Phoronix IRC Log: 2008-12-31
Kano: Deanjo: do you know zero ballistics?
redeeman: hey kano
redeeman: do you know where i can find out what the different "NVRM: Xid" things mean?
redeeman: i get sometimes:
redeeman: NVRM: Xid (0001:00): 13, 0003 00000000 00008297 00000980 00020000 0000000c
redeeman: NVRM: Xid (0001:00): 6, PE0003
redeeman: apparently, it's something that happens ALOT with 180.xx, but im on stable 177.82
Kano: thats some nvidia error. you only can change driver, i prefer 180.18. do you want to play a panzer game against me
redeeman: but surely those errors have some sort of implifed info, or they would just go: "NVRM: Dang, we fucked up, but hah! you already paid us!"
redeeman: about game, sorry, i have to work, must get clients app finished till weekend
Kano: redeeman: for testing vdpau you have to use 180.16+ anyway and for normal use 180.18 seems to me really stable
redeeman: yeah i was almost convinced 180.x was the way to go, then i saw the thread on nvnews
Kano: but i disabled composite
Kano: for vdpau
redeeman: i don't use compositing at all
Kano: best disable the extension then
Plouj: michaellarabel: I find it increasingly annoying to read phoronix articles with links leading back to phoronix or even phoronix search results. It's not very useful if I want to find some "original source" information about a developer or a project which the article mentions.
Plouj: michaellarabel: and in the "Triangles On The ATI RV770" specifically, I think you meant "It *was* expected that there will be a modest level of open-source 3D support for the ATI R600 and R700 series in the first half of 2008 for end-users"
Plouj: on the same article:
Plouj: "can be found in the r600_demo Git repository at FreeDesktop.org" should link to the actual git repository URL and not just freedesktop.org
michaellarabel: Plouj: Phoronix isn't allowed to link to FD.o Git repositories because everytime we have previously we ended up overloading the server with traffic
michaellarabel: Their cgit/gitweb is on a different server from the main page
Plouj: well, that explains a lot
Plouj: they should up their infrastructure
Kano: michaellarabel: is it possible to record the statts for cpu usage for each core in one graph?
michaellarabel: Hmm not sure if I ever tried to read the usage from each individual core. What's the easiest way to access that?
Kano: good question,what did you use before
Kano: because when i record a h264 playback i see 48% max, but thats a dual core cpu
GNU\colossus: /proc/stat is where you'd loke for those stats, I'd say
Kano: one core @ 100 would be not that good
Kano: GNU\colossus: right, there is cpu + cpuX
Kano: but where is the load in that line?
michaellarabel: Kano: I might have a way
michaellarabel: I am using /proc/stat already in PTS to compute the load, so I just need to add support for reading an individual core instead of just "cpu"
Kano: would be good
GNU\colossus: just have a look at top from procps
GNU\colossus: it's using the same interface
michaellarabel: Just committed the support to pts-core for being able to read individual cores. Now just to make the system_monitor module support reading multiple CPU usages
Kano: also did you think about setting a max to 100 % for the scaling? it look a bit bad when the max is 115% ;)
michaellarabel: It should be like a 2-line patch. I can probably get to that once I get done with the system_monitor
Kano: a last thing, maybe add
Kano: git clone http://phorogit.com/repo/phoronix-test-suite.git
Kano: to the download section of pts
Kano: i always have to google for the url ;)
maligor: the catalyst 9.1 betas seem to have lots of opengl 3.0 required extensions
Kano: did you ever look at kernel.org for the git?
Kano: i like that because it always writes the git url when browsing
Kano: you see the URL line
Kano: pretty good
maligor: git.freedesktop.org is shinier
Kano: maligor: i am sure linus wants to hear that ;)
Kano: but it shows the clone lines too
maligor: the kernel git browser is maybe a bit cleaner tho
Kano: maligor: glxinfo|grep -o -e GL_ARB_texture_rg -e GL_EXT_texture_swizzle -e GLX_ARB_create_context|sort -u|xargs
Kano: what does that show?
Deanjo: Kano, you were asking about zero ballistix?
