Phoronix IRC Log: 2008-06-16
starkmjo`: lo all
starkmjolk: (just trying to wake up the channel, have nothing interesting to say myself atm ;) )
d2kx: GeForce GTX 280 launch is in 4.5 hours... but I don't want to wait so long to have something to laugh about :(
maligor: what's there to laugh about gtx 280?
maligor: aside from maybe that it has two power connectors
maligor: and uses 236W?
maligor: that'd be twice as much as a quadcore
starkmjolk: just after I said hi I succeeded in killing my emacs :(
d2kx: maligor, the pre-reviews from yesterday have shown that its performance is not as good as expected
d2kx: it's 599€, while the HD 4870 for 349€ will nearly as fast
Exopaladin: no big loss then, starkmjolk ;P
starkmjolk: Exopaladin: irc is in my emacs!
Exopaladin: silly emacsOS
maligor: d2kx, heh
maligor: a flop then
maligor: fantastic news for ati I guess
starkmjolk: not silly at all, the only silly part is all that linux below :P
Kano: hi, did somebody test fglrx with intrepid kernel? or older nvidia?
redeeman: does anyone know what happens if i put in both a radeonhd and nvidia pci express card?
redeeman: will i be able to just use 1 at a time?
maligor: for 3d acceleration, yes
Exopaladin: I wonder if you could run one 3d accelerated X server for each card
copper: yes you can
copper: What would be the point though?
Exopaladin: I have absolutely no idea, I was just curious
copper: nvidia driver 173.14.09 for Linux x86/x86-64 released
maligor: probably support for their new card?
copper: erm, just click on the link already
dante_2core: maligor, don't think so, they're not so fast
maligor: unlike amd, eh? ;)
maligor: the fglrx driver lists 4800
redeeman: where does it list it?
maligor: FGL_ASIC_ID(0x9440), // RV770 (RV770 9440)
maligor: 4 different RV770 models there
m-c: I decided to use the open "radeonhd" driver. Works pretty snappy in 2D. No Compiz, but that's okay with me, as long as my dual monitors work. Glad to be using open drivers that are at least functional! :) :) :)
copper: ah crap, now that I'm running a distro, I have to wait for update packages to show up :(
maligor: you can make them too
vadi2: did you enable the pre-release updates? that can get them faster
maligor: weird, I wonder if my memory's borked
copper: the announcement was just made, I doubt someone updated packages this quick
maligor: need your binary driver fix, eh
copper: This is what I'm waiting for, mostly: http://www.nvnews.net/vbulletin/showpost.php?p=1682593&postcount=2
m-c: I followed these instructions and some Ubuntu instructions for adding a Virtual display. http://www.phoronix.com/forums/showthread.php?s=e6059bf57a04b2327f00af4663511407&t=9951
m-c: I think I can say that I am going to be off binary-blob drivers from here on out. One more piece of the open source puzzle falling into place.
vadi2: that's pretty cool.
copper: they're not talking to you, m-c - put me off your /ignore list already
copper: some dialogs might make more sense once you do that
m-c: Seems like WiFi is the last, great frontier now, and that's mostly a legal thing.
copper: (I'm really talking to myself here, since he can't read me)
vadi2: m-c: copper wants to be friends again
copper: I don't even know why I'm on his shit-list
m-c: AMD is now REALLY on my Buy List. I liked them before, but now with the open 3D drivers and their firm stance against Treacherous Computing, they are the only thing I will buy from here on out.
copper: m-c: can you run the gtkperf PTS benchmark ("Total Time" option) and post your results?
m-c: copper: Let me check on that - I got an error running glxgears a minute ago, not sure what that's about yet.
copper: "Release Date: June 11, 2008" uh, they're only announcing it now?
m-c: copper: Compiling the gtkperf source did not work. The configure script exited silently when it was checking gtk+-2.0 versions.
copper: try running ./configure on your own, on a separate copy of the source code, see what it says
m-c: That's what I am doing - at least, I downloaded the source code and I ran ./configure
copper: and what does it say?
m-c: appending configuration tag "F77" to libtool
m-c: checking for pkg-config... /usr/bin/pkg-config
m-c: checking for gtk+-2.0 gdk-2.0 ... mmm@salamander:~/newfiles/video/gtkperf$
copper: do you have development packages? GCC, gtk-devel etc?
