soc: hi
soc: wasn't it written that nvidia is preparing his own response to amd's opensource efforts?
maligor: where?
soc: on phoronix ...
maligor: I don't remember statements like that
starkmjolk: it was a while back
starkmjolk: I remember, one sec
soc: looks like the resonse is this:
soc: http://blogs.zdnet.com/open-source/?p=2588
starkmjolk: rss was from 10 jan
starkmjolk: but it was more rumors than facts I think, and I wouldn't take nvidia for their word, I wait for some actual work
mha: I will stick to what works best. not because what license it's usder.
mha: under*
mha: just as I use Opera because it gives me more benefits than Firefox. :)
maligor: soc, yeh, a big Coca Cola sign
maligor: you can drink it but how it's made is a secret :P
mha: screw licenses. hail functionality :]
maligor: I don't want my PC to be turned into a Game Console
maligor: Thank you, but no
mha: what about functionality turns your computer into a console?
maligor: I mean game console in the worst possible way
maligor: not the ability to run games
mha: functionality can be what suits your needs best.
maligor: but the entire crappy We Control How You Use It
redeeman: freedom is obviously not a "functionality" to mha
mha: redeeman: the freedom of choice is.
maligor: Microsoft seems to be going the Console Way with Vista
maligor: Hopefully they'll go back with Windows 7 but I doubt it
mha: then again I don't consider Vista that heck much functional :)
mha: what I mean is; if opera works better for your needs, what does it then matter that firefox is open source?
mha: I will still run what cost/functionality ratio gives me a best value.
redeeman: well.. i don't really care what browser you use as long as it doesent crap on the w3c specifications
maligor: I like the functionality of apt keeping my iceweasel upto date
maligor: I use Epiphany webkit and konqueror also sometimes tho
redeeman: but mha consider this
redeeman: freedom is typically much valued
redeeman: only on "software" people don't give a rats ass
maligor: opera is a bad comparison anyway
mha: maligor: howso?
maligor: it's not a part of the supply stack
redeeman: for instance, we have homeless people in mostly every country
mha: redeeman: yes.
redeeman: some which die because they are unable to purchase food
maligor: it's like a computer end user
redeeman: they could merely do a bad crime
redeeman: and they'd get food in prison
redeeman: but they do not
maligor: opera as such doesn't restrict your use of your computer
redeeman: because freedom is MUCH VALUED
mha: redeeman: free will, yes.
mha: redeeman: but then again I must consider what open source gives me.
redeeman: open source gives you freedom
mha: redeeman: I have the ability to modify the program and do so legaly. but then in reality I won't. even if I will I'll have to maintain my own patch tree to be able to update security bugs etc etc. it's just a whole hirarchy of unlikely administrative work
maligor: no, you don't have to maintain your own fork
mha: redeeman: it's true that I CAN. however in reality i'll just use nvidias binary blob and as long as it works it'll be great. :)
redeeman: and when it doesent work your ignorance has doomed you
redeeman: not to mention lots of others
redeeman: take for instance people using IE
mha: maligor: course I have. if I do some off spring firefox patch which only *I* see the use for' I'll have to maintain security patchen in to my branch aswell. :)
maligor: mha, and when it stops working, you'll have a useless bit of plastic and silicon
redeeman: they are actively helping to destroy the web for everyone else
redeeman: or people using flash
redeeman: they are the virtual equivelant to those working against personal freedom 100 years ago
mha: maligor, redeeman: when/if. you take it as that nvidia would actually abandon supporting their products in their viable lifetime'
maligor: viable lifetime is usually longer than they support it
redeeman: they don't support their products, lol
maligor: the graphics card game has become really fast
redeeman: they throw out a tiny piece of crapdriver which mostly allows one to use SOME aspects of the hardware
maligor: 6 month cycle to a new model
redeeman: and then they spend the entire lifetime of a product on fixing huge bugs, if even that
maligor: I don't think GeForce 9 series was even that?
redeeman: maligor: geforce 9 isn't really a new series
mha: maligor: they yet support the first TNT2 models with their legacy drivers. sure they may be a bit less maintained and not compile on your latest kernels today. but then again I wouldn't consider using such card today.
Kano: it can be compiled, i added a patch for 2.6.26 in my script
mha: you see :]
maligor: mha, can you paste lsmod | grep nvidia here
maligor: the line it gives
Kano: did it myself, based on a patch for 9x, bascially 1 extra line removed for 71.x
maligor: just curious about the size
mha: maligor: aha? what would you want to show with it?
Kano: the size of the patch or the module?
maligor: the module running in the kernel
Kano: nvidia 6747812 26
mha: maligor: yes? that's great?
maligor: 7MB, jeez
maligor: my kernel is under 2
mha: how much allocatable RAM do you have?
Kano: well compare it to fglrx ;)
redeeman: lol how big is fglrx?
maligor: mha, oh, memory isn't the problem
maligor: fglrx 1939712 24
maligor: it's just that 7MB of code is friggen lot
maligor: binary at that
redeeman: yeah it is
maligor: even 2MB is a lot
redeeman: but stuff today is excessively big
maligor: I mean my kernel, which essentially handles core services is only 2MB
mha: maligor: so. what *is* the problem then? you hawe two choices. 1. decide that you want to utilize your hardware rendering capabilities of your card or 2. use the "nv" driver and drop to a lesser level of functionality in tradeoff to an open license.
maligor: mha, just complexity
redeeman: nv isn't really what i'd call "open" anyway
redeeman: there really should be laws demanding that hardware vendors give out documentation
maligor: yeah, intel probably has the most open stuff now
mha: redeeman: it's a fully (surely not well documented) available source.
Kano: well nvidia did really more steps to create a binary which could not be hacked so easyly
maligor: biggest issue I have with binary only divers is support for non-mainstream os's tho
redeeman: mha: by that argument, even nvidia can have full source available, just disassemble it
mha: maligor: of what I've heard the intel xorg module is not that well documented of chipset functionality either.
maligor: stuff like AROS, Syllable and Haiku
redeeman: intel has released docs though
mha: redeeman: that would be a difference of readable code and magical code though.
redeeman: mha: nv driver is exactly magic
Kano: objdump -d /usr/lib/xorg/modules/drivers/nvidia_drv.so|grep :$
Kano: compare that against fglrx
Kano: you will not see any function name which shows the use of it
maligor: Kano, actually, fglrx does
mha: redeeman: yes. but atleast it's readable. a dissassembly of the nv blob of a few MB would not be quite the readable thing.
Kano: or at least very few
Kano: maligor: i know
redeeman: mha: there exists pretty good decompilers giving actually readable C
maligor: it's why I was able to make a softfiregl anyway
redeeman: mha: a result not much unlike the nv driver
Kano: everything from nvidia is always _nvNUMBERX...
maligor: yeah, obsfucated symbols
mha: redeeman: then again people do not dissassembly the nv blob and fix a bug, and recompile it again. when security holes appears. which is the strongest argument for an available source.
redeeman: people don't generally modify the nv driver either
redeeman: because it's just not viable
mha: when the open source ATi drivers actually overcome the functionality of the nvidia blob I would definatly switch
mha: until that happens I don't see a point
mha: if ff would work better than opera I would use ff
Kano: you can not fix so many bugs in binary blobs. but when the checks are not so hidden you can for example remove some version checking which prevents the driver from loading
mha: Kano: it's simply not viable; no. :)
mha: what I am saying is that people degrade themselves in terms of functionality and/or performance just because they get the source code of something they will never ever in reality touch.
mha: to compare it to that i have then by making that choice consumed my "freedom" is actually quite inversed of what I'm actually doing.
maligor: mha, no Haiku R2 for you :P
maligor: probably not R1 either
redeeman: you are advocating, that a country is free, if it holds a vote, and changes to dictatorship
maligor: actually I don't know if there's a ati driver for recent cards in it either... probably not
redeeman: that will probably come though
mha: maligor: had to look Haiku up :]
maligor: it's one of the most promising "New" Os's
maligor: ofcourse it's just a BeOS copy :P
redeeman: they are also doing more than copying
mha: ole BeOS. perhaps they make it "this time"
maligor: yeh, they'll have to break ABI compability in R2 tho, unless they want to stick with GCC 2.95 forever :P
mha: But no. That's correct that I would not restrict my availability of freedom to choose from restricted programs to use. Also why I stopped using Debian and went over to Ubuntu (less license anal)
mha: hehe
maligor: what programs were lacking in debian?
mha: maligor: I dislike that I have to go against the creators strong intents just to get a non-branded "icewheesel" (+spelling). Using their semi-updated non-free. Comparing that to Ubuntu which dosen't differentiate with things like that
mha: and apt'ing in vmware or opera is easily done. and dosen't go against anyone's will
maligor: err?
maligor: Mozilla had a problem with it
maligor: I think they want people to only use Blessed Mozilla Releases under the Firefox brand
maligor: and debian wants to add their own security patches
redeeman: debian wanted to ship security updates before mozilla approved them
redeeman: so that YOUR box would be safe
redeeman: and mozilla said: "yeah well.. if you do that, rename the package"
maligor: quite frankly I consider that anal of Mozilla, not debian
mha: maligor: actually, debian aswell had a problem with it. they didn't want to use their trademark logotype. and firefox turned down using the name firefox without the logotype.
mha: that's afaik why they aswell decided to rename the branch
maligor: how does it affect you if you use opera anyway?
mha: it was an example. non-free in debian is "non-wanted"
mha: non-maintained if you'd like.
redeeman: just like non-freedom in the real world is non-wanted
mha: :]
redeeman: software is no different
maligor: 'non-maintained'.. err?
mha: I would like to use my vmware with a repository maintained tree which is not recented by half the maintainers.
mha: maligor: refering to the time I wanted to install ssh.com's sshd. They were well into 2.x branch and the non-free had an outdated 1.x version
maligor: and ubuntu ships ssh.com:s ssh 2.0?
maligor: isn't that commercial
redeeman: theres nothing wrong with commercial
mha: maligor: not sure. haven't checked. point being that they keep packages there which I cannot be sure wether they are actually maintained. the correct thing would be to terminate the package all in all.
mha: maligor: but offering something outdated to the user is quite the sin imho.
mha: this was a few years back. of course. but it still puts the traits in my mind.
maligor: redeeman, yes, well, it would rather stop them from distributing it in official repos
maligor: redeeman, unless you want to make a Pirate Linux Distribution
redeeman: maligor: you are thinking of the word "closed crap thing", not commercial
maligor: really? I mean I though ssh sells ssh 2.0 for Money
redeeman: yes but i also sell gpl stuff for money
redeeman: as a commercial thing
redeeman: that's not non-free
mha: there's nothing wrong with software of different economical financial gains. wether it's support contracts or licensing costs.
redeeman: free/open software is not anti commercial
maligor: redeeman, ah, now I see your point, heh
mha: exactly. look at redhat/suse. they sell support contracts :]
mha: of free software
mha: which is fine.
maligor: so does IBM and Sun
maligor: and Dell and HP now too probably
mha: vmware sells licenses for using their software. which is fine
mha: if you want to use something free however there's several options.
maligor: it's their software
redeeman: vmware is just a piece of crap
mha: in my work I would prefer keeping an up-to-date vmware since that's the working defacto standard. which gives me the motivation for the cost.
Kano: well i used it till i discovered virtualbox, thats bascially enough what i need
maligor: yeah, I use vbox too
mha: i wouldn't want my distribution to "work against me" by not giving me sufficient updates of a package because the maintainers lacked the interest in non-free applications.
redeeman: i use kvm
Kano: i just dislike that sun forced the developers to remove the debian repository
mha: redeeman: me too. actually on my server at home. very nifty and easy.
Kano: well i did never try kvm, but i guess it is much slower
maligor: mha, haven't really seen any commercial software that didn't work in debian
mha: 4 virtual machines running on one Q6600 :]
redeeman: slower?
maligor: except maybe XSI
redeeman: it's faster
Kano: redeeman: i highly doubt that
maligor: XSI was a major pain in the ass to get working (It was a trial or some such)
redeeman: you get very very near native speeds
maligor: Bloody mainwin
redeeman: Kano: why would it be slower?
