kms has nothing to do with dri2.
luke-jr: airlied: radeon+AGP is hit and miss? since when? :/
soreau: airlied: Ok, so why rewrite something funnily when it is already working ok without the funniness?
luke-jr: PCI Express works better, from a stability standpoint?
luke-jr: soreau: it's not
airlied: luke-jr: yes its properly designed.l
luke-jr: soreau: the current architecture does not support many advanced things
airlied: soreau: because we want features we can't get with the current model;
airlied: like flickerfree booting, oops when X is running
luke-jr: airlied: so my crashes and lockups *could* simply be AGP? :/
airlied: better suspend/resume suppor
adamk: soreau, https://fedoraproject.org/wiki/Features/KernelModesetting
airlied: luke-jr: they can be, sometimes its worth trying to decreas agp bus sped
soreau: adamk: Ok, I'll have a read.
adamk: That lists some of the advantages of KMS.
luke-jr: airlied: oh, I remember that. Yeah, I'm using AGP 4x instead of 8x cuz 8x doesn't work
luke-jr: Ran out of GART memory (for 1048576)! <-- is that particular one addressable?
spstarr: airlied: given your chipset is different, you think there's some bugs in the mm?
airlied: spstarr: possibly.
johndoe: im using radeon hd 2600xt graphics card. should i use fglrx or radeonhd as the driver?
moeSizlak: as im sure you recall, my rv350 agp is unusable w/ composite effects
moeSizlak: glitches and lockups GALORE
airlied: is just upgrading the dodgy AGP box to 155
moeSizlak: 155 of what?
luke-jr: johndoe: never fglrx
johndoe: luke-jr: can you elaborate
moeSizlak: im still using kernel.org kernels
luke-jr: johndoe: fglrx is illegal and unsupported
moeSizlak: but hey, i can at least rule out that kms might be causing my problems
johndoe: the bigger issue to me is which one will give better support
moeSizlak: typically, the free one, as the nonfree one rarely works
luke-jr: johndoe: fglrx will deny you any support
adamk: Of course, fglrx is the only one that will give him 3D acceleration and Xv support.
luke-jr: adamk: orly?
moeSizlak: ummm, u get Xv support w/ radeon.o
moeSizlak: and 3d accell
adamk: No on a hd2600xt.
adamk: Not, rather.
moeSizlak: meh, we all have problems im sure, or we wouldnt be in here
johndoe: adamk: so fglrx it is?
luke-jr: johndoe: no
luke-jr: never fglrx
adamk: johndoe, If you want 3D and 2D acceleration. radeonhd and radeon will only provide shadowfb support.
johndoe: adamk: so whats the downside of fglrx
adamk: johndoe, So, yes, I would suggest fglrx in that case.
luke-jr: johndoe: it's illegal, unsupported, buggy, etc
johndoe: luke-jr: why would i care if its illegal
luke-jr: johndoe: because the law exists to be obeyed
adamk: luke-jr, Unless you can cite some legal ruling, please stop calling it illegal.
johndoe: luke-jr: no such laws in my country
luke-jr: johndoe: your country has no copyright law?
johndoe: luke-jr: it doesnt have stupid software laws
johndoe: like yours does
luke-jr: no copyright?
felipec: luke-jr: it's not illegal
luke-jr: so it's perfectly legal in your country to copy a Windows disc and install a single copy on 13833 PCs?
johndoe: i do it
luke-jr: sigh, "do it" doesn't make it legal
moeSizlak: i smoke 6 foot bongs daily
luke-jr: "nobody sued me yet" doesn't etiher
moeSizlak: doesnt make it leagl
adamk: Can we please keep this on-topic. The legality of fglrx is clearly not a subject for #radeon
johndoe: people outside the US dont care so much about software patent laws
felipec: luke-jr: what makes you think fglrx is illegal?
johndoe: not my fault you have a stupid government
johndoe: i think there was a court case "European Union vs. Microsoft"
johndoe: guess who won
adamk: Guys, seriously, enough.
luke-jr: johndoe: we're talking COPYRIGHT, not PATENT
adamk: Just drop it.
luke-jr: felipec: the licensing terms of Linux
johndoe: so its against the law to reverse engineer the driver?