m-c: installs libgtk2.0-dev .
m-c: okay - built
m-c: Okay, I clicked start with the defaults. It did "stuff". What do you want to know?
m-c: Oh, the Total time was 7.61
copper: either run "gtkperf -a -c1000", or "phoronix-test-suite run gtkperf" and select "Total Time"
copper: the former takes less time than the latter, FYI
m-c: It's cranking away.
m-c: This is very taxing on the CPU
copper: all the more taxing whenever some graphics function isn't accelerated by the GPU/driver
m-c: Hmm - I have a total time of 116.15 , but I also see many Gdk-CRITICAL errors as output: gdk_draw_pixbuf: assertion `GDK_IS_PIXBUF (pixbuf)' failed
vadi2: the last release was 4 years ago, so things might've changed
copper: I'm waiting for d2kx, I'd like him to explain his extravagant gtkperf results
m-c: I did see the Pixbufs test run, and it did not crash, so I do not think it was all that critical of an error.
copper: nah you're fine
copper: it's still a mediocre result, I was hoping it would fare better than nvidia drivers
m-c: Anyway, I am going into this knowing there are limitations and there are faster / better driver solutions out there, but I am primarily concerned about the open drivers. Plus, they are just going to keep improving. I am frankly surprised they are as useful as they are, but, as you said, maybe a lot of it is due to unloading the work to the CPU.
copper: "Mozilla Releases Alpha Version Of Firefox 3.1"
Deanjo: Hmmmm, FireStream 9250 news but nothing on the Tesla C1060 ........ CONSPIRACY!!!! :P
starkmjolk: can the c1060 deliver a teraflops? :)
starkmjolk: oh, it could :) must be a conspiracy
Deanjo: Ya conspiracies run rampant @ Phoronix :D
copper: it's the Illuminati, I'm telling you
Deanjo: I heard that Micheal collects cheques from MS and is really a double agent
vadi2: ruh roh
copper: says the fudzilla double agent
Deanjo: no more kickback cheques for you
Deanjo: It's not like I revealed you being the editor for The Inquirer
maligor: yay for ati!
maligor: or amd
vadi2: bridgeman came in?
copper: where's d2kx damn it
maligor: what's the C1060 based on?
Deanjo: The GT200
copper: Deanjo: so, nobody answers me in #xine
maligor: ah, yes, the power sucktion device :P
copper: I want my money back
Deanjo: No refunds
Deanjo: Or they will charge ya a 20% restocking fee
maligor: would be nice if amd and nvidia would standardize on one language tho
Deanjo: Such as OpenCL?
maligor: I mean for GPGPU
copper: they both claim their standard is open
Deanjo: That is what OpenCL is for
maligor: or stream processing
maligor: copper, well, AMDs is
maligor: copper, their sdk comes with sources
copper: doesn't CUDA's, too?
maligor: mm.. never looked
copper: way to bash nvidia, dude
maligor: I think it required you to link against libCg
maligor: but it was a while back
vadi2: since I'm not much into the specifics, this standards proposal by apple looks like the c# proposal by microsoft
vadi2: i hope i'm wrong.
Deanjo: I just want to be able to render 1080p realtime dammit
copper: lots of C# apps exist on linux, fwiw
copper: yeah, me too
vadi2: don't have anything against the programming language itself, tomboy is awesome. but it sure does cause a lot of time wasting debates and suspicions
copper: what's silly, is that you can't buy cheap but fully supported hardware, vs. expensive but partially unsupported hardware
copper: and video app developers don't help with their conservatism, either
copper: "multicore is useless" <---- mplayer dev
copper: somebody buy the hungarian dudes some multi-core rigs
maligor: copper, I don't see much sources in cuda
maligor: unless cuda is made from 2400 lines of code
copper: it's THAT powerfull :P
starkmjolk: copper: that citation made me sad :(
maligor: brook+ compiler is just 75k lines
maligor: I mean just the compiler
maligor: ofcourse maybe they just hid it somewhere
starkmjolk: where the sun don't shine?
