Phoronix IRC Log: 2009-02-27
Kano: hi michaellarabel
Kano: hi redeeman
Kano: does anybody know how to use create isohybrid iso images
redeeman: i have no idea what that is
Kano: within 1 h i have got a test iso
Kano: you can do cp image.iso /dev/usbstick
Kano: and the usb stick will boot
Kano: i found out why this never worked here... the debian syslinux package is broken
Kano: it works since syslinux 3.72
Kano: moblin used that trick
Kano: you create it the usual way you would do, then you run isohybrid on the iso
Kano: and thats all
soluser: hey dudes
soluser: is Christian Cruz here
mattmatteh: does not know who that is
soluser: he wrote this post on solaris jumpstart dvds
Ivanovic: michaellarabel: uhm, is r6xx branch merged into radeonhd master?
Ivanovic: or am i missinterpreting this stuff?
Kano: hi michaellarabel , did you fix those issues that when you preset no upload in the gui that there will be no question if you want to?
michaellarabel: Kano: Soon I will, easy fix.
Kano: or at least preset it with N
Kano: most users press enter
michaellarabel: BTW, will be adding it so it automatically enables the sym linking from download cache when detected its running on a LiveCD
Kano: if they changed their mind allow y, but default should be like the user wanted
Kano: do you check for unionfs or aufs or how do you detect?
Kano: grep -q -e unionfs -e aufs /proc/mounts
Kano: is pretty easy, how do you do it?
michaellarabel: I already added the support to Git today where if no normal file-system is detected (right now it's been reporting Unknown in the system info area when running on LiveCD), it will detect SquashFS and AuFS. So now I will do a filesystem check for SquashFS from that code already introduced
Kano: you can try union=unionfs with live-helper images
Kano: i guess it even works with ubuntu
Kano: that option
Kano: i told em a while ago ;)
Kano: just compiling wine for lenny, for etch i already uploaded it
Kano: next iso images will have 1.1.6
Kano: err. 1.1.16
Kano: if you like add some wine tests with external wine
chain_: michaellarabel, are you there?
michaellarabel: Hi chain_
chain_: I was just about to go to sleep, and I've found some fresh information that you might find interesting
michaellarabel: What information is that?
chain_: S3 graphics seems to have released a linux driver for Chrome 530GT and 540GTX
chain_: it seems to be only x86 at the moment though
michaellarabel: Wow, you're right.
michaellarabel: Video decoding + OpenGL 3, wow.
michaellarabel: Thanks for taking note of this
redeeman: and it's some binary only crap right?
michaellarabel: I wonder what video interface they are using...
michaellarabel: Looking at it now.
redeeman: also, i don't think their hardware supports duallink dvi
michaellarabel: Oh great, another video interface...
michaellarabel: Actually, three
michaellarabel: S3 now has a Va, VaNox, and XvMCNox interfaces for video playback
chain_: it's good that they have finally released it anyway
chain_: I just hope it isn't as crappy as early ATi drivers
redeeman: at this point they offer NOTHING over ati or nvidia
redeeman: just another crappy closed driver
michaellarabel: Technically they do have more video acceleration than ATI
chain_: I kind'a thought that you know about those drivers, but then I thought - I'll tell him anyway :)
redeeman: yeah nobody gives a damn about their acceleration
redeeman: nvidias probably works better
chain_: I personally think that if S3 wrote linux drivers that would be more efficient than the ones by ati and nvidia, they could sell more of those cards
Kano: redeeman: are you using vdpau now?
redeeman: Kano: nope
redeeman: i have no need for it
Kano: i like vdpau ;)
redeeman: and i am very concerned by your warning that it may crash my X
Kano: when you use bad media
redeeman: i am not willing to risk that for something i benefit largely nothing from
Kano: that usally does not happen with mkv from hd
redeeman: well.. i don't know for certain, and i don't particularly care
Kano: only when you use live dvb with bad signal quality
chain_: I'm very curious what's the performance of S3 cards on linux, I'm hoping it'll be a nice surprise
Kano: must be a joke ;)
Kano: even ati tells the drivers support opengl3,but not even a test app runs with it
redeeman: yeah well it's not hard to believe s3 could have done that far better
chain_: I don't think that opengl3 is very important just yet, what's important are well written, efficient drivers. It's the first release so if they implement basics in a very efficiet way, it'll be easier for them to add some more features later on
chain_: if the initial release turns out to be crappy, then even opengl3 won't help them
Kano: redeeman: far better with first driver release?
redeeman: Kano: i see no reason why they couldnt do it better than ati
Kano: s3 did not even support xv in older oss drivers...
