Milyardo: michaellarabel: Do you have a repository for the sources to phoronix global somewhere?
evocallaghan: He is a sleep I believe
Zool^: Hello! If i want to do a write/read test on my server, is dd if=/dev/zero of=foobar count=1024 bs=1024 a good way to do it?
Zool^: I feel when i run it twice/several times it runs faster and faster. Does it cache?
nasedo: Zool^, why not do hdparm -t ?
nasedo: but thats just read i guess
Deanjo: damn spam bots in the forum
michaellarabel: Zool^ phoronix-test-suite.com is a good way to do it :)
Deanjo: Shameless plug :P
Kano: hi
Kano: today i really got a new ati card, 3450 to test the latest drivers *g*
Kano: 9-2 works with gl2benchmark, wow 16 month of waiting for that driver
Kano: last working with gl2benchmark was 8.40.4, last with google earth + gl2benchmark 8.39.4
Kano: basically 18 month broken drivers when you wanted both
maligor: 9.1 didn't?
maligor: and apparently people are raving about the 9.4rc1 on phoronix forums, but who knows how accurate they are
Kano: dont know yet
Kano: got a new system with that card, tested only in live mode
maligor: saying it fixed compiz and wine somehow
Kano: i dont play games with wine, as the games i want to play do not work
maligor: I wonder if they've fixed opengl 3.0 also
Kano: i use it for dvd burning like imgburn works great with wine
maligor: I guess they mean resizing slowness in compiz
maligor: err... there are linux burning sw?
Kano: i dont use compiz
Kano: imgburn beats em all
Kano: if you primary use iso images
Kano: say you want to burn with verify
maligor: I wouldn't be surprised if it used cdrecord/wodim and cdrdao
Kano: then the standard tools do a md5sum test on the iso first
Kano: this one does its checksum check while burning, no extra time spent for doing this step
Kano: it definitely does not use cdrecord
Kano: http://imgburn.com/
Kano: it even mentions wine to be compatible
Kano: ImgBurn supports all the Windows OS's - Windows 95, 98, Me, NT4, 2000, XP, 2003, Vista, 2008 and 7 (including all the 64-bit versions). If you use Wine, it should also run on Linux and other x86-based Unixes.
maligor: as opposed to having written in a portable fashion
Kano: one wine version did not allow running it, therefore i tracked down the patch when this happend
Kano: and then it was fixed again
superdump: hello
maligor: err.. you can't do checksum checking while burning except on the buffer
Kano: i do not need many tools,but i want this one
superdump: i know you guys are all over i7 and such at the moment, i was just wondering if you knew what gcc options (in terms of march/mtune and math options) are recommended for i7
Kano: maligor: that tool even allows you to verify already burnt iso images against dvd+r(w), which is not possible using md5sum
superdump: i'm sure it depends on the gcc version one is using, but say current 4.3.3
maligor: no, we're all amd fanboys
Kano: you got that cpu?
Kano: i want it too *g*
superdump: i haven't yet, but i'm going to shortly :)
superdump: http://gcc.gnu.org/onlinedocs/gcc-4.3.3/gcc/i386-and-x86_002d64-Options.html#i386-and-x86_002d64-Options
Kano: well before i need a new board
superdump: that seems to suggest that the gcc cpu-type with the most available relevant instruction sets is the barcelona
Kano: for a Q9300...
maligor: written with Delphi that ImgBurn
superdump: but that also has 3dnow stuff
Kano: maligor: do you like it
superdump: elsewhere i see: -march=core2 -msse4
maligor: oh, I have no intention of actually starting it :P
maligor: it was upx compressed for some reason too
superdump: i don't know if you guys are interested in icc versus gcc
superdump: http://multimedia.cx/eggs/icc-vs-gcc-smackdown-round-3/
superdump: not completely settled as fair yet
maligor: on x86_64 specifying options won't change much
maligor: the generic target enables most of the stuff that affects performance most
superdump: that was a somewhat synthetic test though as we have a fair amount of asm in ffmpeg
maligor: Kano, err.. what's the point of doing checksum check during burn anyway?
superdump: people over there don't trust C compilers with the job of SIMD
maligor: Kano, shouldn't you verify the image before burning it? :P
maligor: unless you specifically want coasters
Kano: maligor: i do so after dl
Kano: but do not need that time with each burn
bkero: leaves a comment on gcc vs icc article about how acovea already sort of does this.