Kano: i have got my own server running, the playground, before you see my callsign ;)
Deanjo: Ya, I ran across that a while back
Deanjo: Oh You have your own server running
Kano: the other is not stable enough
Milyardo: Looks interesting
maligor: rather too simple for my taste
maligor: Kano, GL_ARB_texture_rg GLX_ARB_create_context
Kano: maligor: then play it ;)
Kano: maligor: ok,that seems to be opengl3
Kano: install it an join the playground,will see that
Milyardo: I'm downloading right now :)
Deanjo: Oh dear, I see another Compiz soap opera is developing
Kano: Deanjo: it is boring after a while
Kano: nothing i need to use daily
Deanjo: I know, the only thinh I find remotely useful is the cube
Deanjo: Compiz never had great integration with KDE and as such when Kwin4 came out it caused alot of people to abandon compiz
michaellarabel: Kano: Hoorah, committing system_monitor usage for multiple cores
Kano: michaellarabel: fine will monitor gaming then
Kano: Milyardo: join the other server
michaellarabel: Kano: Everything should be in Git now for system_monitor, 100% graph fix, and core monitoring
maligor: Kano, yeah, it sort of looks like it, it segfaults if you try to use the new context creation stuff tho
Kano: ok, will try
Kano: michaellarabel: i used MONITOR=all but i dont cpu1 in the graph only combined and cpu2?
michaellarabel: Kano: With /proc/stats is there a cpu, cpu0, and cpu1 ?
Kano: yes 0+1, not 1+2
michaellarabel: Well in the key it will show you the real core + 1
Kano: but not in the graph? i see no line for it
michaellarabel: On the graph is what I am talking about. Can you pastebin your /proc/stats
Kano: you see what i mean?
michaellarabel: Kano: No?
Kano: there are only 2 lines not 3 in the graph
Milyardo: Kano: The battery on my laptop died
michaellarabel: Kano I think that CPU1 might be behind CPU2
elfgoh: I am sorry if i missed this but are there any channel logs?
elfgoh: Am referring to #phoronix
Deanjo: Hey michaellarabel, have you started on that retest of opensuse yet?
michaellarabel: I did but had ran into a problems with the barriers option so I postponed it until I had the time to see what was messing up.
Milyardo: elfgoh: http://www.phoronix.com/scan.php?page=irc
d2kx: does anyone else have got the problem that Radeon/RadeonHD dont stop requesting EDID information so that the Xorg.0.log is getting longer and longer and longer when using (K)Ubuntu 8.10?
Deanjo: OK, should have been as simple as adding barrier=0 to the fstab line
elfgoh: Milyardo: TKs!
elfgoh: Happy New YEar!
redeeman: xfs needs nobarrier
Deanjo: The issue is with the Ext3 in suse redeeman
Deanjo: Referring back to the head to head results where the RC of opensuse was tested. suse enables barriers by default
Deanjo: other distro's don't
Milyardo: Deanjo: RHEL doens't enable barriers by default as well?
Deanjo: Not sure. I know the other mainline distro's don't such as fedora, mandriva, *buntu, etc
Milyardo: Yes I know Fedora doesn't :)
Deanjo: I haven't touched RHEL in a long time. I know CentOS doesn't as well
Deanjo: At least it didn't in 3.9
Deanjo: or 4.7
Deanjo: I'm not sure how close Cent copies RHEL with that respect
Kano: michaellarabel: how should the 2nd cpu have got the same line than the 1st when the combined is not also 100 % but lower?
michaellarabel: Kano: That's what I am trying to figure out... I have something similar on an 8-core system with some lines being nested behind and the average being lower
michaellarabel: It might be with how /proc/stats is setup
Kano: which values are the cpu load?
Deanjo: You got that new badass system up and running yet michaellarabel ?
michaellarabel: Kano: Bottom of http://phorogit.com/index.php?p=phoronix-test-suite.git&b=494d781cb85ae99ed95aa8368a9d150fd70cf863 that page with the hw_cpu_usage() and hw_cpu_load_array() functions
Milyardo: now has lots of free time now that school is out
redeeman: "new badass system" ?
Milyardo: I need a project to start contributing to
michaellarabel: Deanjo: Shanghais? I have two of them right now in a Tyan board. Waiting on my quad-socket board, but unfortunately I don't have an MEB chassis for it yet...
michaellarabel: Milyardo: Phoronix Test Suite :)
Deanjo: Ahh, Tyan still hasn't got the new board out yet?