Kano: redeeman: not when you run windows
Kano: because you need optimizied drivers
redeeman: i don't give a rats ass about winblows
mha: maligor: the difference is that the responsability of updating cannot be trusted in the distribution if I use debian. because the developers will not support the non-free tree, and it's fairly limited. is vmware in there for instance? is it updated?
mha: maligor: surely I can grab the tarball and make a package or just throw it in
mha: maligor: but then I'll have to subscribe to their mailing list to grab a new updated version and maintain this
maligor: mha, I dare say vmware would raise hell if debian distributed their software
mha: maligor: perhaps. ubuntu seems to do it without any problems though.
maligor: but when I used it, it worked fine
maligor: oh.. vmware player?
mha: yes. it works just fine. it's a linux kernel with gnu base. course it works fine :)
Kano: vmware does not update the needed kernel modules for new kernels, you always have to use 3rd party patches, vbox even runs with 2.6.26 directly
mha: maligor: no. the actual vmware workstation
maligor: vmware-package - utility for building VMware Debian packages
maligor: that's in debian
redeeman: yeah lol kano, and also, vmware's network modules often create big problems, especially in relation to nvidia
maligor: I guess it's also a question of wanting to spend a huge amount of disk space to mirror software that requires a license
maligor: debian mirrors are immense
Kano: did not notice that, but i use vbox for some time. it has lots of new features, as you can add even sata drives now
Kano: which run with intel ich8-M driver in ahci mode
redeeman: kvm can add virtio drives...
maligor: Kano, does the open vbox have usb passthrough yet?
Kano: the main problem with vbox is that you can not add more or less cdrom drives than your host has
mha: point being that this is not something which is worked against in ubuntu. it's maintained and developers keeps interest in it. you cannot say the same about debian'
maligor: haven't needed it for a while
Kano: i use the prepackaged one
redeeman: i regularly see people cursing vmware when it crashes their box at work on a friday night, and they cant get there fixing it untill monday :P
mha: their developers are a tad.. um.. different in the licensing aspect :)
maligor: debian's developers have their own motives
maligor: if vmware provided a free license for all debian users, I'm sure they wouldn't mind putting it in non-free repo
maligor: else, I really don't see any point in distributing it in debian repos
mha: maligor: and that mindset is just fine. just not something I would feel comfortable using.
mha: maligor: to give people the ease of use, of course?
mha: maligor: you don't have to promote it to distribute and maintain it
maligor: comparing ubuntu to debian has it's problems too
mha: but if a user selectivly wants to use it, why not make it as easy as possible?
mha: course.
maligor: they use lots of debian packages
mha: yes, they do.
mha: why not use what's working good and improve it even better? isn't that what the openness of debian is all about? :)
maligor: anyone wanting to use a virtual machine, won't have any issues with using an automated package generator
mha: except for that the responsability of updating this package would be in your hands
maligor: mm.. didn't vmware want a email clickthrough form for downloading their vm's?
mha: yes they do.
mha: anyhow
mha: opera is another example
mha: opera.com distribute debian packages for download (non-repository-style, just plain files)
mha: you can use them under debian fine
mha: but during any security issue debian will not give a crap about that. :)
mha: it's your software. go update it.
mha: which is perfectly fine in the debian mindset.
maligor: opera could have a debian repo
maligor: but they don't
mha: they don't, correct. and I agree. but luckily my distribution dosen't work against opera, so any security patch will be apt'ed in as a comfy workaround
mha: besides. I wouldn't want to add 10 repositories to my sources.list.d/* when I can have one central archive which is supported and maintained by my distribution's maintainers.
mha: it's supporting, working and easy
mha: and they don't promote opera. they just offer it.
maligor: I don't see opera in ubuntu repos
mha: the user have to choose it.
mha: maligor: http://archive.canonical.com/ubuntu
mha: http://archive.canonical.com/ubuntu/pool/partner/o/opera/opera_9.27-20080331.6hardy1_i386.deb
mha: anyhow. bbl. foods!
maligor: you could probably use them in debian
mha: probably yes
maligor: I only see VMware server tho
mha: but that would be unsupported and going against their maintainers mindset/intention.
maligor: not workstation
maligor: and server is the one they hand out 'Free'
maligor: err.. 'free' I mean
maligor: I don't have any fundamental problems using commercial or closed source software tho
maligor: I'd prefer not to
mha: maligor: I have it in my dpkg --get-selections, perhaps it's from 7.10. perhaps they had to remove it for 8.04. which would be a shame for me as an end user.
mha: maligor: they are however not scared of maintaining a nice selection of closed source products.
mha: maligor: under debian you are the villan if you want to use them to begin with :)
mha: anyhow. food for real
maligor: no, debian doesn't villianize you
maligor: you have to delve pretty deep into the fanatics lair to get to that
maligor: they won't support you for it on debian support sites tho
starkmjolk: because they don't think it could ever be a good thing, using closed software
Kano: maligor: only the server is available when you register for free - the workstation product still consts money
maligor: I know
mha: starkmjolk: correct :)
mha: starkmjolk: which was my argument for it being a bad approach.
starkmjolk: I wouldn't say bad, some people like it and that's fine. But it isn't a mainstream approach, or any approach to go forward. :)
mha: :)
Kano: michaellarabel: when i search for a kernel on pts site, then i only find the direct benchmarks, not the ones which have been done to compare with another benchmark
Kano: how to seek for those?
redeeman: i don't think you can
Kano: then it would be nice if it would work
redeeman: i wish povray was faster
redeeman: it's been running for more than 38 hours now on my quadcore and is only 50% done :|
maligor: you could make it faster by buying 3 more quad cores
redeeman: which boards exists which have quadsocket?
maligor: only opteron and xeon actually
redeeman: i can only find dualsocket xeon
maligor: you'd really have to get 4 new cpus
maligor: http://www.tyan.com/product_board_detail.aspx?pid=619
Kano: redeeman: skulltrail, what about that ;)
redeeman: i thought that was dualsocket?
redeeman: maligor: that's not xeon
Kano: yes but you can use sli on it, the only intel board which support for sli
maligor: yes, but it's the first I found
redeeman: no nvidia chipset
Kano: or get sli and wait for a cuda mod of povray
Kano: skulltrail has got one nvidia chip on it to use sli
Kano: at least some boards, not all
redeeman: well.. no for me then
maligor: interesting, tyan doesn't seem to have any boards with 4x 771 sockets
redeeman: i know
redeeman: i already checked :P
Kano: http://www.xtremesystems.org/forums/showthread.php?t=184718
Kano: 6 ghz only ;)
Kano: 8 cores
redeeman: the fsb isn't nearly fast enough to fully take advantage of that for most stuff
maligor: yeah, use HT 3.0
michaellarabel: uncle_fungus: I think git is working now
starkmjolk: if you need quad quad intel, check supermicro, sure they'll have a few to choose from
redeeman: damn supermicro's site sucks
starkmjolk: btw, good news for those who haven't read it yet. Symbian is going unified and open source! :)
redeeman: where did you read they are opening?
uncle_fungus: michaellarabel: ok. I'm going to re-clone my local copy as I think I may have trashed it somewhat trying to get it working before
starkmjolk: redeeman: it's in the press release, I suppose most sites reproduced it. The goal seem to be to get it all under eclipse public license within 2 years.
starkmjolk: http://www.theinquirer.net/gb/inquirer/news/2008/06/24/nokia-delivers-knockout-blow is one link :)
michaellarabel: uncle_fungus: Great. Though the only bug it looks like right now is its defaulting to trondheim-12 branch. The master branch is still there, but you need to switch to that it looks like
vadi2: 'Nokia has enlisted the support of other Symbian investors '
vadi2: That's really nice.
cxo: the site may suck, but the their boards are the best
vadi2: Hahaha
vadi2: While BBC is reporting "Finnish mobile phone giant Nokia is paying 264m euros ($410m; £209m) to buy out the other shareholders in handset software firm Symbian."
vadi2: 'Nokia said Sony Ericsson, Ericsson, Panasonic and Siemens had agreed to sell their stakes in Symbian. '. Looks like they said "Uh... no. We're out."
starkmjolk: vadi2: two sides of the same coin
starkmjolk: nokia is buying it all, and together they're starting symbian foundation, with nokia contributing S60 and symbian, sony ericsson UIQ, TI some chips, and so on. :)
vadi2: Right. And everybody else said "we're outta here, you're nuts"
vadi2: let's hope nokia's strategy goes through
starkmjolk: It sounds quite solid to me
maligor: what? opensourcing it?
starkmjolk: I think symbian has been kinda off since it changed name from epoc, but it seem to be back on track. Atleast it isn't windows CE :)
maligor: that's fantastic
maligor: beats me why nokia would want to pay all the other owners for it, tho :P
vadi2: because they said you're nuts and we're outta here
vadi2: unfortunately. theinquerer said otherwise, but I'd trust bbc more...
starkmjolk: makes sense enough if they had the money lying around I think :)
starkmjolk: vadi2: take any other news source, most are just rewrites of the press release but you get a decent picture of the whole idea
starkmjolk: what really happens in the end might be a different story, but that's for the future to tell
maligor: I guess they want to drive up symbian adoptation
vadi2: I did. but bbc and inquerer say two different things. "Nokia has enlisted the support of other Symbian investors" and "Nokia is paying 264m euros ($410m; £209m) to buy out the other shareholders in handset software firm Symbian."
maligor: Just doesn't really make all that much sense
vadi2: if it was the first, it would've meant other companies were convinced that this is a good way to go
maligor: what does it really matter if they use an open symbian platform for their phones or android
vadi2: however it's not, as bbc is sorta more reliable.
maligor: I wonder how suitable symbian would be for network appliances
maligor: if they have/add TI OMAP35xx support for it, symbian + beagleboard might be interesting
starkmjolk: the software firm symbian is going the way of the dodo, they're forming the symbian foundation :)
starkmjolk: maligor: interesting thought :) but symbian (and epoc) has always been about being mobile, we'll see if that ever changes :)
maligor: with symbian and android being Free, I wonder what it'll do for Windows CE adoptation
maligor: err.. Windows Embedded I mean
starkmjolk: just as I'll remember symbian as epoc I'll remember windows CE as windows CE :)
starkmjolk: bbl :)
maligor: oh and let's not forget linux + QT which nokia might also start using
maligor: and maemo
michaellarabel: uncle_fungus et al: Another thing in PTS 1.2 is support for monitoring cpu.usage as a percent
uncle_fungus: yeah, just noticed that :)
uncle_fungus: PTS 1.2 has now got a dependency on php-pcntl on some systems though
michaellarabel: Some distributions fork out the pcntl functions into a separate package?
uncle_fungus: yup
uncle_fungus: PCLinuxOS being one of them. Not sure about mandriva, I'll you know when I get home
michaellarabel: pctnl is what's needed to thread the functions.
uncle_fungus: yeah, if you don't have it you'll get errors about undefined functions when you try and use sensor monitoring
uncle_fungus: also, I can no longer see the master branch in the git repo. If I do `git branch` all I get is trondheim-12
michaellarabel: type: git-branch -r
michaellarabel: and then: git-checkout origin/master
michaellarabel: and I think you'll get back to master
uncle_fungus: ah, yes, that did it
michaellarabel: Not sure how trondheim-12 branch ended up becoming the default...
uncle_fungus: michaellarabel: ok, the threading appears to be working well with sensor monitoring: http://fire-salamander.co.uk/temp/pts/system-monitor/pts-monitor-viewer.html#1214312719-0.png,1214312719-1.png
uncle_fungus: that graph shows the unfortunate throttling on my laptop, but also highlights why we need a legend on the graphs as it doesn't say what its monitoring
vadi2: and some anti-aliasing!
michaellarabel: uncle_fungus: I'll add in the legend even when it's a single type, right now.
uncle_fungus: great, thanks
michaellarabel: vadi2: Anti-aliasing is enabled I believe, well, the bit provided through the GD library.
vadi2: doesn't look like so in that screenshot :/
michaellarabel: hmm I had the antialiasing call commented out for some reason...
vadi2: ahh :)
michaellarabel: hmm some reason with the Ubuntu package, imageantialias() is missing.
vadi2: whops.
vadi2: I couldn't find anything in the bugtracker
vadi2: maybe need to check the debian one
michaellarabel: uncle_fungus: I also added in support where if pcntl isn't found on the system, print out a warning message to the user but don't crash.
uncle_fungus: good
uncle_fungus: imageantialiase doesn't work for me either
vadi2: maybe it's in another package?
vadi2: Oh, here's what I found: http://groups.google.com/group/linux.debian.bugs.dist/browse_thread/thread/4137462f81461485/8a0cf5eb5be30bab?lnk=gst&q=imageantialias#8a0cf5eb5be30bab
vadi2: "Please switch to internal GD library". "Availability of additional functions like imagerotate and imageantialias
vadi2: (which are not included in the external GD library). "
michaellarabel: Ah
uncle_fungus: from php.net: "Note: This function is only available if PHP is compiled with the bundled version of the GD library."
