Phoronix IRC Log: 2008-08-26
michaellarabel: Hi Deanjo
Plouj: yay, no more errors when running 'show-results'
michaellarabel: Great Plouj
Erektium: michaellarabel, latest git now does identify my Motherboard, but not Chipset
michaellarabel: what does your Chipset show
Erektium: nothing but that
Erektium: "Chipset: AMD"
michaellarabel: what is the output of: lspci | grep "Host bridge"
michaellarabel: can you paste your entire lspci?
michaellarabel: and what is this chipset supposed to be?
Erektium: NVIDIA nForce® 570 Ultra™ MCP
Erektium: that's what asus site said
michaellarabel: Erektium: Still around?
michaellarabel: Erektium and Deanjo, when you have a moment please checkout git (http://phorogit.com/index.php?p=phoronix-test-suite.git&dl=targz) to see if the Chipset is now properly detected.
Erektium: yep still here
Erektium: let's see
Erektium: michaellarabel, that now doesn't identify motherboard ;S neither chipset
Erektium: so it broke motherboard identification for me
michaellarabel: What does the motherboard report?
Erektium: Motherboard: System Product Name, Chipset: AMD
Erektium: thats the same as with 1.20b3
michaellarabel: hmmm how did it regress...
michaellarabel: can you redowmload it again? I am trying to figure out what could have gone wrong
Erektium: redownload that git that you just posted?
Erektium: It doesn't work
michaellarabel: Doesn't make any sense why when I didn't even touch the motherboard code, but just the chipset
michaellarabel: I'll need to see if it regressed for Deanjo too
Plouj: michaellarabel: can't you just re-run your code on his lshal output?
michaellarabel: I run it through in my head and it works there.
Erektium: michaellarabel, it does work!
Erektium: hmm.. I forgot to but ./ in front of the phoronix-test-suite system-info
Erektium: sorry my bad ;(
Erektium: so the bottomline is that it now works
michaellarabel: Ah, so what exactly does it output
Erektium: Chipset: nVidia MCP55.. well thats not excatly the same.. as asus site says NVIDIA nForce® 570 Ultra™ MCP
Erektium: but it at least says that it's nvidia not amd
michaellarabel: Erektium: MCP55 is the 570 Ultra, that's what it's codename is.
michaellarabel: The hardware usually doesn't know the retail names always
vadi2: impressive actionscript 3.0 demo: http://www.adobe.com/jp/events/cs3_web_edition_tour/swfs/perform.swf
Erektium: well this is nice =) no I have system-info showing my motherboard and chipset \o/
Erektium: michaellarabel, thank you =)
Plouj: slaps Erektium around a bit
michaellarabel: vadi2: Nice
Plouj: it's an IRC tradition, don't worry too much about it
michaellarabel: vadi2, I hope Adobe makes that into a benchmark for PTS
Erektium: never heard of such a tradition
GNU\colossus: trust me, it exists ;)
Erektium: ok =)
starkmjolk: Xemanth: tested the new motherboard yet?'
michaellarabel: Anyone here with PTS 1.2.0b3 setup?
redeeman: mesa 7.1 released
Deanjo: hmmmm, so with 1 usb keychain I get 15.63 MB/sec, two in raid 0 it gets 23.65 MB/sec
Ivanovic: Deanjo: how much of this is limit of USB and how much of the flash memory on the stick?
