Phoronix IRC Log: 2008-08-31
Deanjo: Morning/Afternoon/Evening all
Deanjo: I see your already punching code this morning
michaellarabel: Yep, trying to get it finalized.
Deanjo: hehe, just saw that ioquake3 for windows is being built on openSUSE's build service
Ivanovic: hi Deanjo
Ivanovic: Deanjo: you tried to test performance of a "usb stick raid", right?
Ivanovic: Deanjo: with a normal 2.5" usb 2.0 hd i am able to get 32MB/s on average
Ivanovic: so this is where the cap seems to be over here
Ivanovic: and with the raid you should be able to at least get this much asuming that the sticks are fast enough
Ivanovic: if you don't reach it the overhead per stick is either too large, or the sticks are too slow (that is you need more ;) )
Deanjo: Well the sticks I'm using are one of the faster sticks http://www.ocztechnology.com/products/flash_drives/ocz_atv_turbo_usb_2_0_flash_drive
Deanjo: Wrong link http://www.ocztechnology.com/products/flash_drives/ocz_atv_usb_2_0_flash-drive
Ivanovic: still you should at least get 30MB/s if things are working nicely
Deanjo: Ya stripe size makes a difference on it quite a bit I noticed
Deanjo: smaller stripe sizes seemed to work better, at least for these keychains
Deanjo: I got them up to ~30MB/s
Ivanovic: and somewhere in this area you get to the usb cap
Ivanovic: that is: when you really are on different channels more should be possible
Deanjo: That is on separate channels
Ivanovic: then either it does not really help, or the sticks are too slow for more
Ivanovic: are you reading from them or writing to them?
Deanjo: That's reading
Ivanovic: my guess is that the sticks are limiting there
Deanjo: Well, not sure about that, can get damn near double with windows
Deanjo: 56 MB/s there
Ivanovic: hmm, interesting
Deanjo: (you just have to disable the flip bit so windows sees them as non removable drives)
Ivanovic: which file system?
Deanjo: and xfs on linux
Ivanovic: hmm, interesting
Ivanovic: maybe something with the usb subsystem driver that does cap the results this much?
Ivanovic: (just a plain guess)
Deanjo: Ya could be
Deanjo: CPU usage is negligible 5-7%
Deanjo: I'm thinking of trying later what it does with a older kernel (before they started revamping USB support in the latest kernels)
Deanjo: OS X also saw about the same speeds as windows
maligor: pats his 120MB/s disk
maligor: no, classic
redeeman: then the seektime is still abysmal
maligor: yeah, it is
maligor: hmm.. eSATA key drive?
Deanjo: heh, like you can get a coffee with that extra 7ms saved on seektime :P
redeeman: Deanjo: don't be so analog :P
redeeman: think about 100000000 of those 7ms wasted seeks
maligor: a 30MB/s drive does allow you to go get coffee tho
Deanjo: heh, OMG can I sue seagate for that time back redeeman :P
redeeman: if you move to the united states, you sure can
redeeman: and i encourage you to do it
redeeman: because if you win
redeeman: im gonna go there and sue you for using my idea :)
Deanjo: ha, more then likely "Judge my client lost damn near an hour of his life waiting for his drive to position the heads...."
maligor: usb stick raid huh
Deanjo: "We want 78.4 million in damages for loss of productivity"
redeeman: "I _WANTED_ to go buy the winning lottery ticket in that half hour, but now i didnt, and thus didnt win"
maligor: but now what I want is a eSATA CompactFlash reader
Deanjo: Ya those bastards won't refund defective lottery tickets too :P
maligor: hmm.. maybe buy a cheap board from ebay and build a case
Deanjo: Now if this was only about 80 dollars cheaper http://www.newegg.com/Product/Product.aspx?Item=N82E16813153114
maligor: heh, the sink support looks ridiculous
Deanjo: Dual Gbit lan is nice though
maligor: lots of stuff on the underside too
maligor: I can just imagine stuffing a massive 800gram copper cooler on that
maligor: something bigger than the board
Deanjo: heh, I think the idea of the board though is NOT to slap a 9950 on it :P
maligor: cute board tho
Deanjo: Get a low power version of a X2 and you could passively cool it
maligor: no esata
maligor: and no firewire
maligor: what a poor board
Deanjo: lol, no SLi
Deanjo: err, actually, I shouldn't say that, it would support hybrid SLI
Deanjo: Lets slap a 280 video card on it :P
maligor: ofcourse I just spend similar amount of monetary units on a nas device
Deanjo: If they were cheaper it would be a great "lets try crazy shit" board
maligor: maybe the 8200+nano
Deanjo: Probably be able to fit that in the old G4 cube I have sitting in the corner
Deanjo: Or at least in my old Colbalt Cube
Deanjo: hmmm, looks like acetone ISO is working on burning capabilities now http://www.acetoneiso.netsons.org/index.php
Deanjo: debates switching to the alpha 2 version of openSUSE.
danboid: I've got a HD2400 with DVI out on my laptopand I'm using the fglrx (prop. ATI one) driver under Hardy. I have a TV with HDMI input so could I get a DVI to HDMI cable and expect it to work or is DVI out not supported under Linux?
cxo: try #ati for fglrx driver, and #radeon for open sauce driver
danboid: Thanks cxo!
niklauz: does this mean that dri2 *is* expected with the next major release?
rbrett: I doubt it'll make 7.4/1.5 but it will make 7.5/1.6
niklauz: ah, ok i think i had my timeline mixed up
rbrett: it was originally scheduled for 7.4/1.5 but had to be reworked to use gem instead of ttm.
rbrett: but now krh has dri2 running on gem it should make 7.5/1.6
Dicrodiu1: are there any Intel X3100 tests on the Phoronix site? I didn't have luck searching :(