vadi2: however that was in 2004
vadi2: so..
vadi2: it says that they did it I think. but it in php4
vadi2: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=260790#4
cxo: morning
starkmjolk: did you just sleep cxo? :)
starkmjolk: and morning :)
cxo: yes
cxo: i guess you guys are like 6hrs ahead
starkmjolk: I've said it before and I'll say it again, time differences rock :)
starkmjolk: I'm way ahead :P
starkmjolk: 15:36 here now
cxo: 6hrs
cxo: so does microsoft go bankrupt in the future starkmjolk ?
cxo: (especially since bill is leaving this week)
starkmjolk: cxo: depends on how long of a future, but nope
maligor: future is infinite
starkmjolk: they still have ballmer :)
starkmjolk: maligor: not mine :)
maligor: the manner of destruction is only in question
maligor: I expect the GNU Project will buy out Microsoft once the share price drops below a millionth of a cent
cxo: heh
starkmjolk: very old rms as ceo :P
cxo: ...."Linus Torvalds and his following buy Microsoft Corp for 23cents"
cxo: "...company product strategy reshaped and to focus on toiled paper and other micro soft products"
cxo: "... directX now open source, and fully compatible with Linux..."
cxo: thats all too good to be true,
cxo: so its never gonna happen
starkmjolk: you're just giving me false hope :(
redeeman: lol directx is crap
redeeman: we don't want it opensourced, we want it EOL'ed
starkmjolk: yep, but everyone is using it
vadi2: they shouldn't! and nor should you buy games that use it *hiss*
starkmjolk: in this senario, there will prolly be some nice dx10 games out there too. so even though the future belongs to opengl, it wouldn't hurt with some openness and porting :)
cxo: The Church of the Opener Sources
redeeman: those games should be forcibly pulled off the market untill they had been fixed
starkmjolk: Open Saucers :)
starkmjolk: it's like a UFO with an open roof :)
cxo: convertible
starkmjolk: indeed. :)
maligor: would the windshield be completely around it or just on 'front' side?
cxo: traveling at light speed can seriously mess with your hairdo
vadi2: Why is xorg taking up 100mb of ram? :/
starkmjolk: because it is a big round number?
maligor: vadi2, pixmap caches probably
vadi2: ah. those don't count as "cache"?
maligor: well.. not caches, pixmaps really
vadi2: oh
maligor: cache in kernel memory statistics refers to the kernels caches
maligor: probably
michaellarabel: Just added gpu.freq to git for monitoring GPU Frequency during testing
cxo: gpu frequency? where do you get that information from?
michaellarabel: cxo: PTS supports reading it through the binary ATI and NVIDIA drivers.
cxo: from libnvidia?
michaellarabel: On NVIDIA through the NV Control Extension with interfacing through nvidia-setting queries.
schestowitz: , have you seen this? http://news.cnet.com/8301-10784_3-9975720-7.html?part=rss&subj=news&tag=2547-1_3-0-20
michaellarabel: schestowitz: Yes
michaellarabel: One thing that unfortunately isn't supported by the NVIDIA driver to my knowledge with the extension is reading the GPU load.
uncle_fungus: michaellarabel: can you update-server-info please?
michaellarabel: uncle_fungus: Done
uncle_fungus: cheers
starkmjolk: time to head home, bbl
Ivanovic: michaellarabel: http://www.phoronix.com/forums/showthread.php?t=10991&page=2#post36528
Ivanovic: what have you written about banning and jade in the other thread as a warning?
Ivanovic: please, please, do so
Ivanovic: (at least the coloring is really annoying, the whole post in red, *ARGH*!)
cxo: maybe he likes red
cxo: dont be a colourist
cxo: nvidia settings itself doesnt show the gpu load, s
michaellarabel: cxo: Unfortunately
cxo: on windows you only get the clock speed
cxo: and temp
cxo: so same sort of thing on linux
michaellarabel: cxo: On RV670+ for Linux, we can access the GPU load as a percentage.
cxo: nice
cxo: with the open source driver?
michaellarabel: cxo: Not yet
Ivanovic: michaellarabel: how?
michaellarabel: how what?
Ivanovic: got a rv670 based card...
Ivanovic: "[17:08:57] cxo: On RV670+ for Linux, we can access the GPU load as a percentage."
michaellarabel: Ivanovic: For now, through pplib-cmd
Ivanovic: ah, okay
michaellarabel: aticonfig --pplib-cmd "get activity"
z3ro: michaellarabel: is there a similar command for older GPU's?
michaellarabel: z3ro: Not at this time.
athewk: please, don't ban JAde
athewk: the guy is hilarious
cxo: wtf is JAde
athewk: Jade, a poster on the phoronix forums
Exopaladin: argh, probably the worst thing about gentoo is when I realise I want a package in a hurry that happens to take ages to compile
Deanjo: athewk, Jade's banters get really annoying after a while. It's like listening to the village idiot. It starts to where a persons patience thin.
Deanjo: *wear
cxo: be strong
cxo: direction your passion towards writing the open source ati driver
cxo: s/direction/direct
cxo: instead of fighting with the village idiot
Deanjo: Na, I just put him on my forums ignore list. To bad it doesn't allow to ignore thread
athewk: I have no doubt that they do
athewk: I'm just fascinated by the depth of the man's paranoia
athewk: if paranoia is the right term to describe it
cxo: show me a thread... please
michaellarabel: http://www.phoronix.com/forums/showthread.php?t=7544
michaellarabel: http://www.phoronix.com/forums/showthread.php?t=10902
cxo: red on gray, difficult to read...
vadi2: Haha, you guys titled him "The Conspiracist"
redeeman: athewk: some of what he writes is true
vadi2: he must be one of the guys behind http://boycottnovell.com/ :)
redeeman: you believe novells deal was a good thing?
vadi2: I didn't claim a single thing :)
michaellarabel: Vadi2: Schestowitz is the boycott novell creator
vadi2: ah
Deanjo: Ya he's had to retract quite a few things as well
cxo: damn is this guy related to that woman or what...
cxo: how is this a general linux discussion?
cxo: its a venture into the private life of some linux dev...
Deanjo: rantings of a moron
maligor: a stalker
maligor: instead of beautiful women, he targets a linux kernel developer
Deanjo: Although he did actually post one semi-intelligent post recently
schestowitz: , I would have never called it "boycott anything". The site was created by Shane. Think of it as a Watcher over fight versus FOSS.
vadi2: hm. this java webstart thing is really nice.
maligor: just now discovered it?-)
vadi2: no but just remembered about it, this game uses it (http://www.tommytwisters.com/?page_id=6)
vadi2: even works with openjdk!
vadi2: which I have only because I can't figure out how to get the normal java to work..
maligor: running amd64?
vadi2: yeah
maligor: does the plugin work too?
vadi2: installed the 32bit java and everything.
vadi2: yeah it works.
vadi2: except for one game I play the fonts are horrifically small. but it's quite usable as it turns out
vadi2: runs the nord game too (http://www.nordgame.com/)
m-c: Could you recommend a good tutorial for using KVM ? Are there repositories for preconfigured KVM disks?
m-c: The documentation is kind of lacking for complete information on OS installations.
vadi2: yeah, sec
m-c: I know Phoronix did some benchmarks on it a while back, and I figure maybe some people got into it
vadi2: https://help.ubuntu.com/community/KVM
vadi2: I tried it, but for some reason the windows installation would always fail
vadi2: since I need virtualization just for that I went back to vb :/
redeeman: kvm works for winblows
vadi2: (but kvm regularly annoys me as I have to do "sudo rmmod kvm-intel" on every single bootup)
redeeman: so blacklist the module?
vadi2: thanks for the troubleshooting redeeman
vadi2: it, uh, helped.
redeeman: it works if you don't follow wierd uggabugga tutorials using libvirt and xml config files and stuff
redeeman: and wtf
redeeman: ubuntu-vm-builder
vadi2: I didn't, I just installed kvm and opened it
vadi2: didn't follow that tutorial.
vadi2: well
redeeman: well.. believe me, it works
vadi2: I did spend $300 on a CPU upgrade
vadi2: to get kvm working...
redeeman: what cpu? :P
m-c: Yeah, I have been using the ubuntu tutorial, and that worked.
m-c: Me too, vadi2
redeeman: i just use my Q9450
redeeman: one doesent even need to setup bridges or anything
redeeman: unless you want it.. bridged
vadi2: "T7500"
redeeman: user net works fine
m-c: I just got told to RTFM in the #kvm channel. I guess I forgot what it was like to be chatting with the red hat guys.
vadi2: oof
m-c: Fun times!
redeeman: well.. the documentation is quite okay
vadi2: basically all you do is "sudo apt-get install kvm libvirt-bin"
redeeman: vadi2: you don't need libvirt
redeeman: all you need is kvm
vadi2: then paste the commands to add yourself to the user
vadi2: and make a new machine in kvm... but it has 2 modes for it, I forget which.
m-c: The documentation is there, but it is hardly comprehensive. It's the kind that makes sense only as a reference.
redeeman: m-c: erhm, the qemu documentation is very comprehensive
redeeman: not to mention that the kvm/qemu stdout help is enough for what one needs 99% of time
vadi2: in short, vb is much easier to setup and use.
vadi2: atm at least.
redeeman: lol yeah right
redeeman: fiddling with weird modules
vadi2: *shrug* I did no fiddling.
redeeman: kvm is ALOT easier
vadi2: installed the .deb, and windows in it.
m-c: the fact that kqemu qemu virt-manager libvirt etc etc etc all have different documentation sources make it, by definition, not comprehensive
redeeman: as i said, you do not need libvirt of virt-manager or such stuff
redeeman: and as for kqemu, that's not even in use for kvm
redeeman: and since kvm is a modified qemu, it makes very much sense to use their documentation
redeeman: it's extremely simple, just install kvm, load the kernel module(your dist probably does this), have permissions (add to kvm group, dist possibly? does this)
m-c: So, where is this comprehensive documentation?
redeeman: ok
redeeman: http://bellard.org/qemu/qemu-doc.html
redeeman: then how to use kvm is:
redeeman: create vm: qemu-image create -f qcow2 Blablablaba.img 20G
redeeman: then to start:
vadi2: yeah.
redeeman: kvm -hda blablablabla.img -m 512 -cdrom gah.iso -boot d
redeeman: and that's it
vadi2: stop there. where will I find this code to paste?
m-c: No virt-install ?
redeeman: m-c: that is not required
redeeman: m-c: one ONLY needs kvm, and the kvm kernel module, all the rest is just weird clutter for "enterprise setups"
m-c: what's the difference between the two?
redeeman: difference between what?
redeeman: kvm is a modified qemu, to make use of the cpu's virtualization features
redeeman: thus qemu's documentation is used for kvm
vadi2: redeeman: I don't see the commands you said mentioned anywhere in that doc
vadi2: chances of me figuring out that combination of letters is close to nil
redeeman: vadi2: look at "invocation"
m-c: is it for "enterprise setups"? what's the difference between using qemu-image / kvm and using virt-install ?
redeeman: m-c: virtmanager and stuff, is a thing to supposedly help administrating virtual machines
vadi2: redeeman: that's not "simple"
redeeman: vadi2: it's quite easy, just invoke "kvm", and you see the options which you need
m-c: Oh.
redeeman: it's quite easy, -hda is to set the harddrive 1 image, -m 512 gives 512mb ram, -cdrom gah.iso makes gah.iso the cdrom, -boot d makes it boot from cdrom
vadi2: redeeman: sorry dude, but I can't agree. looks like a whole lot of options to be that I don't know how to use.
redeeman: virt-manager also have some gui's
vadi2: redeeman: what would be simple is a "create new virtual machine" button.
redeeman: vadi2: you only need to use options you need
redeeman: as simple as a qemu-create command?
vadi2: redeeman: you think I know what I need?
vadi2: I have a .iso, and I need to make a vm.
vadi2: at most, I should get a dialog asking for the .iso.
redeeman: lol
redeeman: then use the virt-manager gui crap
redeeman: it's much easier just going by kvm/qemu-create directly
vadi2: yeah... right.
vadi2: calculus is easy too when you know it
vadi2: but not when you're in grade 6
redeeman: you are one of those "OHOHHH MUST BE GUI TO BE EASY?!?!?!?!!?" type of persons?