Ivanovic: since i know that it is possible to get >1GB/min with usb2.0 (real amount of data)
Deanjo: That's what I'm wondering
Deanjo: Plus how much of an effect does stripe size play on it
Ivanovic: (experiences from the p2p system named "lecture hall")
Ivanovic: and there with those HDs the bottleneck is definitely usb
cxo: usb2.0 is 400mbits
Ivanovic: cxo: uhm, definition is 480MBit
Ivanovic: *but* in fact it is slower than 400MBit firewire
Ivanovic: cxo: in normal useage it is
cxo: firewaire has burst of 400mbit
cxo: usb cannot do that
cxo: but will reach equal throughput
Deanjo: firewire tends to have less overhead as well
Ivanovic: cxo: do not underestimate the overhead by usb
cxo: fw has a higher voltage bus spec, so the signal is more suited for applications reading raw data without error checking
Ivanovic: cxo: regarding "transmitting data per minuite" (400mbit) firewire is faster than usb 2.0 (high speed)
Ivanovic: that is transmitting data from external devices
cxo: thats debatable
Ivanovic: like harddrives in cases
Ivanovic: cxo: that is what all benchmarks i ever heard about have shown
Deanjo: Ya firewire 400 tends to lay a lickin on USB 2
Ivanovic: cxo: no, "good marketing" for usb2
cxo: well thats why our boys at intel are working on usb3
Ivanovic: the reason for no broader adaption of firewire are the high license fees for it
Ivanovic: cxo: it will take a *long* time until usb3 devices are widely available
cxo: thats what they said about optic fiber
Ivanovic: the first chips are expected for late 2009, more likely 2010
cxo: gigabit lan
cxo: 1Tb hard drives
cxo: even Flight
Ivanovic: cxo: in technics 4 years *are* a long time
Ivanovic: and it will take this long till usb3 is widely adopted
cxo: you live in year 2005?
cxo: how'd you go back in time?
Ivanovic: cxo: ???
Ivanovic: cxo: for usb3 the specs are not final so far
Ivanovic: so *any* chips on controllers are not available
cxo: the first chips are expected for late 2009, more likely 2010
cxo: what year is it in your world? right now?
Ivanovic: yeah, "the first chips" != wide usage
cxo: thats what they said about the phillips tuner
Ivanovic: wide usage requires usb3 on the installed base
cxo: and the AKM4441 DAC
Ivanovic: and *many* will get it when they buy new computers with new mainboards
cxo: everything is marketing baby
Ivanovic: hardware vendors will take some time to port things and so on
cxo: look at 802.11n
cxo: the draft spec
Deanjo: You also have to remember that no USB 2.0 device can use more then 2/3's the available bandwidth. So true maximum throughput is about 320Mbps (and that is with a very good usb bridge)
cxo: they had routers for that shit last year
Ivanovic: cxo: sure
Ivanovic: but having stuff available and broadly used is a huge difference
Ivanovic: cxo: how many people do have draft n networking at their homes?
cxo: what do i look like bestbuy or something?
Ivanovic: how many (public) access points eg at universitys do support it?
cxo: how the fuk should i know
Ivanovic: the number is really tiny since the installed base for the "old" b/g compatible net is high
cxo: great that you know such things
Ivanovic: cxo: have you found any access point with draft n in use?
cxo: Ivanovic, how many people have german shepards in their homes?
Ivanovic: that is beside seeing that they sell things
Deanjo: That will change with time as well though Ivanovic, took years for g to become more or less the standard as well over b
Ivanovic: Deanjo: correct
Ivanovic: Deanjo: and so it will take time for usb3.0 to become standard
Deanjo: Ya, heck when USB 2 came out it took a couple of years to see devices using it
cxo: man you are naive, everything is driven by deman
cxo: PCI-E 2.0?
cxo: DirectX 10.1?
Ivanovic: cxo: yeah, and where *is* the demand for usb3.0?
cxo: if people want to adopt something, it will happen over night
Deanjo: Ya and what device takes advantage of PCI-E 2.0 yet?
cxo: Ivanovic, i dont know, did i say there is?
Ivanovic: you know, it is just faster (and, since it is new, more expensive) so why should there be a high demand?
cxo: Deanjo, go pick some cotton, cant you see i'm busy trolling here
Ivanovic: cxo: this is why i estimate that it will take 4 years from now until usb3 is widely adopted
Deanjo: RS-232a is the future imho :P
Ivanovic: Deanjo: yeah
cxo: yeah lets supercharge it
cxo: at 48v, we can switch it at 50ghz
Deanjo: Anything over 2400 baud is overkill
Ivanovic: but i really do not understand why anybody wants to put fiber optic cables down into the atlantic to get "more bandwidth" for internet between the US and europe
Ivanovic: that is: if you really want bandwidth, why not just completely fill one of those huge super tanker like ships with eg blue ray discs
Ivanovic: you get a *huge* bandwidth with this
Ivanovic: though the latency is bad, too...