Kano: and now they should be working well in first release?
redeeman: if they really wanted i see no reason why they couldnt do it
redeeman: it's not as if it takes some magic
michaellarabel: This driver package is a mess.
michaellarabel: The OpenGL 3.0 support looks like it might be OpenGL 2.1....
chain_: I've downloaded the package too, and it seems to support x64 although it's advertised as x86 only
michaellarabel: And it looks like their video API might also be VA-API
redeeman: chain_: why do you say that? because of the install64 thing?
redeeman: also, theres no such thing as x64
redeeman: stop using that non-existing term
chain_: redeeman, I'm reading install64.sh now
chain_: redeeman, I know it's not a correct term but it's so easy to type, and I'm so lazy... ;P
redeeman: i don't think it supports 64bit
michaellarabel: The kernel module also claims that it's GPL.
redeeman: they should be severely punished for that
Kano: the kernel module can be gpl if it does not contain a binary part
redeeman: but it does
Kano: then it is incorrect
Kano: libS3G.a.GCC4 requires to use another licence
chain_: maybe they want to release the full source code ? ;P
Kano: but it is definitely not for x86-64
chain_: Kano, you're right - it's not for x86-64, I was wrong
michaellarabel: Thanks again chain_
chain_: michaellarabel, no problem, I'm very happy that I could help
chain_: michaellarabel, do you think you'll get a 540gtx to actually test on linux?
michaellarabel: Only if I end up buying one. I don't know whether or not I'll do that, I'll need to see how interest is from this newest article to see if I can justify the cost.
chain_: I hope there'll be a lot of interest then, fortunatelly these cards aren't too expensive
Kano: and you can really buy em already?
chain_: I've bee thinking about getting a Radeon HD4670, but maybe I'll wait and see how engaged in developig these drivers S3 really is
chain_: Kano, yes - https://s3gstore.s3graphics.com
chain_: I'm not sure these cards are available in Europe(s3gstore ships only to US, and I haven't seen these cards in Europe)
Kano: chain_: i would prefer a nv gtx 260
redeeman: serious price difference there
chain_: Kano, choosing a gfx card is very difficult for me, I know that nv would be the most logical choice, but I just don't like Nvidia
Kano: chain_: i dont get why they compare cpu usage with a nv 8600 and speed with a ati card
chain_: Kano, now have a look at 530GT performance comparison graphs
chain_: ooops, they're the same
chain_: which means that those graphs are meaningless
chain_: goodnight everyone!
Plouj: what's a fanless ati card that is or will soon be supported by any of the Free/OpenSource drivers?
Kano: i would go for a fanless nv card ;)
Kano: because it supports vdpau *g*
Plouj: I'm sure you would
Kano: well all my ati cards are dead, till now nobody wanted to send me one working pci-e card
Kano: and i will not buy it, thats 100% sure
Plouj: I'd rather look forward to gallium and other non-proprietary thingies
Plouj: yeah, one good thing about nvidia cards is lifetime warranties
Kano: never heard of that
Kano: michaellarabel: how about checking wine --version and when it is wine-1.1.16 then use external before compiling it?
michaellarabel: Kano: There is a better way to handle it, but needs some changes in pts-core first
Kano: what would be better?
michaellarabel: The WINE support is provided by a base test profile through the Cascading Test Profiles feature. It would be easy to write a new base test that would use WINE installed into the system (or even the stable branch or cxoffice or anything else), but first pts-core needs support for prompting the user when selecting between multiple possible CTP bases.
Kano: how did you do your wine comparisions?
michaellarabel: which ones?
Kano: the ones you did
michaellarabel: In a Phoronix article? Wasn't using PTS
Kano: running 3dmark 200(1) would be fun too ;)
Kano: btw.didd you notice that your wine benchmarks do not work when you use nvidia with vsync on
Kano: maybe you should check vsync state, disable it while testing and put it to stored state after test
Kano: usually it is good idea to have vsync on for flash video in fullscreen or for playing games
Kano: just not for benchmarking
michaellarabel: There is already vsync handling support in a module I think
Kano: also how about a live update feature
michaellarabel: live update of....?