vadi2: and you're one of those "it's easy lol, how can you possible not get it?" persons
redeeman: okay let's break this down
redeeman: you OBVIOUSLY know you wanna create a vm
m-c: Well, that helped me a lot to understand these are tools layered on each other. I thought they were parts to a big puzzle. Knowing there are only 1-3 programs now, yes, I can see how the documentation is comprehensive for just those.
redeeman: you obviously know you want to boot an iso
redeeman: you obviously know you want a disk image
vadi2: disk image?
redeeman: the virtual hard drive
vadi2: ? I want a virtual machine.
redeeman: if you just wanna BOOT an iso you don't need it
vadi2: the thing should take care of the virtual harddrive itself..
redeeman: a virtual machine doesent need to have a harddrive
redeeman: if you ONLY want to boot a livecd, for instance, you just go:
redeeman: kvm -cdrom gah.iso
redeeman: and it boots
vadi2: and where's my virtual machine that I can work in?
redeeman: i don't understand what you are asking?
vadi2: or is it "some assembly required?"
redeeman: no assembly required
vadi2: but I don't get a virtual machine with "kvm -cdrom gah.iso", do I?
m-c: he he - I just booted puppy linux live cd with that command
redeeman: yes, you get, it starts up a window, and runs the livecd
m-c: vadi2: yes you do
vadi2: and I can save and everything in it?
m-c: no, you can save nothing
redeeman: if you add a harddrive to it, yes
m-c: oh, neat
redeeman: vadi2: i think the problem is that you are thinking in much too complicated terms, virtualbox is actually a great deal more complex than kvm
m-c: that's what I would like to know, how do you add a drive to a running VM
redeeman: m-c: one can actually do that!
vadi2: right, so there's the assembly..
redeeman: m-c: you can also add network interfaces and such
vadi2: and the network interface..
m-c: that's okay, I have that set up
vadi2: I hope the mouse is automatic?
redeeman: vadi2: you mean the same kind of assembly as the wizard to create a vm in vbox?
vadi2: redeeman: at least that thing is easy
redeeman: as is kvm
vadi2: redeeman: figuring out this isn't
redeeman: you are impossible to deal with
redeeman: m-c understands it
redeeman: m-c: you can enter an administration console inside kvm/qemu
m-c: I've been looking at it for days, though. Also I deal with VMware at work. ;)
redeeman: m-c: where you can for instance add hardware, or migreate a running VM to another box
m-c: So what would be the command for adding disks. Just point me in the right direction.
m-c: I need to know how to "swap cds" while a VM is running
m-c: Oh, an administration console (reading above)
redeeman: yeah
redeeman: qemu monitor
redeeman: http://bellard.org/qemu/qemu-doc.html#SEC12
m-c: vadi2: Did you try to do that simple KVM command on a live cd yet?
redeeman: vadi2: this is ALL explained in the simple output from "kvm"
redeeman: just like the wizard explains
vadi2: I don't have any windows .iso's anymore, so no
m-c: Windows has live cds? No, I mean from an Ubuntu CD
vadi2: don't worry, I got the machine figured out in kvm gui before. it just refused to work. but figuring out this via the terminal would've been lol
m-c: Or download puppy linux
vadi2: don't need either virtualized..
vadi2: actually I got an opensuse .iso. sec
m-c: you do it for fun and learning
vadi2: er heh. I do things for fun and other things for learning
redeeman: vadi2: believe me, kvm is ALOT simpler, you are just getting confused because it's alot simpler than virtualbox
vadi2: of course. how could anyone not figure it out from typing 'kvm'
vadi2: it's dead simple.
redeeman: it's like.... EVERY other commandline application?
m-c: Actually, it's a lot better than a GUI that reveals nothing
vadi2: yes, with their [-23trtrsarfsdgtraesrtgrstdff] help file syntax
redeeman: i don't understand what you mean? this lists the most used options and tells what they do
redeeman: it seems a very useful thing
vadi2: yeah, it's very useful and simple.
redeeman: and if one wants it a bit prettier laid out, one reads the man page
redeeman: and if one wants to watch the man page a tad prettier, one uses it via konqueror like: man:kvm
vadi2: nah there's the yelp thing for that, but unfortunately it still prints [-dsffrwertdgtr]
redeeman: i don't understand? are you saying it doesent print the help digest?
redeeman: http://rafb.net/p/9QlAVO89.html
m-c: I was not even looking at the "kvm" program before just now.
vadi2: opensuse is stuck at the green screen forever now. doesn't seem to be using any cpu either
m-c: *head smack*
vadi2: (while the same iso did work, so it's not corrupted for sure)
m-c: Try puppy linux
vadi2: I don't need it
m-c: So what
m-c: It's tiny
m-c: and you can see a VM real quick
vadi2: I'd like to be able to see it work with opensuse, for a valid .iso
m-c: walk before you run
m-c: opensuse is a big honking distro that might need special parameters
m-c: puppy is small flexable ... and cute!
m-c: or get Damn Small
vadi2: if it needs any special parameters, then i'm out. 'cause i already got it working without any. anyways downloading the puppy thing
maligor: opensuse has a good looking... Installer!
m-c: ;-)
maligor: infact it's pretty fantastic looking in 11.0
maligor: I'm almost tempted to install it somewhere to see it running
vadi2: Yeah, I can agree. although there is still so much technical info everywhere for home installation :/
m-c: Yeah, one day KVM will have a nice shiney front end... not yet.
m-c: no point - it is still changing very fast
cxo: shiny front end?
maligor: cxo, you know... glossy buttons
cxo: to a kernel module?....
m-c: like virtualbox
vadi2: easy buttons stupid people can press. like "next" and "start"
cxo: kvm like the VT kernel module?
m-c: the last year has seen lots of new features - no point to put a lid on the user interface
vadi2: is puppy supposed to give a black screen for a long while after it asked to choose the x server?
m-c: not a long time - it came up for me quickly
cxo: i dont understand what you guys are talking about, how do you put a gui to a kernel module thats just supposed to sit there and translate binary ?
m-c: do you have the VT kernel modules installaed?
vadi2: well, if I didn't, was it supposed to get to that point?
m-c: cxo: yes, I get the joke - KVM is technically a module. It is also a linux command. ;)
m-c: vadi2: yes, I think the command will switch to software emulation
cxo: oh kvm is a command, i didnt know that
vadi2: don't know, but it's not working in software emulation.
m-c: when you chose the x server, did you choose Xvesa?
vadi2: no cpu usage, and it's still black.
vadi2: no, xorg
m-c: try again with vesa
m-c: Yeah, it says on it that it will take a long time to poll when you choose Xorg
vadi2: that worked
vadi2: now, opensuse?
m-c: you're on your own from here out! ;-)
vadi2: ok. since vb got it working fine, I'll stick with that.
m-c: *cough* proprietary operating system *cough*
m-c: You realize that Virtualization is putting a whole new fight for open source computing, right? All of a sudden, with the advances of Linux, especially on servers, VMware and VirtualBox comes in with a proprietary operating system to run Linux - and boxes up the Free Software into a little window on a non-free operating system.
m-c: Remember: He who owns the hardware drivers, controls the computer.
vadi2: ya what?
vadi2: for me it was the other way around :P
vadi2: VMware and VirtualBox comes in with a proprietary operating system to run Linux - and boxes up the Windows into a little window on a free operating system.
m-c: I am just saying that it will be important in the next couple years, because a popular virtualization solutions might become the new standard operating system, five years from now.
m-c: While Xen is open source, the company Xensource was just bought by Citrix which is strongly proprietary. So, that leaves KVM tied with Linux as the best bet for the future. And Red Hat has thrown itself behind it, so that's terrific, too.
m-c: redeeman: I really hate how I can hit CTRL-C and stop the VM. Is there a way around that?
m-c: Cutting and pasting becomes a lot harder when you crash computers with it. ;)
m-c: Well, thanks for that link and the kvm explaination. What a relief.
cxo: you are using qemu?
cxo: i had a similar discussion with the folks at qemu, about having something to properly shutdown the vm, but apparently its not needed, and kill is fine
cxo: They fear that if NVIDIA provides an open-source alternative, kernel developers may become even more hostile towards binary blobs.
cxo: heh
redeeman: that's just stupid
cxo: to fear kernel devs becoming hostile over a binary blobs that you dont supply anymore because you just moved to open source
cxo: logic these days....
michaellarabel: cxo: No, if they did a strategy like AMD.
redeeman: that's still a completely stupid fear
michaellarabel: redeeman: It's a fear that some have.
redeeman: yeah obviously
redeeman: and im saying that's it's an invalid fear
cxo: reminds me of the movie 300
cxo: 300 free people against paid employees, no match
cxo: lights up the torches
cxo: we're going to nvidia hq!
redeeman: http://www.phoronix.com/forums/showthread.php?p=36592&posted=1#post36592
redeeman: it's brilliant
Exopaladin: can we stop for pie on the way?
vadi2: er, isn't that what the open source nvidia driver is doing?
redeeman: my bet is that they don't give a rats ass about the hdcp stuff
cxo: "dedicated warfare"
cxo: i like that
redeeman: man it would be so sweet
redeeman: to go say to nvidia "complete docs for all but hdcp within 1 month, or we supply docs for hdcp!"
redeeman: and then say: "and just to show you we are serious, its not 14 days!"
vadi2: that'd be great, lemme know when that's happening
redeeman: lol now nvidia claims they invented the gpu
Deanjo: I guess that would all depend on what definition a GPU is.
Deanjo: Postscript r.i.p's could be a GPU.
Deanjo: http://www.webopedia.com/TERM/G/GPU.html
Deanjo: Before that they were called graphics accelerators
redeeman: i don't see how a gpu would need t&l
Deanjo: So then what is a GPU? A chip that outputs video? A chip that does primitives?
starkmjolk: it's a processing unit with the main purpose of processing graphics related data by my definition
Deanjo: By that definition, like I said before, could also mean a postscript rip
cxo: a gpu is basically a vector register
cxo: all it does is handle vectors
redeeman: http://icanhascheezburger.com/2007/01/24/we-cant-stop-here-this-is-bat-country/
Deanjo: How can you limit a definition of a GPU to just one function cxo?
vadi2: oo. nice. ubuntu mid is released
vadi2: now, just for hardware that runs it to come out..
starkmjolk: vadi2: I'm seriously concidering getting a nokia e71 and let this all mature a bit right now
starkmjolk: vadi2: both nvidia and TI have great platforms on the way it seem
Deanjo: As far as I know Nvidia was the first to market a product using the term GPU
vadi2: starkmjolk: i'm gonna wait for an umpc to come out
starkmjolk: vadi2: you heard of any early ones except for m528 and the other intel power hogs?
cxo: for example, if you had a matrix of vectors that defined a hot chick in a game, to pass that into a regular register, would mean, each dimension would have to sit in a register of its own, and be access separately
vadi2: starkmjolk: not yet, but I suspect a lot just didn't make the move yet from the hints dropped
vadi2: starkmjolk: and they will be based on the intel atom anyway, no?
cxo: you would need 3 registers for 1 vector, and multiple adds/multiples just to do a vector multiple, since you are accessing more than 1 register, and have to accumulate your result
cxo: while a vector register, can be loaded with the entire vector, and needs only 1 instruction
Deanjo: OK so then by that definition a instruction set can be a GPU
Deanjo: Such as altivec
starkmjolk: vadi2: for the moblin things, yes. I'm starting to suspect it isn't as good of a thing as intel wants it to sound like
starkmjolk: vadi2: the battery times for the m528 really put me down, and it has a fairly beefy battery
cxo: an instruction set can be a gpu, but it doesnt do much justice if the hardware underneath is doing linear math
vadi2: starkmjolk: time will tell. what is the battery life on m528?
vadi2: starkmjolk: and we'll be able to use powertop and whatnot, remember?
starkmjolk: like my laptop, 3 hours if you're really lucky
cxo: as the microcode gets really complex then
starkmjolk: 2 hours to be fair
vadi2: hm that's not so exciting, but those are windows benchmarks too
starkmjolk: no, m528 is only delivered with linux moblin core
starkmjolk: no win even available
vadi2: ah
Deanjo: Right, the point being though is that there is a very real shade of gray as the term can mean so many things
vadi2: hrm. I definitely hope that improves, I was hoping for ~5 hours
cxo: Deanjo, yes
starkmjolk: vadi2: that's what TI has delivered over the last few years and will continue too, intel can't match it with x86, noone can. atleast that's what I'd like to think :)
Deanjo: Marketing wise, Nvidia was the first to market a product as a GPU
vadi2: starkmjolk: does the TI thing run ubuntu mid or mobile?