Ivanovic: WOW: http://www.holarse-linuxgaming.de/news/ut3_und_es_lebt_doch
Ivanovic: something like "UT3 port still alive"
Ivanovic: it is referencing the UT3 ml where ryan said that the linux and osx port is still being worked on with optimizations happening and annoying bugs still being fixed
Ivanovic: no release date ready yet but tyan wrote that the the port will be released as soon as possible
maligor: In other news: DNF is going to be released next week!
Ivanovic: jupp
Kano: maligor: haha
redeeman: ffs, only ONE nic left
redeeman: must order more soon
Kano: you need more than onboard nic?
bkero: KillerNIC
redeeman: Kano: yes, for routers
maligor: buy mb's with two nics onboard :P
redeeman: what makes you think i dont?
maligor: oh, routint to more than 1 net
maligor: or want to handle switching on system?
redeeman: many networks
Xemanth: redeeman: what kind of harddrives do you have if your network is the slowest part ?
redeeman: what makes you think my network is the slowest part?
redeeman: anyway, a simple raid is much faster than gigabit
Kano: redeeman: now my quad core runs
Kano: CPU[Intel Core2 Quad Q9300 @ clocked at 3000.000 Mhz] Kernel[Linux 2.6.28-8-generic i686] Up[-13min-] Mem[-204.2/2012.4MB-] HDD[-760GB(70%used)-] Procs[-173-] Client[Konversation 1.0.1]
redeeman: when did you get quad?
Kano: today
Kano: replaced my E6600 by this one
Kano: wanted a new board for it first,but then i thought, i should exchange it first to test if the cpu is ok
Kano: i need faster ram
Kano: 3 ghz is max with 400 fsb
redeeman: doesn't matter
Kano: well i want 4 ghz ;)
Kano: whats the max with your 45nm cpu?
redeeman: don't know, didn't try
Kano: i got a 2nd system with E8400 too,but thats a board for business pcs, no oc possible
Kano: therefore i put it in my standard board with oc features
mattmatteh: wonders how many watts kano's computer is
mattmatteh: s/computer/oven/ ?
mattmatteh: s/computer/space heater/ ?
Ivanovic: Sysinfo for 'rechner1': Linux 2.6.25.20 running KDE 3.5.10, CPU: Intel(R) Core 2 Quad CPU Q9300 @ 2.50GHz at 2000 MHz (4999 bogomips), HD: 343/1113GB, RAM: 2686/3952MB, 153 proc's, 14.2h up
Ivanovic: mine is fast enough for me without overclocking...
mattmatteh: goes for efficiency
mattmatteh: has never over clocked... doesnt see the point, thats amd/intel's job
raevol: agreed
mattmatteh: it should also automatically clock to what is needed
mattmatteh: what would really be cool is if i could tell it to go as fast as it can with a need for a fan or set a max fan speed. most computing doesnt need the fan on, and neither does idle
mattmatteh: lazy monopolies
raevol: hehe
bkero: I'm doing a 340MHz FSB
bkero: With a 90nm cpu
bkero: That's the limit of my PWMs :/ Past this, and I'm going to need a water rail going over my FETs.
bkero: is serious about his cooling, and has considered looking at chopping and tapping fuel rails for it.
redeeman: has watercooling
bkero: My watercooling setup used to involve a bucket, a 255LPH pond pump, and an MR2 radiator complete with dual SPAL fans.
mattmatteh: thinks about removing the fan all together for a quiet system
mattmatteh: roles his eyes at water cooling.... waste of energy
bkero: mattmatteh: Not necessarily, look at the G5's use of water cooling.
mattmatteh: bkero, but the g5 are not efficient ?
mattmatteh: and still hot ?
bkero: The g5's aren't efficient, but the cooling system is.
mattmatteh: i considered one of those
mattmatteh: bkero, yeah, thats what i was getting at. the cooling might be, but not the cpu
mattmatteh: i would never want a cpu like that
bkero: They're great CPUs
mattmatteh: i have a 45 watt now, and think that is way tooo high
raevol: http://images.google.com/images?hl=en&q=g5%20water%20cooled&um=1&ie=UTF-8&sa=N&tab=wi
bkero: Yea, it is. Mine is 2 watts.