Deanjo: Ya submarine ping time suffers :P
cxo: use the dolphins
Deanjo: Na tuna with friggen lasers on their heads
Ivanovic: cxo: nah, we need those in the tuna fish tins...
cxo: TCP/IP over Dolphins
Deanjo: Broadband Carrier Pigeons?
cxo: now you're talking
Ivanovic: IP over super tanker
cxo: but first i need to finish my PhD on PoIP (Pizza over IP)
Deanjo: Ya but them packet collisions could be messy Ivanovic
Ivanovic: Deanjo: nah, they are not expected to happen
Ivanovic: though the package size would be rather big
Deanjo: sprays some WD-40 on his nic cable to lubricate the bits
Ivanovic: are you insane?!?
cxo: yay Nortel and Cisco branded pidgins flying all over the place bumping against each other
Ivanovic: you got to heat the cables to get more speed
Deanjo: P2P would be a mess with the pigeons
Ivanovic: why else would *fire*wire be faster than usb?
Deanjo: Shhhh, I'm overclocking my built on audio to get Dolby 67.8 Uber-THX surround sound
cxo: so, DOS would be invented by Alfred Hitchcock then?
cxo: Denial of Service
cxo: The Birds?
Ivanovic: Deanjo: man, you got to underclock to be really up to date those days
cxo: get it?
cxo: you guys are slow
Ivanovic: you know, green IT and stuff like this
Ivanovic: cxo: no, multi purpose data transfer
Deanjo: lol, you know someone is going to read the logs and go "WTF?"
cxo: isnt that what AMD did with the athlon
Ivanovic: cxo: when you received the message you can still eat the pidgeaon
cxo: and it smoked the Atom?
michaellarabel: Deanjo, Hi
Deanjo: Ya the Athlon 2000+ smoked the Atom in power consumption and speed
Ivanovic: Deanjo: you really think someone will read this?!?
Deanjo: howdy michaellarabel
Deanjo: Ivanovic, it will probably show up on slashdot lol
cxo: our friendly open source driver liaison
Ivanovic: Deanjo: ah, okay, then noone will read it
Deanjo: And the inquirer will state it as fact
Ivanovic: and regarding atom and losing against the "old and outdated" athlon: is anyone of you really suprised?!?
cxo: that birds carry data? of course its a fact
Ivanovic: that is: the cpu itself is castrated
cxo: Nano kicked its butt too
Ivanovic: (from the features point of view like out of order execution and such)
cxo: they were playing crysis on the Nano at the launch :)
Ivanovic: it might be using not too much energy, *but* the chipset it comes with uses by far too much energy
Deanjo: And at least you can have a bloody PCI-E slot with the Nano/Athlon
cxo: imagine playing Crysis on your next cellphone
cxo: thats more of a chipset limitation though
Ivanovic: cxo: be carefull, maybe crysis will come for the pandora
Deanjo: No, intel is ordering manufacturers to not implement a PCI-E slot
cxo: atom is just a toy, they have bigger stuff in the works
Ivanovic: no "real" extension slot is intel politics
cxo: atom was for the netbooks, sorta to annoy VIA
Ivanovic: beside this i really love intel saying "man, no real user will want to use it, it is to f***ing slow"
Ivanovic: i can just say: even this castrated CPU has more than enough power for everage joe
Deanjo: It would have it's uses if they didn't castrate the board as well
Ivanovic: who just surfs the net, writes some text docs and spreadsheets and that's it
Deanjo: IE small home server
cxo: imagine if you scale a 386 to 45nm
cxo: you wouldnt be able to see it
Ivanovic: Deanjo: for a small (yeah, really tiny and small) home server i am using an nslu2
Ivanovic: combined with a 2.5" external hd
Ivanovic: nothing fast and such, but enough as "remote backup" and irc bouncer
Deanjo: Right but it would have been nice if they would have at least put gbit lan and 4 sata connectors
cxo: needs a server
Ivanovic: (plus a small svn server when needed)
cxo: i have a sound server though
cxo: 500 Mhz Cyrix-C7-2, 512mb ssd, 512mb ram
Ivanovic: okay, the nslu2 is just 100mbit, too (as my switch at home is)
cxo: running an maudio1010
Ivanovic: that is the great thing about this server solution is the power consumption
Ivanovic: i am <10W for the whole box including HD and stuff
cxo: what cpu?