starkmjolk: to be honest I don't know right now, debian armel is going pretty well so the base is there, but ubuntu have snuck up pretty close to intel lately so the risk is that it'll be x86-32 only
vadi2: cool. and that's a good thing, they were all lonely for a long while..
starkmjolk: I don't like hardware bound to software and vice versa
starkmjolk: that's one of the great things keeping the microsoft monopoly
vadi2: true, but if it'll run linux, chances are it'll run another linux easily
starkmjolk: but I like ubuntu, and I'd like to have it on hardware I like too :)
vadi2: or with some tinkering (ie eeepc)
m-c_: oooh... nokia has oss plans for symbian
vadi2: yeah it's great. only thing that worried me is that everybody else jumped ship
starkmjolk: indeed m-c_!
starkmjolk: vadi2: the symbian foundation will have quite a few members, but nokia could almost pull this one off themselves :P
vadi2: starkmjolk: no what sucks is that everybody else didn't go along
starkmjolk: since ericsson screwed up in the middle 90s they've been in a pretty sweet position :)
vadi2: starkmjolk: if they did, that would've been great, but shows that it's still a small field of 'believers'
starkmjolk: vadi2: http://www.symbianfoundation.org/ list looks full enough for me, some go WM but they are traitors anyway :)
vadi2: Sony Ericsson, Ericsson, Panasonic and Siemens jumped ship and they aren't exactly small
starkmjolk: they sold their parts of the old, soon to be gone, software firm. sony ericsson will be one of the biggest contributors to symbian foundation by getting UIQ there
vadi2: oh? that'll be good then
redeeman: now i haven't tried to develop for symbian, but isn't it kindof a piece of shit?
redeeman: that's what everyone ever having developed for it tells me
starkmjolk: I've heard good stuff and bad stuff
starkmjolk: kinda fat to get out the essential part, but page 9 shows the real deal: http://www.symbianfoundation.org/files/AnnouncementPresentation.pdf
starkmjolk: though this won't happen over night, they'll launch the whole foundation thing first half of next year, and it'll take two years or so before the last row of code will be open
starkmjolk: 1. Nokia to acquire Symbian Ltd – 4th Quarter 2008
starkmjolk: 2. Contributions made to Symbian Foundation – 1st Half 2009
starkmjolk: 3. Symbian Foundation launched – 1st Half 2009
starkmjolk: 4. First complete Symbian Foundation release – 1st Half 2010
starkmjolk: 5. Platform assets made available as open source – Over next 2 years
starkmjolk: that's the timeline, and now I've spammed enough for a week! :)
vadi2: good plan
cxo: does the tar -C param actually work?
cxo: doesnt seem to do anything...
starkmjolk: cxo: never tried actually :)
starkmjolk: nite all!
cxo: gnite
vadi2: *opens ekiga for the first time* *sees the title Ekiga Configuration Assistant (1 of 10)* *closes* *adds PPA, installs snapshot version* *sees (1 of 7)*
vadi2: just a few more releases, and hopefully we'll get that number down.
redeeman: Kano: hey
Kano: hi redeeman
Kano: did you really ever try ati oss drivers?
redeeman: well yeah i have it on my laptop
redeeman: but i don't really use it
redeeman: im seriously considering buying an X1650
redeeman: what performance level you think ill get on it, compared to an nvidia 6600gt?
Kano: i dont know any benchmarks for that card
redeeman: i think that the free drivers will have same performance delta % as they have on yours
Kano: well it depends on the functions used
Kano: it is surely fast enough for compiz
redeeman: any information you can give about what is slow, and what isn't
redeeman: i don't care about fancy smancy stuff as compiz
Kano: you will not be able to play doom3
Kano: that engine is too new
Kano: maybe up to quake 3 engine
Kano: and not at really high res
cxo: http://www.tomsguide.com/us/nokia-symbian-cellphone,news-1769.html
redeeman: Kano: but how can mesa state opengl 2 support if it doesent have glsl? isn't it part of the spec?
Kano: glsl is a language which needs to be "compiled"
redeeman: yeah i know that
redeeman: i thought amd said they would provide code to do that?
redeeman: for the radeon hd stuff
Kano: you can be sure that will not happy this year
redeeman: so the opengl 2.x support in mesa is only on pure software?
cxo: booo nvidia and yay amd
cxo: what a twisted fate
Kano: nope
Kano: the basic things are of course accellerated
Kano: just not the full spec
cxo: damn my 7600gs is gonna be worth nothing now..
Kano: when you use software only rendering usually everthing is rendered
redeeman: doesent it fallback to software for unsupported stuff?
Kano: when you have got partly supported hardware rendering, then parts can just miss
Kano: do you know glxinfo?
Kano: that shows all supported extensions
redeeman: yes but as said, haven't bought the card yet :P
cxo: you cant partly soft/hardware render things
Kano: you can use it for playing videos, compiz, old games, but really nothing new
redeeman: cxo: you can fall back to software for many things
cxo: yes, but you cant mix
cxo: you either send your commands through libGL or mesagl
cxo: not both
cxo: they implement the same api
cxo: ubuntu MID looks great
cxo: i wanna try this...
cxo: hey michaellarabel
michaellarabel: cxo: Hi
cxo: a few colleagues and i were wondering what you meant by "free software zealots could make the decision (or several decisions over time) that would block critical elements of the kernel that are necessary for the binary drivers to work"
michaellarabel: expanding the coverage of GPL_ONLY symbols and some patches that have been proposed before that would significantly break the binary drivers.
cxo: the "free software zealots" are the people eg on that kernel driver blob statement?
michaellarabel: Some could be called that.
cxo: why would a kernel dev want to do that?
michaellarabel: Because of their beliefs in free software
cxo: they aren't just mythical hippies with long hair and smoke weed... love/peace etc.., they are all heavily paid employees
redeeman: believe me, the availability of a free driver for nvidia/amd hardware will change nothing
redeeman: if at some point something nvidia uses, is considered an internal interface, they _WILL_ export it with gpl
redeeman: and nothing nvidia says is gonna change that
redeeman: and torvalds won't interfere, he already said that
cxo: all significant kernel dev is paid by various companies that use linux in their products
redeeman: cxo: all kernel developers are significant
Kano: that gpl only thing is something really stupid. like gregkh forced a patch for usb in 2.6.25
Kano: and there are still not replacement drivers
cxo: kh can be an asshole sometimes
redeeman: which replacement drivers?
Kano: for avm hardware mainly
redeeman: well.. export_symbol_gpl is only a guideline
Kano: his point was that those binary drivers could be written in userspace
redeeman: was it an internal thing?
redeeman: gregkh is no fool
Kano: that will be the truth, but a company that did not update the binary only drivers after release will never create new drivers for hardware which is +10y old
Kano: then the hardware is just no longer supported
redeeman: that's not the kernels problem
Kano: well it forces that you do one of 2 things
redeeman: the kernel can not slow down because some weird external company doesent want to play by the "rules"
cxo: hmmm
Kano: a) revert the patch and be able to use the hardware you have got
Kano: b) buy new hardware as you can not use it anymore
cxo: the "weird external company" is what writes the kernel
redeeman: no it isn't cxo, because then the driver would already be upstream
redeeman: and there'd be no issues at all
Kano: so what would you do, if you would be affected
redeeman: Kano: probably get new hardware
Kano: well i prefer a)
redeeman: but really, even if it's an annoyance for users, it's NOT the kernels fault
redeeman: the kernel is more than willing to host even the most obscure drivers
redeeman: and maintain them indefinetly, as they are used
redeeman: really, hardware vendors that are not assholes have an EASY time
Kano: do you think when a vendor bought licences from other companies to write drivers they would allow em to write open source drivers?
Kano: i highly doubt that
redeeman: i think that how the company decides to deal with drivers for the HARDWARE they are selling, is not my concern
redeeman: as far as im concerned, if their drivers arent free, and merged, they arent supporting linux for their hardware
Kano: you can do what ever you want with code when everything was created by your own
redeeman: well that's not the consumers problem
redeeman: then they will simply have to purchase the code, or write it themselves
Kano: well would you support hardware which is 10y old with new drivers?
redeeman: if they had made proper drivers back then, they wouldn't need to write new drivers
cxo: something is going to happen soon
cxo: the kernel cant keep going like this
redeeman: "like this" ?
Kano: only linux has that gpl only thing, no commercial system ever had that
cxo: yeah, its not a desktop operating system kernel to replace win3.0 anymore
redeeman: Kano: because commercial systems doesent document the internal interfaces
redeeman: so people don't use them
redeeman: and IF they do
redeeman: the commercial vendor doesent care at all if it's broken
cxo: its a big bloated image covered in obscure hardware support for some big blue chip
redeeman: cxo: drivers are pretty much self contained
Kano: well driver development works a bit different. usually commerical systems have got even a much more stable abi, so that you can reuse drivers more easyly
cxo: there is a 9800GT, 9800GTX, 9800GTX+ and 9800SSC
cxo: wtf , did they run out of names...
redeeman: linux is different
redeeman: linux isn't your run of the mill closed system
Kano: yes abi are changed really often
redeeman: and for good reason
redeeman: and it's a BENEFIT
Kano: not much, but enough to need patches for older drivers
redeeman: the drivers are updated as the api changes
redeeman: vendors are just gonna have to realize that linux is NOT some commercial system
redeeman: if they wanna support it, they cannot use the EXACT same model as on the closed systems
cxo: its a free boat ride
cxo: thats what it is
Kano: as those binary blobs are surrounded by an open source loader usually anybody who knows the changes can make em run
cxo: thats true
cxo: the wrapper is very simple to write, its just function pointers
Kano: it is not so much different from firmware, just that firmware is uploaded to the device itself
Kano: and those binary blobs are inside the kernel module
cxo: yes, having your scanner crash is not a big deal, but your entire pc....
Kano: when the module is fully open you could look for errors inside it - but i highly doubt that so many users are willing to do so
redeeman: Kano: for performance i think actually that X3100 is fine
redeeman: shouldn't an R500 with free drivers match that?
cxo: heh, there was only 1 patch from v2.4.36.5 to v2.4.36.6
Kano: redeeman: i guess so.
redeeman: oh well, i guess ill order the card tomorow
redeeman: i mean ffs, it's only half an hours work :D
cxo: but the power users and devs like you and i can fix it and push it back Kano
Kano: i can only do very simple changes
Kano: i never found a fix for a crash myself
cxo: oh
Kano: creating some patches for kernel abi changes is the max i did
cxo: about 50% of my work deal with kernel space
cxo: but anyways, that wasnt my point
Kano: this took enough time for me as i have to diff the kernels to find the differences
cxo: it doesnt matter if the n00b running ubuntu cant debug his driver, his distro maintainer will
Kano: and some other patches i made, like supporting the guitar hero guitar to the xpad driver, this patch is from me
Kano: but that whole patch was about 2 lines huge
cxo: so tomorrow the 4870 is coming out...?
cxo: is getting one!!!!!!!!
cxo: has been putting together his pennies for a while now
cxo: stares at the porcelain pig
Kano: i made it because i bought the game + guitar and wanted to use it with frets on fire
Kano: *g*
Kano: http://kanotix.com/files/kernel/unused-patches/xpad-2.6.24.patch
Kano: i would like to know how to revert the Y axis values as they are reversed
Kano: jscal can fix this however
cxo: http://www.theinquirer.net/gb/inquirer/news/2008/06/24/nvidia-price-protects-channel
cxo: negate the value or something, or find its range and invert it
cxo: are u reading this...
cxo: nvidia is totally loosing it
Kano: well intel will be a really hard competitor when they begin to produce gfx cards
redeeman: i hope they beat nvidia to death
Kano: that will not happen too soon, as they need to learn - and in most cases you have to learn from your own failures
Kano: but maybe in 10 years it could be really problematic
Kano: maybe then nvidia sells cpus too, they already began to produce arm cores for mobile devices
redeeman: i still hope they beat them badly
redeeman: and i hope the nvidia CEO gets fired
redeeman: and have to live on the streets
redeeman: and then has to beg for money
Kano: hehe
redeeman: then i would fly there
redeeman: and shove an objdump of his crappy binary driver, into his begging hat
cxo: did you read the thing on the yields of the wafer?
cxo: the channel partners are not making anything
cxo: yeah serves the mother fathers right
cxo: keep your binary blob bitches
redeeman: one can only hope that the persons responsible at nvidia comes to regred it, badly
cxo: " Nvidia has no effective countermeasure to the R770, the GT200 was quite simply botched, and now they are going to pay for it."
cxo: my first ATi card, what a time to buy....