luke-jr: johndoe: it's against the law to distribute a derived work of Linux under any terms other than the GPL
johndoe: just answer my last question
luke-jr: I'm not an IP lawyer, but afaik reverse engineering is generally legal unless a license forbids it.
airlied: guys off topic
soreau: fglrx discussions should not take place here. Please take this elsewhere
felipec: luke-jr: johndoe: #radeon-offtopic?
luke-jr: felipec: if you need details, just ask an IP lawyer. I am done discussing it.
felipec: luke-jr: ask an IP lawyer, yeah, very realistic. Some lawyers disagree that proprietary linux kernel modules are derivative works. And lawyers don't decide if something is illegal or not, that's decided in court. And please don't reply in this channel.
johndoe: "Reverse engineering software or hardware systems which is done for the purposes of interoperability (for example, to support undocumented file formats or undocumented hardware peripherals), is mostly believed to be legal, though patent owners often contest this and attempt to stifle any reverse engineering of their products for any reason."
johndoe: had no luck with fglrx, now onto radeonhd
luke-jr: johndoe: chances are fglrx screwed up your system, so nothign will work now
luke-jr: or at least that's how nvidia worked on … I think it was Mandrake I was using then
johndoe: i bet the IP lawyers screwed your brain in a imaginary lawsuit
felipec: johndoe: I think it is legal unless you have accepted a license agreement which says you can't do that
johndoe: i cant wait to see the day when windows dies.. if that ever happens
moses: question: what radeon settings make video quality better
moses: can anyone please help me with this!
luke-jr: why would the video card have anything to do with that?
moses: my video quality is horrid
moses: and i think i have my 3d settings
moses: screwed uop
moses: or something
moses: the quality really is terrible
luke-jr: video isn't 3D
moses: well then do i just play with the video settings?
moses: there are no video settings on this driver
moses: they are very weak
moses: where are the video settings?
moses: under avivo video?
luke-jr: moses: I don't think avivo works on Linux
moses: what do you use to play with video settings then?
moses: this card is a pile of shit
luke-jr: I don't.
luke-jr: everything works out of the box
luke-jr: well, everything I care about anyhow
luke-jr: software decoding works fine
moses: my video quality is low
moses: its fuzzz
moses: like all fuzzz
luke-jr: so download a better file
moses: no thats not it
moses: i have hi-deff videos
luke-jr: "hi-deff" means nothing
moses: im palying good quality now
moses: it looks pretty good
moses: but not as good as my last cared
moses: nono something is wrong with the video
moses: i mean the card
moses: the card is having rendering problems
joecool: luke-jr: don't feed him please
joecool: figures, that was fast
joecool: luke-jr: i don't believe he was even on linux, he came in here asking for better windows drivers before
bridgman: moses; are you still running windows xp ?
airlied: wow fail, I had my dvi cable half plugged into the monitor.
airlied: so DDC worked, but ntohing else
luke-jr: joecool: you know, I think you're right about him being a Windoze guy
luke-jr: joecool: he's a regular in #anime though, so I assume not *trying* to troll..
joecool: luke-jr: he's using a web client, i couldn't get accurate version information
joecool: but my sense tells me i'm 99% right
joecool: especially the avivo comment
luke-jr: well, I beat him at Yandere earlier; maybe he decided to try to piss me off by trolling in here? :/
joecool: he was trolling another time
joecool: i mean, i'm a professional troll, i can spot them a mile away
joecool: i reserve that talent for special targets though :)
bridgman: moses was definitely running windows xp yesterday
bridgman: or someone pretending to be moses was pretending to run xp, or whatever ;)
joecool: anyway, this channel is no fun to troll anyway, you need to aim for #debian or #wrongplanet for real fun.... i should tell him that if/when he comes back in
joecool: but back to serious topics d|-_-|b
joecool: i'm noticing weird artifacts with firefox now since i got onto this new xvideo vsynced driver
bridgman: there's an option to sync exa, do you have that turned on or off ?