Ivanovic: (that is, the power device is just able to deliver 10W)
Ivanovic: cxo: in the nslu2 a 266MHz arm chip is used
Deanjo: Ya but it's too damn small for the pigeons to land on Ivanovic
michaellarabel: Deanjo Hi
Ivanovic: Deanjo: yeah, this is right
Ivanovic: michaellarabel: you repeat yourself
Deanjo: Hi michaellarabel
cxo: wonders i michaellarabel is going through a relapse
Ivanovic: [01:13:50] Deanjo, Hi
Ivanovic: [01:21:25] Deanjo Hi
cxo: or has become a cyborg
Deanjo: Na he just has a short attention span :P
michaellarabel: oops, was just getting home from a business meeting and forgot if I typed a message or not before running away from the computer
Ivanovic: hmm, is michaellarabel really just an irc bot?
Deanjo: elisa on IRC :P
cxo: could be
michaellarabel: Anyways, is the motherboard detection good now for you?
cxo: if you cant tell who your mother is you got problems
Deanjo: Ya, all except for the 780a chipset
Deanjo: But that's not a PTS problem
michaellarabel: Can you run a command for me? I found a very concerning regression today that I know what's going on but now sure how it happened...
michaellarabel: MONITOR=all ./phoronix-test-suite benchmark idle and then type idle of like 3 or 4
Deanjo: Sure as long as it doesn't involve rm -rf :P
Deanjo: NOTICE: php-pcntl must be installed for the system_monitor module. <--- is that supposed to come up all the time?
michaellarabel: When that runs, I need to know if it's ending the monitoring process prematurely... where its monitoring ends but test still running
michaellarabel: hmm how many times does it show for you?
Deanjo: Just once
michaellarabel: are you able to install that?
Ivanovic: michaellarabel: http://gitweb.freedesktop.org/?p=mesa/mesa.git;a=tag;h=96842e4a2667227f8a2232412072c7ca7a354008
Deanjo: Ya just installing it now
michaellarabel: Ivanovic, yes, thanks... typing it now. Was in a meeting the past two hours
Ivanovic: michaellarabel: might be worth a news that mesa 7.1 is finally tagged and that xorg should maybe be out soon, too
Deanjo: when it's done, you want me to paste the output to pastebin?
michaellarabel: sure, thanks
Ivanovic: time for 'lil Ivanovic to head off to bed, got to prepare for exams in the next weeks and should get as much sleep as possible...
Deanjo: Here you go michaellarabel http://pastebin.com/m24b0bab
michaellarabel: Damnit, this confirms the bug. This is a bad regression
Deanjo: blames Ivanovic
Deanjo: Be back in a few minutes, got to check something
Deanjo: Some days I tell ya my one daughter makes me think about putting her on craigs list :P
Deanjo: Any news when the new lm-sensors is going to be out michaellarabel ?
michaellarabel: Deanjo, not sure off hand. Well, I may know but too busy thinking right now about this PTS bug to check my notes.
michaellarabel: The difference between 1.2.0b2 and 1.2.0b3 is about a 2,200 line patch... and I just looked over it and nothing jumped out for causing the bug
michaellarabel: Decided to learn git-bisect to help debug this problem :)
michaellarabel: Fixed in git :)