Exopaladin: eh, they had a good while selling tons of 8800GTs :P
redeeman: but you cannot use it properly? why buy now?
cxo: who bought the 8800gt? it came a month after the 8600gt and 8800GTX
cxo: everyone bought one of those
cxo: only a bunch of poor linux users bought it when the prices dropped
Exopaladin: heh, most people I speak to in other chans didn't buy the 8600GT or the 8800GTX and ended up getting the 8800GTs when they came out
cxo: heh i was right then
Exopaladin: mainly because the 8600GT wasn't a massive improvement over the last gen and the 8800GTX was EXPENSIVE
cxo: k the 4870 is $329 and is coming out tomorrow
redeeman: how much will it be in europe? double?
Kano: why dont you buy now the 9800 gtx after the price drop?
Kano: i am sure you will have got more fun with it, no wine problems specific to fglrx...
redeeman: i agree
redeeman: the price drop hasn't happened in .dk, lol
cxo: redeeman, 249 pounds, including vat
cxo: Kano, i plan to buy a hard drive as well and install windows
cxo: heh the 8800gt is like sub $150 now
cxo: haha
cxo: man.......no one is preordering the 4870 yet
cxo: whats a bit scary is, cos the 3870x2 is $440 ish, and the 4870 beats it, will the 4870 price go up, or the 3870x2 go down
cxo: holy shit the 4850 is selling for over $370 in canada
Kano: the 3870 does not work with fglrx yet, i published a hack however to try
Kano: err the x2 i mean
cxo: (oh that price was for 2)
cxo: statement from bfg http://channel.hexus.net/content/item.php?item=13957
m-c: How many people are still out there saying, "Oh gosh, if only I can squeeze another 2 frames per second!"
m-c: Well, Michael, for one, that's a given, of course. ;-)
m-c: People can run 2x-4x AA graphics at super high resolutions, even on mid-range ($100) cards, today. Is NVIDIA's strategy: "we will be fastest!" Seriously?
Kano: play crysis ;)
cxo: Maybe so, but one thing’s for sure: any partners who aren’t happy with NVIDIA’s handling of the situation now have a good alternative GPU option.
m-c: no thanks
vadi2: Play Spring RTS
Kano: what is that
vadi2: The game handles up to um.. 16 * 500 units on screen. brings any card to it's knees
cxo: poor channel partners
m-c: Spring ran great on my old 6600GT
vadi2: a really nice strategy game.
cxo: imagine you are in their situation
cxo: ok so atis best card is a 3870x2, over priced, no comparison to even the 8800GTX, so you go ahead and buy the next gen 9800GTX and the 280GTX cards, then all of a sudden you got #199 cards making $600 nvidia cards look stupid
cxo: what do you do...
Kano: but the 3870 x2 does not run with fglrx without hacks yet, the pci id was not added - would be really easy for ati to run it as single 3870, but for some strange reason they prefer not run it at all
michaellarabel: cxo: The Linux benchmarks of the Radeon HD 4850 will probably be on Thursday...
m-c: Fantastic.
michaellarabel: Though unsure yet whether I'll be allowed to publish the Linux CrossFire benchmarks early...
Kano: michaellarabel: do you use wine for some games?
michaellarabel: Not for this article coming out Thursday.
Kano: well i setup wine to play trackmania nations forever
cxo: michaellarabel, the 4870 is coming out tomorrow, do you think you can get your hands on one
Kano: which worked really nice with nvidia
Kano: no go for fglrx...
Kano: nothing changed only gfx card
redeeman: yeah wine doesent work with fglrx
redeeman: it's really weird
redeeman: and fglrx seems to be put in weird states by wine
cxo: "Looking into the future, it appears that this may be one of the last times we’re discussing the GPU market in terms of NVIDIA versus AMD/ATI. Intel’s discrete graphics offering – codenamed Larrabee – is expected to make an appearance in 2009 and it will have the backing of a company that’s several times the size of AMD and NVIDIA added together."
michaellarabel: cxo: I can't comment on the 4870 for another 2 hours, 18 minutes... Though I will say I won't have the Linux results ready this week.
redeeman: so that you can start a game in wine, then next time you start, it won't run
cxo: you're playing with the big boys now...buaahahahaha
redeeman: and then one needs to run a native opengl app to "reset fglrx"
cxo: they issue people like you NDAs as well michaellarabel ?
michaellarabel: cxo: Of course, been under various NDAs with them for ages.
cxo: so you have an agreement with them?
cxo: do they send you hardware?
Kano: redeeman: well sometimes i need 2 tries for trackmania but then it runs with nvidia. it only runs with working internet btw
cxo: "... warning: “A lot of companies have gone out of business trying to compete with NVIDIA on GPUs.”
cxo: HEXUS.community :: your
Kano: redeeman: if you want to try it on your own i can help you get it running
redeeman: Kano: i never had problems with nvidia and stuff needing "2 tries", i first saw this with fglrx
cxo: S3 and Savage dont count mr Scott
redeeman: Kano: what is trackmania?
Kano: a very cool racing game, fully free variant available for download with 56 tracks
redeeman: can a quadro fx 3400 drive it?
Kano: i played it on X3100 with win ;)
Kano: should be not problematic the card
redeeman: wishes he had X3100
redeeman: can it be played in highres? :D
Kano: you have to set it to medium details, but i play it with 1600x1200
Kano: you need one registry hack and 2 dlls
redeeman: that might be fun to try
redeeman: but it will have to wait untill probably 1.5 days
redeeman: since povray is still only at 62%
redeeman: and it's kindof eating 400% cpu
redeeman: so i don't have much to spare
Kano: redeeman: i sent you all you need
redeeman: okay :)
Kano: first you regedit the regfile
Kano: after setup of the game put the 2 dlls in the zip files into game folder
redeeman: and just a nonmodified wine can do it?
Kano: sure, used 1.0
redeeman: ok
Kano: dont forget the 3rd file to accept
cxo: wonders if michaellarabel is avoiding his questions
redeeman: i have autoaccept
redeeman: it appears to be stuck
redeeman: ill cancel it
redeeman: resend please
redeeman: Kano: i remember in the old days where wow didn't work in vanilla wine
redeeman: that was when i started developing my hate for nvidia
redeeman: i was the person that solved the nvidia crap wow+wine bug
Kano: not bad
redeeman: the crappy nvidia driver reported incorrect information
redeeman: so wine crashed
Kano: but i dont play wow ;)
redeeman: i don't either anymore
redeeman: haven't for a long time
redeeman: the nvidia driver reported wrong drawables, lol
redeeman: that's actually really something, even fglrx did that part correct
redeeman: that's got to be a first where fglrx does it right :)
redeeman: anyway, kano, ill let you know how it goes
Kano: do you know the download url?
redeeman: but probably it will take another 24 hours for povray to finish
redeeman: no but isn't it on the site?
Kano: it is easy to find
redeeman: if i cant find it, ill ping you :)
redeeman: 63% finished it is :D
redeeman: and it's spend 52 hours 22 minutes
Kano: http://files.filefront.com/TrackMania+Nations+Forever+Free+Full+Game/;10013754;/fileinfo.html
Kano: well you have got a quad core
Kano: you can easyly play a game
Kano: while rendering
m-c: cxo - a big part of signing a non-disclosure agreement is not disclosing the information ;)
redeeman: it's using all the cores
Kano: until you nice it to -20 it will not hurt wine
m-c: I signed one recently that explicitly said that I could not even discuss that signed the agreement.
redeeman: are you sure?
redeeman: it seems to slow down paludis considerably
Kano: then renice it a bit
Kano: with a positive value
redeeman: povray seems to be out of control
redeeman: i tried to bg it with ctrl + z
redeeman: but it ignores the signal, lol
Kano: do you know renice?
redeeman: yes
redeeman: but renice in my experience doesent typically have enough effect
redeeman: sched_idleprio was alot nicer for this
m-c: redeeman: I saw a demo of sending a rendering job to google's app engine
Kano: Meep = mtippett?
redeeman: seems so
Kano: hi Meep
m-c: redeeman: they created 15 VMs and made a 30 minute rendering job take 4 minutes :)
Meep: whistles quietly and looks around innocently.
redeeman: m-c: yeah, this ones been going on 52 hours already
m-c: do a search on youtube for the demo
Kano: redeeman: hopefully your pic will render nicely then *g*
m-c: and you pay by the hour for the VMs, so when your job is done, there is no monthly fee ;)
redeeman: Kano: actually it looks to be a failure :(
Kano: is there a cuda optimized povray?
redeeman: doubtful
redeeman: povray is moving very slowly
redeeman: the devs may only first be hearing about cuda right now
vadi2: isn't cuda proprietary?
redeeman: yes
redeeman: which is another good reason against it
Kano: pah
Meep: http://www.khronos.org/news/press/releases/khronos_launches_heterogeneous_computing_initiative/
Meep: CUDA is to OpenCL what Cg was to GLSL
vadi2: yes, so cuda should not be used! it's proprietary *hiss*
m-c: CUDA? How about all of NVIDIA? I am ready to have an NVIDIA bonfire party after their announcement today.
vadi2: what did they say
m-c: They said the kernel developers did not know what they were talking about, in regards to drivers.
vadi2: in the business sense sure
m-c: Actually, in a technical sense. NVIDIA believes keeping their drivers closed is the best approach for linux graphics.
redeeman: ill bet they believe it's the best way for them to assrape their customers
m-c: Despite the fact that a petition was created, signed by linux developers, saying otherwise.
vadi2: m-c: is it not? they made the best linux drivers about
m-c: But what do linux kernel developers know... they probably have never played Crysis
Deanjo: but not signed by the project leader lol
vadi2: m-c: prolly powered by hollywood, but still consumers get a nice piece of it.
vadi2: it's worked great so far
vadi2: the only open-source card isn't in general production. best open-source driver is the "ati" one that runs on my t40. *shrug*
vadi2: (it's the best because it runs compiz and suspends/resumes/hibernates without a hitch)
redeeman: the linux kernel team maintains drivers for a very wide variety of hardware, which generally is of alot better quality than nvidias driver
m-c: with intel and amd open graphic drivers, nvidia is out-moded, as far as I am concerned
vadi2: I mean graphics.
Kano: redeeman: http://fastra.ua.ac.be/en/index.html
Kano: do you know that
vadi2: still, atm, so far, they've been best.
redeeman: they are not the best for stability
vadi2: say that when they'll be proven wrong, then it'll be of some value. this is all too early imo
redeeman: or security
vadi2: no comment, I don't experience issues
cxo: oh sweet
cxo: the first 4870s have hit hong kong
redeeman: so you never had the drivers installed those YEARS where the nvidia root security issue existed?
cxo: $379+shipping
redeeman: or the MONTHS where it was KNOWN to exist, but nvidia didn't bother patching?
Kano: Meep: could you explain why fglrx does not add the 3870 x2 pci id when it is possible to add that id to the fglrx kernel module + chipid override to 3870?
m-c: Meep: Any idea what is up with that Khronos Group? They formed back in 2006, but I've never seen anything produced by them
vadi2: aren't they behind opengl?
m-c: ... other than OpenKODE that was sort of not really used
m-c: http://www.khronos.org/about/
cxo: http://translate.google.ca/translate?u=http%3A%2F%2Fwww.tomshardware.tw%2F574%2Cnews-574.html&sl=zh-CN&tl=en&hl=en&ie=UTF-8
Meep: A couple of lead-in comments... I lurk everwhere... But don't talk often... I *will* reserve the right to not response... Don't take it personally.
Meep: The 3870X2 is not supported - there are a few reasons.
redeeman: bridgman said it would work but only as a single card
Kano: yes some verfied that already using chipid override
Meep: Also remember that the external - ie: what you guys see - is a small part of what goes on.
Kano: well when you want to use fglrx you must have lots of time ;)
redeeman: that piece of information doesent really help users, Meep
Meep: m-c: the Khronos group is the new ARB. They run the standardization for *all* of the OpenGL variants.
Meep: redeeman: I understand that, but at this stage all I can say is that it currently is not supported.
redeeman: that seems to be the response for alot of things
redeeman: don't take this wrong
m-c: Meep: Ah, I see, thanks
redeeman: but how can you really claim ati hardware as supported?
Meep: There are a lot of things that I cannot say publicly.
Meep: Define "support"
redeeman: as in.. being able to properly use the product one bought
m-c: ...like george carlin's seven dirty words...
redeeman: as in.. it renders the opengl stuff properly
redeeman: (kano has info on that)
redeeman: as in.. it works with the same opengl stuff as OTHER drivers does
Meep: *sigh*... Be careful of which paths you want to go down...