Kano: that would download the current snapshot and put it into your user dir
michaellarabel: for the test profiles?
Kano: of pts of course
Kano: for everything
michaellarabel: I don't know about a live update of pts-core
Kano: well then for the rest
Kano: how about that
michaellarabel: but of test suites and test profiles, it should be possible with 2.0. All of the infrastructure is in place and I think might even be done in my 2.0 branch
Kano: did you think of instaling the depends from the gui, with prompting for pw if not in live mode
Kano: su-to-root works on debian as wrapper
Kano: dont know what could be used for ubuntu, i dont test that so long ;)
michaellarabel: gksu works on Ubuntu and is used already
Kano: dont think that is installed on a kde only system, then it would be kdesu, but su-to-root might be there too. works with -X option for x apps
Kano: btw. i have got now testimages with wine 1.1.16 ready. for every arch
Kano: hybrid iso
Kano: you can run : sudo install-nvidia-debian.sh
Kano: even in a x console in live mode
Kano: same for: sudo install-fglrx-debian.sh
Kano: that really works fast when running from usb
Kano: a cd has longer seek times
Kano: anybody interested? just pm me
Kano: and of course: pts gui preinstalled *g*
Kano: maybe i should see fast usb sticks *g*
Kano: but i dont have got the really fast ones
Kano: michaellarabel: you could stop the bootup time from stick or from cd or different sticks
Kano: it is not really speed optimized, too many services running by default
Kano: but seek time is reall something that matters
Kano: michaellarabel: do you use a gui editor for pts?
michaellarabel: Kano: No, write all the GTK2 manually.
Kano: ah, ok
Kano: Deanjo: did you test the gui yet?
Deanjo: A few issues with it
michaellarabel: Deanjo: what issues have you had?
Deanjo: but at least with the gui the tests complete
Deanjo: Just mostly trying to open the results in konq
michaellarabel: do they not open or what?
Deanjo: No you just get a text summary with no grpahics
Kano: michaellarabel: did you see 16 gb usb sticks are not so expensive anymore, you could run that fully from stick *g*
michaellarabel: Does Konqueror support XSL?
Deanjo: "In theory" in real life not so well
Deanjo: It may be different in Konq4
Kano: you use kde 3.5 too?
Deanjo: But of course
Kano: well i prefer that too
Deanjo: it's hard to improve perfection :P
Kano: michaellarabel: do you prefer kde 3.5 or 4.x ?
Deanjo: There are just too many apps missing for KDE 4 that I use
Kano: well you could run those apps within kde 3.5 too, thats only a matter of installed libs
michaellarabel: Kano: GNOME :) Though if I had to choose, probably 4.x.
Kano: gnome does not like mc
Deanjo: Ya well if I have to run a hybrid though Kano I might as well just run a tried and proven desktop
Kano: but the gnome founder wrote mc...
Kano: the use of f10 for menu just like inside win ist really stupid, nobody uses that hotkey, alt basically is enough
Deanjo: I might revisit KDE 4 with 4.3 if apps like Amarok, kaffiene, k3b, etc are ported and are as feature rich as their KDE 3 counterparts
Kano: i dont think that kaffeine will switch to kde 4 so soon
Kano: the next kaffeine will support dvb-s2, as the svn version does but it does not require new qt libs
Deanjo: Well I would say apps like k3b are more of a priority then getting kaff completely ported over.
Deanjo: One of the few apps that actually handled the transition gracefully and was ready was ktorrent
Deanjo: But George started on the port almost immediately after the kde4 release plan was announced
Kano: i can wait till squeeze release, so 1-2 years from now ;)
Kano: maybe kde 4.5.0 then *g*
redeeman: Deanjo: i suppose theres also no reason to bitch about nvidias drivers which run KDE 4 slower than vesa driver in KVM ?
redeeman: worse yet, displaying the vesa-kvm via vnc
Kano: some kanoti users had those xid errors recently too, but with lowend boards
Kano: from the thread you mentioned,but the rest is happy
redeeman: yes it's not everyone that's affected
redeeman: but some are
redeeman: and nvidia does nothing
Kano: i guess less than 5%
redeeman: they cant even bother actually answering
redeeman: those 5% paid 100% as much as the remaining
Kano: but it is impossible to fix all errors for everybody. like kernel 2.6.28 introduced some keyboard/mouse issues. no fix yet
Kano: some switched the keyboard because of it
Kano: thats also very rare
redeeman: ill bet you far more users than known, has the issue
redeeman: kde4 is just expert at triggering it
redeeman: i only managed to half trigger it 2 times in like.. ~3 months with kde 3.5
redeeman: i can do it in 15 min with kde4
Kano: well then dont use kde4 ;)
redeeman: wtf is that kind of answer?