bkero: Yours is way tooo high
raevol: leaks :o
mattmatteh: bkero, i want one, but not unless its efficient.
bkero: As soon as you get yours down to below 10 peak, let me know. :P
bkero: Then you can tell me about efficiency.
mattmatteh: bkero, 2 watts for the cpu ?
mattmatteh: a g5 ?
bkero: No, 2 watts for the entire board.
bkero: Not a G5.
mattmatteh: bkero, sorry, i missed something. i thought you were talking about a g5 cpu
bkero: Also
bkero: http://arstechnica.com/old/content/2002/10/ppc970.ars
bkero: G5 at 1.8GHz is 42 watts, below your current.
bkero: and that's at a 130nm process
mattmatteh: bkero, ahh
mattmatteh: nice
mattmatteh: i did consider the g5
mattmatteh: but....
mattmatteh: its getting old
mattmatteh: there was some power server
mattmatteh: that was on sale
mattmatteh: that used a 2 year old cpu
mattmatteh: some monstrous power supply
mattmatteh: i didnt look at the details
mattmatteh: just didnt seem like a good idea
mattmatteh: so i opted for amd instead
bkero: So go up to IBM and ask for some RoadRunner nodes
bkero: AMD is pretty terrible at efficiency too. The Athlon 64 series never got better for per-clock efficiency, the Phenom negligably so.
mattmatteh: bkero, the 4850e isnt bad
mattmatteh: i am sure not the best
mattmatteh: but good cpu's are hard to come by
mattmatteh: i had high hopes for g5 and powerpc
bkero: x86 is inherently inefficient :(
redeeman: ssh's into his 0.65W computer
mattmatteh: now i am waiting on arm
mattmatteh: bkero, agreed !
bkero: mattmatteh: ARM is already here.
mattmatteh: bkero, but the monopoly
bkero: Remember that 2W I was talking about? http://www.beagleboard.org/
mattmatteh: bkero, for a desktop ?
bkero: Yes
mattmatteh: bkero, where ?
mattmatteh: link
mattmatteh: bkero, beagal board is not an end user device
mattmatteh: its a dev board
redeeman: bkero: i read there was an issue with the current beagleboard
redeeman: anyway
bkero: No real issue
redeeman: i wish it was possible to actually use the beagleboard
bkero: What's wrong with the current one?
redeeman: but theres no proper way to use the dsp or graphics core
redeeman: bkero: don't remember, i was reading the wiki, and there was some issue that was putting me off
bkero: The secondary USB port didn't work on the Rev A boards, so they took it out in the Rev B boards, but you can still use the miniUSB port
redeeman: anyway, theres still the matter of the dsp and graphics core
bkero: DSP is open, the graphics core is the same poulsbo shit
redeeman: the dsp requires a nonfree compiler?
redeeman: does it not?
bkero: The graphics apparently works for basic framebuffer drawing, but no 2d/3d accel without a blob
mattmatteh: bkero, yeah, i thought about the beagal board, but wasnt impressed
bkero: DSPs are just standard realtek shit
redeeman: you mean TI, right?
bkero: Sorry, been working with a lot of DSPs in terms of audio. yea, it's a TI OMAP3
bkero: The only thing left a blob is the graphics
redeeman: bkero: the only way to currently use it, as far as i know, is by using some crappy nonfree ti compiler to generate code for the dsp
redeeman: but the graphics is far worse
redeeman: it's essentially USELESS as it is
bkero: redeeman: Nope, you can build the entire thing crossdev with an openembedded toolchain
bkero: I have one, it's sitting around 20gb
redeeman: hmm okay
redeeman: well that's good then
redeeman: but still, useless graphics
bkero: So are all poulsbo chipsets
redeeman: yes
redeeman: but we are talking beagleboard with OMAP3 now
bkero: Same graphics
redeeman: yes, and same uselessness
bkero: I'm using the blob right now
bkero: and am prodding the moblin people to un-deprecate the xf86-video-psb tree
mattmatteh: bkero, ok, besides the beagal board , any other arm devices that could be used for desktop use ?
bkero: Not really, unless you're blind.