Deanjo: Grrr, is it impossible to find a bloody freenode server that isn't laggy as hell?
Deanjo: OK I guess this one now is OK .1s
m-c: copper: I realized that direct rendering is turned off because of a missing file. That slowed down the speed test, I bet.
copper: try again
m-c: Cannot figure out how to install this library
m-c: It is not in the mesa package with the other DRI libraries
m-c: Ah, I see a Phoronix post on this topic - it is for software rendering
m-c: I am probably using some funny X settings, too, but I am still not comfortable with all this auto-configuration going on within Ubuntu, yet.
copper: upgrade to Arch
copper: starkmjolk: I was paraphrasing, but the actual mailing list post isn't hard to find
Deanjo: Hmmm I wonder how well Cuda works with the 8200 chipset?
copper: I haven't been able to try CUDA
copper: for one thing there's no CUDA package for Arch Linux
copper: also, why the hell does every distro need to have their own packaging system
starkmjolk: copper: I like mplayer, and concurrent programming is the future. Just hope they don't get lost in some sequencial thread :)
Deanjo: conspiracy copper conspiracy. That's why
copper: For that matter, why is there a need for so many distros
Deanjo: Same people have been holding back the flying car for decades as well
copper: I understand different distros have different goals, but I don't see why they can't just provide their own "groups" of packages for a common distro
Deanjo: There is no real need copper for different distro's other then the fact that there is a need to sooth various egomaniacs
copper: you could have the conservative group based on GCC 3x or whatever, the bleeding edge group with compiz and KDE4, etc…
vadi2: packagekit hopefully will please everyone and work.
vadi2: autopackage was really good but it was anti-commercial and nobody agreed :|
copper: frackin' egos
copper: Deanjo: no more BSG until 2009, pal :(
vadi2: there's mojosetup meanwhile for games, and that's pretty good. if everyone used it. quite a lot still use the loki installers
Deanjo: Ya I know shitty TV until 09 now
copper: no more The Wire either
copper: no more Rome
copper: I need my fix!
copper: either that or I get myself one of those "lives"… yuck.
Deanjo: Dammit that Star Wars TV series is taking too long to get on the air
copper: I tried to get a life before… I only met uninteresting jerks, greedy bitches and baby-craving chicks
copper: what's with women, money and babies, really
copper: it's like they WANT sleepless nights and loss of sex drive
Deanjo: Women are definitely not "Free as in beer" lol
copper: ok I'm done with the misogynic rant
Deanjo: I'm really liking this Opera 9.5 64-bit
copper: I'm just too lazy to import my bookmarks and find an alternative to adblock
vadi2: "Larry Dignan: The Windows XP era ends June 30 and soon hardware vendors will be shipping you all Vista all the time in most cases. The save XP effort failed. And now it's time for Vista to sink or swim. "
vadi2: Right beside it
vadi2: "Reduce stress and speed up resolutions with these command references right at your fingertips. TechRepublic editors have updated and expanded our popular Linux Commands chart and made it downloadable for instant access. You'll receive a PDF file covering Linux, packed with the most common commands you'll need and use daily. "
copper: more noobs for Ubuntu
vadi2: Well, crap.
copper: lol: http://global.phoronix-test-suite.com/index.php?k=profile&u=albatros88-27664-3700-11608
copper: Do OpenSSL developers intentionally cripple Intel CPUs?
copper: Jade, what do you think?
vadi2: no, they cripple arch!
copper: vadi2: I have the high score, and I'm running Arch
copper: I'm apaige
copper: Intel fanboys will say Nehalem will fix that, with its dedicated crypto instructions
copper: that's when VIA fanboys come by and show their old C7's results
Deanjo: And then the GPGPU dudes come in and blow them all away :P
copper: This is disturbing btw: "Given that Khan's revelations were made in early 2004, does that mean it took the IAEA 1-2 years to brute-force the encryption?" http://it.slashdot.org/article.pl?sid=08/06/16/0440225
copper: odly enough the slashdotters didn't comment about the encryption being cracked by brute force in such a small amount of time
copper: Of course, I didn't RTFA.