Milyardo: begins to brush up on his PHP skills
michaellarabel: The new revised one has its next production run in early January
Deanjo: Paper launch of the board?
michaellarabel: Do you happen to know where to find any MEB cases that aren't that expensive?
Deanjo: heh, I don't know if such a beast even exists
michaellarabel: Cheapest MEB case I found so far was like $700
Milyardo: "Description: Unnamed repository; edit this file to name it for gitweb." :o
Milyardo: michaellarabel: Is there a bug tracker? or some sort of task list I can find?
michaellarabel: Nope, just the forums. Since usually when a feature request comes in I have it done within an hour or two. There is though "Future Work" @ http://phoronix-test-suite.com/?k=features
Deanjo: MEB is 13"x16" is it not michaellarabel ?
michaellarabel: Deanjo: It is
Deanjo: Wouldn't this suffice? http://www.coolermaster.com/products/product.php?language=en&act=detail&tbcate=17&id=13
Deanjo: M/B Type Micro-ATX, ATX, E-ATX 12"x13", W-ATX 13"x16"
michaellarabel: Hmmm... I looked at it a few days ago on NewEgg, but it had only mentioned E-ATX
michaellarabel: ith%20Side%20Panel%20Window%20-%20Retail 13 x 16 in there might be a tight fit
michaellarabel: Deanjo thanks though, I'll probably buy it and hope it all fits :)
michaellarabel: Actually, no it won't since it's 16" wide as opposed to 16" tall
Ivanovic: wow, that was difficult
Ivanovic: i had not imagined that finding some beds for fosdem could be this difficult
Ivanovic: (that is from wesnoth we needed 8 beds and chab as well as bruegel and most other hostels are booked out already)
Deanjo: That was a S4992 you were getting michaellarabel
Deanjo: I'll see what I can find, who knows ebay might even have something
michaellarabel: k thx
Deanjo: Geez michaellarabel, I'm pretty sure that board will fit in the coolermaster http://www.coolermaster.com/products/product.php?language=en&act=detail&tbcate=17&id=13 Take a look at the board they have in the pic
michaellarabel: From the other pics though, it didn't look like there were holes for standoffs at the very top
Deanjo: If you look at the back the IO slot seems to be in the right place
Deanjo: You might have to do some creative drilling for extra standoffs
michaellarabel: http://www.tyan.com/archive/support/html/r_s4882.html the S4882 has the Stacker 810 as being a supported option
Deanjo: Ya, so it should work then
michaellarabel: Yeah, thanks for mentioning it
Deanjo: Save ya some money anyways lol
Deanjo: I lucked out on my server case, the store owner was actually throwing it out because it sat on the stockroom floor for to long. Who actually makes it is beyond me
Deanjo: The thing is huge. 36 " tall
Deanjo: Not made for lan parties :D
Milyardo: I got mine from work
Milyardo: an old Sun 5U
Milyardo: It's edges are deadly
Deanjo: I've got an old IBM server somewhere that would probably fit one of those MB's too
d2kx: i think i'll take a rest of using computers now. i will stop using any computer until the next year. bye
Deanjo: He lies
d2kx: i dont. oh its already 8pm *goes offline now*
Deanjo: y2k9 bug in zune 30G http://techreport.com/discussions.x/16133
Kano: win will not work 2009 ;)
Milyardo: Yeah a 3 digit year...
Milyardo: That means the Zune Developers had to go out of their way to make that bug
Deanjo: Well to be fair I guess that was their first zune
Deanjo: Quirks happen on 1st gen stuff
Milyardo: Did I say digit? I meant bit :P
Milyardo: but yeah, I don't even know how that bug begins to make sense as an accidental oversight
Deanjo: Not surprising though, after all MSI did manufacture them as well
Milyardo: I wonder who makes a regular habit of shorting their ints to odd number of bits though
Milyardo: Its probably for something like signing the Firmware, if the source was stolen, its unlikely that someone would bother to rebuild the firmware with 3bits for random date variable
Deanjo: Hmm, looks like Alan Cox jumped ship
Milyardo: Deanjo: Yesterday's news
Milyardo: get with the times
Deanjo: Well sorry for not following "The Life and Times of a kernel dev"
Lightkey: you only need to follow the IT news sites to know that :p