Kano: Meep: do you use hardy currently?
Meep: Intrepid
redeeman: Meep: i just don't see how you(by that i dont mean you personally) can in all seriousness claim to support linux, it's just misleading customers
Kano: hard to test with live cds as it currently does not exist
Meep: My use of "support" is that I will put an engineer on the product and the issue to resolve an issue.
Meep: redeeman: There are *many* people that it does work for.
redeeman: Meep: because they only do a small set of things, which works
redeeman: Meep: ill bet for instance, that wine doesent work for ANYBODY at the same level it does with other drivers (hint: nvidia)
redeeman: anyway, i gotta sleep
Meep: ponders how to respond...
redeeman: you don't have to
Kano: Meep: did you try my lenny repo for gl2benchmark?
Kano: deb http://kanotix.com/files/fix/gl2benchmark.lenny ./
m-c: WINE is an intermediate, jury-rig solution while more applications are matured on the linux platform. Personally, I would rather not see much optimization or development effort focus on WINE.
Kano: verified with hardy, maybe it works for intrepid too
Meep: Kano: No. I contacted the author. gl2benchmark isn't really something that he supports.
Kano: well i know the author of course
redeeman: Meep: i can see how you support linux by that definition of "supported", since it doesent actually means that stuff must work NOW, but rather "we will work on it... whenever"
redeeman: Meep: but i believe most people think of support as "stuff works now, and the few bugs that MAAYYY be, will be fixed promptly"
Meep: Work on it relative to priorities and mixed in with other issues.
Kano: basically the newest versions are on my site
Meep: Can't make everyone happy.
redeeman: Meep: for instance, ill bet, that if some big game, like crysis, didn't run on your windows driver, that would get fixed an order of manitude faster than opengl compliance we see with fglrx
Meep: Quite possibly. But that would be mainly financial.
redeeman: m-c: wine itself doesent matter, compliance does, which wine seems to expose
Kano: Meep: do you know the people who wrote this: http://ati.amd.com/developer/rendermonkey/
Meep: If point-sprites were heavily used by the workstation market, it would be fixed an order of magnitude faster than being visibly a problem in 1 obscure (sorry Kano) benchmark and 1 game (flightgear).
redeeman: Meep: so then, shouldn't the relevant litterature say: "Linux: supported (but only at 1/10000 the speed of our other supported OS's)"
m-c: redeeman: Well, let's get a benchmark that exposes all these compliance points, as you say, and use that. Let's not measure ourselves on how well we run windows applications.
Kano: Meep: they are used in simulations more often. that benchmark tool is definitely not obscure, it is open source and based on open scene graph
redeeman: m-c: wine is the benchmark, if you wanna spend time re-inventing a testcase, that's your thing :)
Meep: Sure... Who do you think pushed PTS to have a GL compliance test.
m-c: redeeman: wine is hardly a benchmark - wine is a hack
Kano: Meep: and interestingly the render monkey tool is based on osg too
redeeman: m-c: wine is an implementation of an api
m-c: the hell it is..
redeeman: m-c: there are lots of native applications using wine
Kano: Meep: saidly the code for the viewer does not compile with gcc
redeeman: m-c: for instance, i have ported some inhouse winblows apps to linux using winelib
redeeman: m-c: those would undoubtedly have the issues, had they used opengl/directx
Meep: I'm going to go quiet for the moment... A suggestion, read between the lines in most of my comments... Draw your own assumptions.
Meep: Redeeman: What literature are you referring to?
m-c: no distro will be redistributing your ported windows applications running through wine any time soon, so excuse me for not caring
Kano: Meep: well a bug introduced about 1 year ago and nothing fixed in the meantime is really good for marketing ;)
Meep: Which bug?
Meep: Pointsprites?
Kano: yes, the modeline bug is even older
redeeman: Meep: i don't know, that new fine stuff talked about on phoronix with boxes with tux logos on, and driver releases and stuff, saying "Supported"
Meep: Meep: If point-sprites were heavily used by the workstation market, it would be fixed an order of magnitude faster than being visibly a problem in 1 obscure (sorry Kano) benchmark and 1 game (flightgear).
Meep: r
redeeman: but why doesent it work with wine? surely there are LOTS of other non-working opengl stuff
Meep: Don't know the Wine devs most likely use NV. Consequently they release based on that, and choose to interact with NV.
Kano: but shouldnt a driver render everything correctly? there is a standard and the driver definitely does not fullfill it in that case
Deanjo: ponders what to do with all this old NFR software
redeeman: Meep: i believe that wine also works with a few of the open drivers, just slower
Meep: Sure.
Meep: points to my comment 2 or so lines ago.
Kano: but 1 year?!
cxo: are u guys still arguing about wine
cxo: just buy a bottle already
Meep: NV has lots of bugs, Windows has lots of bugs... How would it be prioritized?
redeeman: Kano: well :P nvidia has spend more than 2 years not fixing an edid bug i reported
redeeman: Meep: i would think fixing a rendering bug would be higher prioritized than a fancy looking amd control center
michaellarabel: redeeman: I have an NV bug going on 3+ years old that's still not addressed...
Kano: well the visible bug fixes are really small in fglrx, even more bugs seem to be introduced with new releases
redeeman: michaellarabel: yeah i believe that, they don't seem to fix ANYTHING reported, unless they personally care about it
michaellarabel: They'll give you a bug tracking number... which is useless.
redeeman: michaellarabel: the edid issue i got, i reported, along with all relevant information, ieven wrote an edid parser which CORRECTLY parsed my monitors blob, but they haven't fixed it
Kano: well in most cases nvnew.net should be used i think for nvidia bugs
redeeman: every few nvidia driver releases, i hook up the monitor and test if they have fixed it, but they never have :)
Meep: Kano. Can you enumerate the publicly accessible applications that expose the pointsprite bug?
Meep: Is Brains on?
redeeman: Meep: i imagine a rendering compliance suite will :P
Deanjo: sends Kano a still sealed copy of Partition Magic 2
Meep: redeeman: Do you have a rendering compliance suite?
redeeman: no, use kano's thing
redeeman: Meep: i do have one piece of information that MAY be helpful for you
Kano: Meep: for the modelines: any modeline crashes xorg 7.1.1 - base install of etch is all you need to verify
redeeman: Meep: when i tested fglrx last, i was able to start wine+wow sometimes, but after a successful start, fglrx got in a broken state, only fixed by running neverball, then i could run wine+wow again
Meep: *sigh*. Redeeman: "gl2benchmark  wasn't developed further since mars 2007 , but it workes very good for my daily testing needs.  I just made gl2benchmark to test if our VisSim-applications could run under Linux etc. " From Markus Hein, the author of gl2benchmark
vadi2: redeeman: what video card are you using? You hate both ati and nvidia
redeeman: vadi2: i currently have an nvidia card in
vadi2: redeeman: why?
redeeman: Meep: so what about the purpose? does it matter? it exposes a bug?
redeeman: vadi2: because it was the only card i had laying around
Kano: Meep: it was developed as a showcase tool for live images with 3d driver autoinstall feature
vadi2: redeeman: you're being hypoctirical.
vadi2: redeeman: stop using it right now!
redeeman: vadi2: how so? i cannot say something stinks while using it?
cxo: heh ati likes to release cards without drivers, thats nice
Meep: Redeeman: I can find lots of applications that expose lots of bugs in lots of things... The eventual impact to the market is the important thing.
vadi2: redeeman: but it's sucks badly. don't use it.
cxo: apparently the suckage on the 4870 on windows is because of no proper drivers
redeeman: vadi2: actually it does suck so bad that i have refrained from using it
Meep: Kano: Your use of gl2benchmark is that. There is a big difference.
vadi2: redeeman: and continue refraining. why did you put it in?
Kano: Meep: as his entrepreneur uses open scene graph - to show how well it works with linux
Deanjo: didn't they just release a hotfix for the 48xx series cxo for windows
Meep: Kano: Yes. But he wrote it for testing the *NV* drivers.
Meep: My discussions with him indicated that he has never run it on ATI.
cxo: Deanjo, i'm reading a review in the future, and apparently that didnt do anything to performance
Kano: sure because i told him that it does not render correctly there
Kano: i could not recommend him to buy ati, do you get why?
Deanjo: lol, STOP advancing your computer clock cxo, that's cheating
cxo: haha
Meep: Kano: For the modeline issue.. Just put a one line bugzilla in ati.cchtml.com.
cxo: the asians have the cards already, so there are a bunch of reviews sprouting everything
Kano: Meep: but there is already a modeline entry
cxo: we should get our english reviews tomorrow
Meep: Kano: Which bug? I don't want to invest a developer in attempting to reproduce a problem.
vadi2: redeeman: i'm waiting for an answer, or your talk will mean nothing since you yourself are using it.
redeeman: vadi2: wtf are you talking about? why cant i say something is crap, while use it?
Kano: Meep: you can add any modeline, therefore the sample should crash etch
Meep: Kano: We *do* have modeline testing.
Kano: i basically add about 100 modelines
cxo: here's some marketing stuff from ati http://64.233.179.104/translate_c?hl=en&sl=zh-CN&tl=en&u=http://www.tomshardware.tw/550,news-550.html
redeeman: vadi2: also, what is that weird seemingly-utf8 weird sign at the beginning of all highlights you do?
vadi2: redeeman: you sure can. "it sucks badly, don't use it" "(but I'm using it myself)"
vadi2: that's pidgin.
Meep: Kano: I can't help you if you want to push back too hard.
Kano: http://kanotix.com/files/fix/xorg-conf/xorg.conf.ati-aiglx
Kano: that is an example
Kano: lots of modelines
vadi2: redeeman: being a hypocrite won't get your advocating anywhere.
Kano: also you can try somethign without modelines too
Kano: when you select res 1152x864 - works without modeline too
vadi2: redeeman: but you can continue, no prob. just warn people ahead.
redeeman: vadi2: i didn't write here "it sucks badly, don't use it"
Kano: then you can try amdccle (or xrandr -r 100)
Kano: that will be applied
Kano: but not used
Kano: when you check the monitor
vadi2: redeeman: if pidgin logged this channel (dunno if it did), I surely can find lots of examples of you furiously bashing nvidia to an obnoxious point.
redeeman: vadi2: yes BASHING
redeeman: vadi2: and how would i know it sucks, if i didn't use it?
vadi2: redeeman: don't use it if it sucks so badly and keep quiet. nobody's forcing you to use it.
Meep: Kano: How many other users do you know of that use that many modelines?
Shahid: lets make an open source computer, as in the electronics
cxo: nvidia and microsoft sitting in a tree, p.i.s.s ing me off
redeeman: vadi2: i am warning people that the drivers are crap
Kano: Meep: all kanotix users, as this is the default here
vadi2: redeeman: instead you're forcing yourself to use and you continue bashing it because you keep using it.
redeeman: vadi2: and there doesent really appear to be many alternatives on the market
redeeman: except apparently the new R500 dri support
Meep: Kano: No offence, but outside of Kanotix?
redeeman: for which i will order the hardware tomorow for
Shahid: we all complain about the price of oil, yet we still buy it
cxo: whats funny is, there is more official product literature on non amd.com about the 4800 than on amd.com
vadi2: redeeman: great.
redeeman: Shahid: oh you must not! vadi2 has spoken!
Meep: Kanotix is not a supported Operating System.
Kano: Meep: kanotix is based on debian etch, that should be supported
Kano: the xserver is the one from debian
redeeman: Meep: lol. that's another crap thing about fglrx, it's useless to most people, if you "dont support" anything but your customers enterprise distributions
Shahid: I'd like to see gpu's integrated into the same physical chip with the cpu
Meep: Redeeman: Out of interest, what sort of work do you do?
cxo: soon the cpu will become part of the gpu, not the other way around
redeeman: Meep: software development and server solutions
cxo: computing is becoming more visual
Meep: Custom work?
cxo: math has always been math
redeeman: Meep: yes
Meep: Redeeman: COotract?
Meep: Erm Contract?
Shahid: i don't know, floating point logic is scary
redeeman: Meep: no, im a consultant
Meep: Basically. Support is very different from "working"...
redeeman: Meep: i also wrote this on the forums, the thing of saying "rhel, sled is supported!!" is useless to us, what we need is "works with X server version blablaba, kernel version gah gah"
Meep: Support for me (and whenever I use it), implies that it has sufficient priority that I am expected to, and will invest engineering effort in resolving issues.
redeeman: support to mean, means that it works, and if issues should arise, outside of normal scope of "working smoothly", it will be fixed
Meep: The "it will wok on" is a completely different perspective.