Deanjo: Actually I was reading an article a few months back showing how they were able to bring down that time frame down to days using just conventional cpu's on a new algorithm
copper: So what's safe, really?
copper: How safe is AES-256?
copper: or maybe the terrorisss' used some weaker algo
Deanjo: Bottom line is, if you have something life threatening on your machine, keep it off networks
copper: Bruce Schneier says RSA-1024 (as far as asymetrical encryption goes) is already quite safe
starkmjolk: what terrorists? but AES is concidered safe by most, but since the US millitary encourages the use of AES I don't trust it :)
copper: Bruce recommends it, I think it's safe to use it
copper: or maybe Bruce is a jew too 0_o
copper: Schneier… that definitely sounds jewish.
starkmjolk: I don't really follow you (:
copper: inside joke, look up user "Jade" on the forums
Deanjo: And then add to your ignore user list :P
copper: geez, Arch users do post a lot
Deanjo: That's not the article I was thinking of but still interesting
copper: hmmm, since when do commercial websites use asymmetrical encryption?
copper: wait, doesn't it use both symmetrical and assymetrical encryption? I'm confused.
copper: vadi2: you're just hurting yourself http://global.phoronix-test-suite.com/index.php?k=profile&u=vadi-6246-31929-31046
copper: just my inner-AMD-fanboyism teasin' you
copper: just don't bring out the LAME benchmark results or I might start sobbing
copper: merged: http://global.phoronix-test-suite.com/index.php?k=profile&u=apaige-26371-19160-25236
copper: Monkey's Audio is so much slower on his system only because he doen't have yasm on his system (I need to make it a requirement or something)
m-c: Can the Phoronix Test Suite not be run without first installing it as root ?
copper: no need to install it as root
copper: just run ./install-sh $HOME or any directory you have access to
copper: you're cheating
copper: compare it to a similarly overclocked Intel CPU at least
copper: still, using my multi-process transcoding script on your beast should be nice
copper: I can already decode FLACs at 200MB/s, I wonder if you could max out your RAID array's bandwidth?
vadi2: Bleh. Getting chain letters from your mom is bad enough.
vadi2: But getting them from your 11 year old sis is even worse :(
copper: the latter is excused, the former is not
m-c: not liking this root access required thing
copper: quit cheating, that guy's setup is obviously underperforming
copper: m-c: root access is not required, did you not read what I said?
Deanjo: Can't find a audio test with a Q6600 @ 3ghz
m-c: yes, but I did not believe you after reading that it was ;)
copper: Deanjo: try the universe benches
copper: m-c: uh, just try it?
copper: I installed PTS in my home directory myself
copper: I don't touch my root-owned directories with anything other than offical Arch packages
m-c: Oh, I see - the error I was getting was not a permissions issue.
m-c: But, anyway, the documentation does say, "Root access is required." for the installation instructions.
m-c: pointing that out for what it's worth...
Deanjo: unfortunately kabage is the only one who has a Q6600 @ 3Gz and he has not ran the test's with final PTS
copper: I guess that needs a clarification
copper: Deanjo: well this one is already better: http://global.phoronix-test-suite.com/index.php?k=profile&u=root-18074-10297-4948
m-c: what's a good test (and systax ;) ) from the PTS for checking the CPU and memory?
Deanjo: Unfortunately we won't get a real true accurate proc vs proc until there is a livecd version of PTS
copper: What interests me more is test results from *optimized* systems with optimized benchmarks
m-c: oh, I see - it installs all the bits it needs. clever!
copper: just switching OS installations made all benchmarks run faster on the same hardware for me
m-c: how much (%) faster?
vadi2: Eh, what is this:
vadi2: Monkey Audio Encoding:
vadi2: 78MB WAV File To APE
vadi2: 26.13 Seconds 39.73 Seconds 26.16 Seconds
copper: not sure, a couple of seconds in several benchmarks
copper: (that was addressed to m-c)
copper: vadi2: what did you do during the second run?
Deanjo: Have you tried bluewhite yet copper ?
copper: what's that?