joecool: but i can't blame the driver because i updated all of X to latest git
joecool: bridgman: i have no option set
joecool: i only see two attributes in xvinfo, the bicubic scaling and the xvideo vsync
bridgman: ok, so I *think* that means EXA doesn't get delayed, so first guess would be "unrelated to tearing changes"
bridgman: its an xorg.conf option, not related to xv
joecool: yeah i figured as much... lemme check my log and see if anything like that's enabled
bridgman: looks like the option is EXAVsync per Alex's blog
joecool: well it's weird... it's like links flicker when i go over them... i don't know what to blame
joecool: certain buttons flicker when they did not before
joecool: the only other non-X update was a minor update to openbox, so i'm not entirely sure that's not to blame either
bridgman: yeah, that is weird. I vaguely remember hearing about that a couple of months ago but don't remember what the problem turned out to be
bridgman: how old was the radeon driver before you updated, ie how much changed ?
joecool: i also recently got the 64-bit flash.... but i'm not going to blame that because i'm sure it didn't do this before
joecool: bridgman: a month at most..
bridgman: so basically the power supply is the same and everything else is different ;) ;)
luke-jr: proprietary stuff installed? I'll blame it ;)
luke-jr: Ran out of GART memory (for 1048576)! <-- my most recent problem
joecool: luke-jr: i'm not a huge fan of having it either... but it is used most everywhere and i was without it for months (this is a no-multilib system)
luke-jr: joecool: I've been Flash-free for years.
bridgman: ok, if you only updated the driver by a month my guess is that something else is the problem, but that is only a guess and not a high quality one at that
joecool: bridgman: i'll poke around, it could be something else in X doing it
luke-jr: my system is now 100% Free
luke-jr: with exception of Python and certain parts of X.org
bridgman: sounds good; if I can find the earlier report about link flicker I'll post back
joecool: luke-jr: sure you don't wanna include firmwares to the list
luke-jr: but those are ambiguities
luke-jr: joecool: unless they're part of Linux's tar.gz itself, I don't use any
luke-jr: not on my desktop at least
joecool: kinda stuck with closed source firmwares for most wifi on laptops
luke-jr: athk and madwifi seem safe now
luke-jr: though my work laptop was just replaced with an ipw ☹
luke-jr: which has a really crummy driver too, even though that bit is open
joecool: i replaced the ipw3945 chip in here with a broadcom one, simply because intel sucked so hard on this... the range was pitiful and the drivers even worse
luke-jr: the broadcom range/drivers were bad?
joecool: the ipw3945 was SHIT
luke-jr: yeah, isn't that odd?
luke-jr: the vendor-open stuff works like crap
luke-jr: yet the reverse engineered is great
joecool: usually i expect broadcom to be bad
luke-jr: yet another reason vendors aren't going to support Linux themselves I guess
joecool: but before i used the broadcom i used the super hackish acx100 driver for TI's chip
luke-jr: adm6996 switch is pretty crappy
luke-jr: but that's the switch itself, not the driver
joecool: luke-jr: i even jumped off linux for a bit to try openbsd's wpi driver for ipw3945 and it was halfway decent until it segfaulted every hour or so
joecool: openbsd is really hardcore about reverseengineering wifi
joecool: though it seems linux has caught up with the new mac80211 stack, things are much better
joecool: before that, every driver was very different and very unpredictable
joecool: you never knew what you were gonna get... i loved to walk into a store, buy a card and it could be anything from orinoco, to prism 2, 2.5, 2.75. 3.0, to broadcom, to TI
joecool: and companies like D-Link had like 10 different chipsets in ONE model
luke-jr: too bad 802.11 isn't like USB
luke-jr: and define a fixed hardware "protocol"
joecool: now you have this draft-n shit... that's a whole new mixed bag
luke-jr: Portage 2.2 (Gentoo) added a nice ACCEPT_LICENSE feature
luke-jr: so I can filter by license
joecool: heh, i'm on paludis right now
luke-jr: that's how I am confident my system is Free
luke-jr: does Paludis have ACCEPT_LICENSE? :þ
airlied: wonders what channel he is in again
luke-jr: airlied: #Radeon
jcristau: airlied: #offtopic, obviously
joecool: i wouldn't be suprised if paludis did.... quite a bit
joecool: airlied: don't worry i got this :P
joecool: reading about malicious hardware.... scary stuff, embed backdoor's into IC's that sorta thing
joecool: really no defense against it, and nearly impossible to detect without complete teardown
joecool|sleep: on that note: gnight
spstarr: bridgeman: cancel the request for erratas, airlied has seen corruption with his intel agpart