redeeman: beagleboard also desperately needs ethernet
mattmatteh: redeeman, yes it does !
bkero: That would make the board bigger :(
Deanjo: And PCE-e X16 :P
mattmatteh: but useful
redeeman: i mean ffs, nobody that respects themselves uses usb ethernet in their desktop
redeeman: bkero: not nessecarily
bkero: Deanjo: That would mean it needs a PCI-e bus
mattmatteh: agrees with redeeman and Deanjo
redeeman: that fugly expansion connector could go
bkero: redeeman: Or a complete redesign of the board. Have you looked at the PCB traces on it already?
mattmatteh: again, its a dev board, not for end user
redeeman: nope, i didn't look at closeup
redeeman: mattmatteh: others dev boards have ethernet
redeeman: ethernet ftw!
redeeman: ffs, everything should be ethernet
redeeman: death to usb
bkero: Other dev boards didn't have the goal of a 3"x3" board size.
redeeman: also, using inch size is ugly
redeeman: must convert to metric round number
Deanjo: Screw that, gimme back my 16550 uarts, anything more is overkill :P
redeeman: yes, i also want an uart
redeeman: and some gpio's
redeeman: have come to like gpio's
bkero: It has GPIOs
redeeman: how many?
redeeman: and do they support interrupting?
bkero: Seems so
mattmatteh: steps away for a while
bkero: 26 pins
redeeman: http://dkc1.digikey.com/us/mkt/beagleboard.html
redeeman: "Programming: free, non-commercial TI compiler for C64x+"
redeeman: but you say we have an actually free way to compile for it?
bkero: Yea, the one right above it. Programming: gcc for ARM Cortex-A8
redeeman: but that is only for the arm core?
bkero: Note: If you will use only OpenEmbedded (OE) to build code for your Beagle, you don't need to download the CodeSourcery compiler. OE builds a cross-compiler from source as part of the bitbake process.
bkero: http://elinux.org/BeagleBoard is a pretty extensive article
redeeman: ill take your word for it then
redeeman: but the graphics is still ruining it
redeeman: plus the lack of a vital peripheral(ethernet)
redeeman: i mean seriously
redeeman: if i saw someone sitting on a small computer
redeeman: with an usb hub sticking out
redeeman: with an usb ethernet adapter
redeeman: i'd point and laugh
redeeman: HARD
redeeman: i do not wish to be in that position
bkero: I use an SDIO wireless card
redeeman: but i want to boot from sd :P
bkero: Erm
bkero: There's another SD slot on the expansion header
redeeman: so then i'd need a breakout board
redeeman: lol
redeeman: also i like wired ethernet :P
redeeman: i COULD get an spi-ethernet chip
redeeman: but that would be quite ugly
bkero: There's no way to hook up a keyboard and mouse without a USB hub, so if you're using it as a desktop, you have one anyways.
redeeman: you are forgetting the possibility of me having ethernet interfaced input devices ;D
bkero: You're also forgetting the possibility of me taking a bone saw to your skull. :P
redeeman: that would be pretty difficult
bkero: an E-bonesaw
redeeman: if only beagleboard had graphics that worked
redeeman: then i'd other one
redeeman: despite the extremely insane usd-eur conversion rate
bkero: You could write drivers for it. :P
redeeman: do you have documentation for the device for me?
bkero: There are old-ass drivers that don't compile against anything newer than 2.6.24, and have half of the drm stack hardcoded in them
bkero: As much as I could get, yes
redeeman: well.. is it complete enough to write a proper driver?
bkero: Wouldn't know, never written a proper driver myself
bkero: Well, a video card driver at least
redeeman: me neither, but i might just pay others to do it :D
Kano: michaellarabel: did you have time to look at the pts error?
Kano: that hanging on end of benchmark
michaellarabel: Kano: with smokin-guns? The only test system I had on today was an R500 with Mesa and there the game doesn't run. So no, not yet.
Kano: i had the same issue with fgl_glxgears
Kano: just boot my live image
Kano: install fglrx live and test
Kano: i would like to do some benchmarks of my new systems *g*
redeeman: bkero: http://elinux.org/BeagleBoard/DSP_Howto
cxo: http://www.tomshardware.com/news/Via-VX855-chipset,7261.html
evocallaghan: michaellarabel: Hey, asleep yet ?