Deanjo: 64-bit only version of slackware
vadi2: copper: nothing important really.
vadi2: I'll try the test again..
m-c: sounds like <10% - I am excited about the PTS because it will give us general trends. +/- 10% is okay in my book
copper: Deanjo: dude, I just installed Arch Linux, unless it fails me big time in the near future, I won't switch to anything else
copper: vadi2: make sure you have yasm installed before you compile (install) the Monkey's Audio benchmark
Deanjo: Well if you ever are curious http://www.bluewhite64.com/news.php
copper: yet another victim of the NIH syndrom?
copper: "Bluewhite64 Linux General Info (Probably The Best Pure 64-bit GNU/Linux)" lol
vadi2: copper: how do I do that?
copper: well if it's the *best*…
copper: vadi2: type "which yasm", see if it finds it
vadi2: doesn't find anything
vadi2: i'll get it from apt-get
copper: Deanjo: they don't say what makes it the best
copper: vadi2: yeah and after it's installed, delete the encode-ape directory in ~/.phoronix-test-suite/installed-tests, and run "phoronix-test-suite install encode-ape" again
Deanjo: Blue & White = racing stripes :P
Deanjo: racing stripes = go faster :D
copper: I can't find anything polite to answer to that
Deanjo: All packages are compiled using 64-bit flags
copper: oh, and Arch x86_64 packages aren't?
Deanjo: Don't know
copper: well they are, duh
copper: that's the case with any pure x86_64 distro
copper: I'm running Arch Linux x86_64 (pure 64-bit system), and a chrooted i686 minimal install for Firefox 32 with flash and games
copper: works surprisingly well, especially thanks to dchroot, which allows for launching chrooted apps with a single command, no password asked
copper: About optimized benchmarks: maybe I should just start a thread on the forums and call for results from people who would compile PTS benchmarks with optimizations tailored to their system
vadi2: arch is like for humans +1
vadi2: ubuntu is for normal humans
vadi2: mint is for humans -1
copper: i.e. it would be sanctionned cheating in a particular context
vadi21: copper: wow, that yasm thing really helped. the test is now at solud 14s
copper: yeah Monkey's Audio benefits greatly from hand-coded ASM
copper: I figured yasm would be installed by default on 64-bit systems with build tools already installed, but I was wrong
Deanjo: Hmmm according to the Arch64 wiki the packages are only compiled with up to SSE2 optimizations
copper: Deanjo: I had to modify the PTS benchmark installation scripts so that it wouldn't link against my optimized libraries
copper: well you won't get more optimized builds than with Gentoo
Deanjo: Well LFS :P
copper: You can't ask package maintainers to release packages for each CPU
copper: yeah well I just come from *LFS, it's not really worth the trouble after all
vadi21: How can I delete / reinstall a test?
copper: i.e. my builds would be very optimized, but switching to GCC 4.3.x would require a lot of work, and I think some of my core libs were fscked up or something
copper: rm -r ~/.phoronix-test-suite/installed-tests/PROFILE && phoronix-test-suite install PROFILE
copper: So what I did, is that I installed Arch Linux x86_64, and compiled the few apps I wanted optimized myself, in my home directory
vadi21: Hm I don't know the individual names of tests for the 'audio-encoding' one
copper: e.g. audio encoding apps, pbzip2, etc…
copper: vadi21: just look in the ~/.phoronix-test-suite/installed-tests/ directory, it's the encoding-* profiles (directories)
vadi21: ah okay
copper: Deanjo: that way is much more straight-forward and painless
copper: because I don't care if man or vim were compiled with SSE9 instructions
copper: sorry, SSSSE 9.1n'
Deanjo: I would love to see a distro compiles with Pathscale optimizations
copper: aaaah yeah that would be really nice
copper: especially since a lot of benchmarks in the press are compiled with Intel's compiler, giving Intel CPUs a serious advantage
Deanjo: Gentoo w/pathscale
Deanjo: that would be sweet
copper: shut up, I'm feeling funny in my pants
Deanjo: even just a distro compiled with intels compilers would be interesting
copper: yeah definitely, especially since so many people use Ubuntu i686, which is hopelessly *un*optimized.