Kano: kanotix stays with kde 3.5 too
Kano: much less problems
redeeman: so what?
redeeman: nvidia claims to not be able to reproduce
redeeman: they are flat out lying
redeeman: or blatantly incompetent
Kano: maybe tell em a board which has that problem + exact card model
redeeman: that's not really required, lots of other people have stated that
redeeman: and it affects a broad range
Deanjo: Sure and a broad range ( and majority ) have no issues
redeeman: or the majority doesn't ever find out because they don't run the software that triggers it easily
Deanjo: There is not 1 product out there that does not have a "in this circumstance it fucks up"
redeeman: i upgraded a few machines to kde 4.2 to test out after i had the problem on one box
redeeman: ALL had the issue
Kano: all gentoo?
redeeman: as a matter of fact only one gentoo
redeeman: also tested is ubuntu, debian, opensuse
Deanjo: Oh no, gentoo never fscks up Kano :P
redeeman: Deanjo: i did suspect gentoo initially
redeeman: because of my somewhat old xserver 1.3
redeeman: as you will also notice on the nvnews forum, it affects all distributions
redeeman: oh btw, i also tested arch linux
Deanjo: Well redeeman I've got all the opensuse mailing lists on tap and the nv issues are far and few
redeeman: it's entirely possible many don't know, or don't bother reporting
redeeman: either way
Deanjo: Well considering there are 4000+ messages that come in a day on the lists and like I say rarely an issue isolated to nv
redeeman: you may think your stuff runs perfect without issues, but just don't get too comfortable - you might find yourself with a whole slew of issues once you too decide to switch to slightly different software
Deanjo: Runs much the same with bluewhite
Kano: redeeman: does fglrx run better with kde 4 for you?
Deanjo: and I do find issues from time to time, I throw them at Roland@nv and they are usually fixed in the next driver release
redeeman: i haven't even bothered trying it
redeeman: and i don't doubt fglrx is broken beyond belief
redeeman: Deanjo: well do me a favor then, throw some xid's in his face and have THEM fixed
redeeman: it took nvidia more than 1 year to fix an edid misparsing i had
redeeman: i even mailed them sample C code to parse my edid blob
redeeman: i must backup my workstation
redeeman: fucking seagate disks are of the affected firmware
redeeman: and i don't trust their updater one bit
redeeman: Deanjo: how did the upgrade go for you?
redeeman: hmm i probably should still do the backup
Deanjo: flashed the 500's on a system with running in ide mode and then slapped them back into the raid.
Deanjo: No issues
redeeman: i have 2x500gb in raid1 aswell
Deanjo: Some say you can do them still in raid just switch the controller to ide but I didn't want to take the risk
redeeman: well i don't see why that shouldn't be possible, obviously it will work, or data is lost
Deanjo: Didn't really feel like restoring a TB+ of data
Deanjo: Ya but you know as well I do about "theory" vs "real practice"
redeeman: well theory or not, if it doesn't work, it destroys data
redeeman: and then it wouldn't work the other way either
redeeman: atleast i see no way it could
Deanjo: I preferred to stay on the safe side
redeeman: unless theres some bug doing two at a time
redeeman: which ofcourse given it being seagate is not unimaginable
Deanjo: Killed a new WD 1 TB at work today too
Deanjo: It at least lasted 4 days
redeeman: which model?
Deanjo: The Caviar Black model
redeeman: the wd10eads looks nice
Deanjo: Got some Hitachi's coming in, hopefully they can take the abuse
redeeman: what abuse?
Deanjo: Abuse being a sustained full write
Deanjo: Meanwhile the old Maxtors keep ticking along with heavy loads without breaking a sweat
redeeman: there are 1tb "maxtors"
Deanjo: No the old 500's and 250's
Deanjo: Those things are like tanks. They just keep going and going
redeeman: damn, linux gpio sysfs sucks
redeeman: atleast in speed