Meep: Er. work:
redeeman: for instance
redeeman: all the time(even still here) the tearing issues have happened
Meep: So, it should work with debian, gentoo, Fedora, (and even Kanotix).
redeeman: that's NOT "supported" in my mind
Meep: But it is not supported on those distributions.
redeeman: it's broken for various people on all dists
Meep: Redeeman: You can define supported however you want.
redeeman: well your definition seems extremely unique
redeeman: by your definition, having NO driver is supported aswell
redeeman: because you will assign an engineer to create one with 3d support
redeeman: surely you must know, that atleast in the consumer market, NO consumers share your(atis?) definition of "supported"
Meep: In the consumer market, who are ATI's customers for Linux?
redeeman: for instance if i were to buy a 4850
redeeman: for my own personal entertainment and gaming
Meep: Go on.
redeeman: what?
redeeman: i would be the customer then
Meep: You mean a Sapphire 4850?
redeeman: well.. whichever
redeeman: i realize that im not directly dealing with ati
m-c: Saying that you will support something with an "out of the box" linux distro (like RH4, SUSE, or Ubuntu8.04) should be good enough, even if just one specific example. Would give the customer something on which to test.
redeeman: bht that is kindof how it works
Meep: So, have you asked Sapphire if they support Linux?
m-c: Not like it is hard to download and use a new linux distro
redeeman: m-c: no but even if it works there, it would be no solution and an unviable workaround
redeeman: Meep: no, i haven't bought anything from them yet either
redeeman: Meep: i understand where you are going at
m-c: Could even make a test package that could be downloaded and run in an isolated environment
Kano: 58M 2008-06-18 18:55 ati-driver-installer-8-6-x86.x86_64.run
Kano: 33M 2008-06-25 05:03 ati-driver-installer-8-6-x86.x86_64.run.new
Kano: what do you think about a much smaller installer using lzma
Kano: same content
redeeman: Meep: so you are basically saying, that AMD/ATI's graphics dealie, is that you assign that responsibility 100% over to the actual manufacturer of the card, which the consumer(me) buys from?
Meep: Since Sapphire actually bought the chip.
redeeman: well sure okay
Meep: Not quite.. But that plays into it.
Meep: Likewise OEMs.
Meep: Michael captured it in an article some time ago.
m-c: I always here people say there are too many linux "versions", when the fact is there is just one kernel. There are a lot of distros, sure, but just pick one to support.
Meep: m-c: There is not one-kernel. There are many variants of kernels.
Kano: Meep: do you want to test a full lzma compressed installer?
redeeman: m-c: they shouldn't support "distros", but just say what versions of software are required, distributions WILL take care of the rest
Meep: Redeeman: We need a commercial entity to interact with.
redeeman: why?
michaellarabel: redeeman: This may be the article, http://www.michaellarabel.com/index.php?k=blog&i=121
redeeman: why cant you just write a driver that works against specific software versions?
redeeman: nvidia seems to be able to do that?
Meep: m-c: We support Red Hat, Novell, Ubuntu
Meep: and Red Flag.
Meep: Because each version of the software is different.
m-c: meep: you are saying those all use different kernel variants?
Meep: Yes. In the areas that we are exposed.
m-c: huh
Meep: RH has 2-3 variants for large memory.
Meep: How many X versions do you think we support?
m-c: I guess there is 32-bit and 64-bit...
Meep: 6.8.2, 6.8.99, 7.0, 7.1, 7.2, 7.3, 7.4.90
redeeman: well.. if you only support a limited number of X versions, that's fine, just don't go about saying "redhat", say xorg versions like THAT
redeeman: this we can use
Meep: We seem to not currently work with 7.3.99 (in Fedora).
redeeman: Meep: it would be quite helpful if the release notes or something, showed such exact information
Meep: There are hundreds of patches for Red Hat for their X.
Meep: It isn't the same as upstream.
Meep: Unless you are a LInux-from-scratch kind of guy
m-c: I understand the need for a company to have interaction -- You cannot be the only guys on the line for the entire OS working
redeeman: then write "7.1, as in redhat's patches"
Meep: That would be a ridiculously long list.
redeeman: Meep: but if you say that it works with UPSTREAM versions 7.1 for example, that's something _I_, and most other distributions can use
Deanjo: lol ok now that's just getting stupid
m-c: On the other hand, I do not agree that every kernel works differently - like a different type of operating system.
redeeman: well.. so smash them in a tarball
Meep: Since We support RHEL4, RHEL5 (2 versions of that)
Meep: The same for the other distributions.
redeeman: Deanjo: well.. i too would prefer not knowing about it, and it just working with all versions, but so obviously this is not possible for fglrx
Meep: UPSTREAM is mostly irrelevant. Since each distro repackages with their own patches.
Deanjo: redeeman, show me any device driver for any OS that specifies a list patches are needed to be applied before it works
m-c: Do the differences between RHEL4 and Ubuntu linux kernels, really support the statement, "too many versions of linux" ?
Meep: Deanjo: Well an OEM made me do that for RHEL once.
redeeman: Deanjo: well.. that's what i said, i would prefer NOT having to have such a list, like with nvidia, but OBVIOUSLY amd is unable to produce fglrx in such a way, and thus people not "blessed" by running a "supported" distribution, needs some way to know what software to get, to be able to use the driver
Meep: Internally NV will have the same structure.
Meep: We say XOrg 6.8 and later, 2.6 and later
m-c: Anyway, I still say that if someone wants to "support" linux, that they only need to choose their favorite distribution. All the other distros will follow in their own way, if the source code is open.
Deanjo: so you could look at the patches that were created for that distro redeeman. Having ATI repeat what is already available elsewhere is pointless
Meep: m-c: yes. Which is why gl2benchmark is not available on all distributions.
redeeman: Deanjo: how would i know where redhat or sled or anything else keeps their weird stuff? that's why i would prefer to know which versions of UPSTREAM code it works with
Kano: well i only packaged it for debian, i do not know rpm packageing
m-c: Look at the Red Hat model, for example. People found the process was too long to get applications "supported" by red hat, so they recommended they "support" them in Fedora, and RH would acquire them from there as needed.
redeeman: m-c: yes but fglrx isn't open, that cant happen the same way
m-c: If a distro and it's users want the code, then they will find a way to get it - as long as it is F/OSS
m-c: redeeman: so, fglrx has no place in linux, as far as I am concerned
redeeman: m-c: agreed, distributions should simply package themselves
redeeman: m-c: well.. hopefully soon it will be only a myth of what once was
Kano: i can accept binary blobs, but they have to work ;)
Kano: when the new ati 2600 AGP cards came out all i could say to my users was to return the card
m-c: And, I do recommend binary applications - like Opera and games - but that is a tough road and it's not long term sustainable - and it'll never be distributed in distro repos
Kano: they are happy with nvidia now ;)
redeeman: i don't care what APPLICATONS people use, as long as it works with standards
m-c: and what I want is the device drivers for the hardware I purchase to be open so it can be used and improved as I see fit
redeeman: me too
Kano: the time between product available and driver available and working was really too long there
m-c: redeeman: Someone tell it to the Apple users ;)
m-c: Somehow, a whole bunch of linux users got confused and switched to Apple with closed *drivers* and open *applications*... pretty much backward to what we're trying to do here.
redeeman: they probably saw their ipod and bent over, lubed up, and limped to steve jobs
m-c: Saw a guy wearing a Debian t-shirt working on a iLaptop in a coffee shop the other day. Talked to him for a while, and he said he used Linux for years but just wanted an Apple one day.
redeeman: lol
redeeman: must be some sort of mental lapse
m-c: Strange days.
Kano: well os x is boring when you used it more than 2 days ;)
Kano: but ok, some users like it that way
redeeman: m-c: i actually find it very shameful to live in a society which just turns their back on those people
redeeman: those people obviously need help
redeeman: we should be trying to cure them
m-c: with drugs?
redeeman: well.. i don't know what a treatment would be, but.. whatever cures them
m-c: lol
Kano: redeeman: did you every try os x?
redeeman: Kano: yes
m-c: I agree - they should be watched by a team of doctors - monitored 24 hours a day.
Kano: redeeman: the gui is sometimes a bit unlogical. and the apple keyboard is notthing for pc owners - when they heavyly use special keys...
redeeman: the gui is abysmal, it's not even logical or easy to use, and it's FAR too limited
Kano: but ok, it is not really hard to install new apps or surf the web with it
redeeman: not to mention that, it acts extremely unintuitively too
redeeman: and the application install thing is the most insane thing ever
Deanjo: Oh please
redeeman: well sure.. surf the web is pretty much the same everywhere
Kano: well it has nice animations and the new backup feature is interesting
redeeman: the animations also seems mostly weird
redeeman: for instance the way apps go into the dock
redeeman: wtf is that?
redeeman: Deanjo: that's right, not everyone is a steve jobs worshipper thinking apple can do no wrong holy god of ui design
Kano: well i would not say that windows is pure logical nor kde or gnome
redeeman: kde is alot more logical than osx
Kano: it is always a differnt view to the problem, when you are used to one thing you have to seek something else on another plattform
Deanjo: Everybody has their opinion on how a GUI should function, nobody is right or wrong
redeeman: as of gnome - well.. im not sure theres enough left to actually draw a conclusion
Kano: but withing a few days you know all what you need to know and then you get bored
Kano: or you are not really interested in customizing then you are happy that you know how it works -> these are the real apple customers
Kano: former linux users most likely only buy it because the systems look so nice
m-c: Yay, I have finally got debian installed in a VM. *whew* redeeman - would you please walk me through how to save off a snapshot of this?
Kano: m-c: close it, cp the vmdk/vdi just to another filename
Kano: thats all
redeeman: m-c: snapshot, do you mean like.. while it runs, or just like.. a snapshot of the disk to have a base image?
cxo: i agree with Kano
m-c: kano - Do I just do a shutdown?
cxo: after a certain stage, it doesnt matter even matter if you're on bare metal
m-c: because I tried that before and when it came up again there was disk corruption
Kano: sure, to backup the vm you only have to backup the hd image
redeeman: m-c: you can have kvm/qemu base disk image off another, so only changes are written to the new file
redeeman: you can even have the base image compressed
Kano: sure many things are possible, just some are more easy than others
m-c: So, I hear what you guys are saying, but I am still not clear on the steps.
redeeman: these things are really really easy with kvm/qemu
redeeman: m-c: okay, ill explain it
redeeman: these are things you can do:
m-c: I want to be absolutely sure I am making a point-in-time-copy
redeeman: 1: copy the entire disk image:
m-c: because that install was a nightmare
redeeman: just copy the disk image file, that will give you a 100% identical copy
redeeman: (while VM is shut down)
m-c: Can I copy it while it is running and I am logged in?
Kano: not good
redeeman: well.. in the qemu monitor i believe you can freeze it to make a such copy
redeeman: but i haven't tried this
m-c: okay, so when I shut down - do I need to do anything special? Or just shutdown normally?
redeeman: just shutdown normally
redeeman: and you can backup the image file, and make copies as you please
redeeman: if you need lots of identical copies
redeeman: you can use it as a base for new images
redeeman: so that only changes are written to a new file
redeeman: ill give an example with pseudocommands, so you can understand:
redeeman: qemu-img create -f qcow2 base.img 20G
Kano: yes you dont need imageing software as you already work on images
redeeman: kvm -hda base.img -cdrom gah.iso -boot d
redeeman: (now you install and do complex setup)
m-c: fantastic
redeeman: now you want 100 identical vms
Kano: images are just files, you can copy em anywhere
redeeman: so you do:
redeeman: qemu-create -b base.img -f qcow2 vm1.img 20G
redeeman: qemu-create -b base.img -f qcow2 vm2.img 20G
redeeman: qemu-create -b base.img -f qcow2 vm3.img 20G
redeeman: etc
m-c: awesome
redeeman: then only changes made, will appear in vm1.img
redeeman: and such
redeeman: you can also have base.img stored in compressed form if you please
Kano: bye for now
m-c: I want to start doing packaging. Thanks for your help.
redeeman: (use qemu-img convert to convert it to compressed format before creating vm1.img and such)
m-c: okay! I shutdown, copied, and restarted the VM, and everything looks 100% thanks again.
redeeman: yeah that is the correct way
redeeman: i really have to sleep now, but if you have any other questions or such, just privmsg me them, ill answer when i wake up
m-c: Thanks!
Deanjo: lol http://www.msnbc.msn.com/id/25272162/