Deanjo: DAMN THE GPL!!! JUST DO IT!!!
vadi21: i686? it says i386 everywhere
Deanjo: Give me a 'balls to the walls' distro, screw licensing
copper: you want to rape the GPL?
copper: screw rape, let's gang-bang it
copper: (I know, I'm deep)
Deanjo: I really couldn't care, my need for making the most out of my system is greater then my need for a 'warm fuzzy feeling'
copper: vadi21: eh, you're right, even worse
copper: Deanjo: don't talk to GNU\colossalPenis
copper: sorry, that's GNU\colossus
copper: GNU\colossus: hey, how's your washing machine doing?
copper: Deanjo: actually, I wouldn't be surprised if the PathScale guys would be interested in setting up such a system
copper: one of them actually wrote to me asking if I needed support for my Cross-Compiled Linux From Scratch system
vadi21: everything went well except the last test:
vadi21: WavPack Audio Encoding:
vadi21: 78MB WAV File To WavPack
vadi21: 17.81 Seconds 17.83 Seconds 27.52 Seconds
copper: it would make for good publicity if they could show a Phenom rig with PathScale-compiled OS & apps
Deanjo: I can see something like that happening in the future copper with services such as the Suse Build service and Rpath
copper: vadi21: I don't know where those slowdowns come from. Is there something special running in the background?
vadi21: not that I know of, but I'll try the test again
copper: is HAL running? That's teh daemon from hell I would shoot down first.
vadi21: /usr/sbin/hald ?
copper: kill that pig
copper: it's evil
vadi21: it doesn't die
copper: it polls all your components every two nanoseconds
vadi21: i'll try powertop
vadi21: these tests provide an excellent opportunity to train the counting seconds skills
vadi21: *just got 18s twice in a row right on time*
copper: lol, "i686 optimized" http://koti.24.fi/sm4tik/shared/08-06-16-21:39_info.png
copper: I need to get off my ass and submit some needed patches for PTS
vadi21: heh, I feel sad for dell. their ideastorm is swamped with linux/foss ideas
copper: my profile on the forums had 171 visits??
starkmjolk: I'm off for the nite :)
copper: good night
copper: amazing, I can't find a single MPD client that I like
vadi21: tried gimmix, guimup, quimup
copper: I looked into all the clients listed on the MPD wiki
copper: their number is impressive, btw
copper: "lookie here, that client sucks, I'm gonna write a client that sucks MORE!"
Klavious: you tried sonata?
Klavious: that one and ncmpc are my favorite
copper: yeah, didn't like it either
vadi21: Wow. This is a pretty worthy list of linux games: http://www.linuxlinks.com/article/20080530054213402/CommercialGames.html
vadi21: this thread is kinda embarassing to reply to after seeing the screenshot: http://ubuntuforums.org/showthread.php?t=831463
m-c: I like this list ;) >> http://www.linuxgamingworld.com/games-catalog/
m-c: vadi21: yeah, not wanting to get my name associated with that thread... ;)
m-c: Maybe it is pictures of Adults.
m-c: "Anyone know what I'm missing?" Umm.. yes, discretion.
vadi21: It's not that bad actually, I had a worst case. In a public library, some guy asked me for help on some adult website
m-c: Were you like, "Dude, so uncool to be doing this in a library..."
vadi21: I should've, but I shyed away and just pointed out his error.
Deanjo: lol got a good heated discussion about SSL debacle and who's to blame.
Deanjo: Going on in the #suse room
Deanjo: I love it..... it's SLASHDOT LIVE!!!!
vadi21: suse? how does that affect them
vadi21: (or is not about the openssl fiasco)
Deanjo: It doesn't effect them, they are just having a heated debate about how some people should not be contributing blindly
Deanjo: and who does most of the key development , etc etc
Deanjo: Good ol' pissing match
vadi21: i see.
Deanjo: Typical, distro troll walks into another distro's room attempting to claim his distro's superiourity
Deanjo: can't understand why ubuntu users seem to think ubuntu was the first to use apparmor
vadi21: are you sure ubuntu? a ubuntu user wouldn't know what apparmour is.
Deanjo: ubuntu was the first distro to introduce apparmor, so how can that be?
Deanjo: Ubuntu is crap, they put ease ahead of security and everything else... provided it seems to work, they're ok with that.
Deanjo: they were the first to use gnome with policykit
Deanjo: Oh vey
vadi21: good marketing ftw :)
vadi21: bad marketing ftl.
Deanjo: Ya, some people will swallow what ever the sheep tell them