Phoronix IRC Log: 2008-10-15
Milyardo: Morning #phoronix
Milyardo: whats new michaellarabel
Kano: the new flash 10 is crap, even the .tar.gz is compiled against too new libs. thats the same problem like the last beta
Milyardo: Hey michaellarabel: I should do you a favour and rewrite all of PTS in a real language like Perl :)
copper: michaellarabel Deanjo : I got an answer from a staff member of Tom's Hardware France about the PTS OpenSSL benchmark results
michaellarabel: copper: What did he say?
copper: he basically says the results aren't out of whack, just not what I "expected", and what matters is that the results are comparable relative to each other (i.e. Intel results can be compared to AMD results)
copper: Whatever they've done, I doubt it's ethical, to say the least
Milyardo: Whats going on with Tom's Hardware?
copper: Milyardo: they used the PTS but their OpenSSL results are completely out of whack
copper: compare http://www.tomshardware.com/charts/desktop-cpu-charts-q3-2008/Linux-OpenSSL,844.html to the results uploaded on PTS Global
copper: michaellarabel: message before last if you want the reference (in french) http://www.presence-pc.com/forum/ppc/Lesnewsvosreactions/performances-meilleurs-processeurs-sujet-31818-1.htm
Milyardo: Yeah there's something wrong with those results
copper: no kidding
copper: I've told them about it over and over again, both in french and in english (Tom's Hardware US)
copper: both redactions assured me they would "look into it"
copper: I've also had to link them to Ars Technica's article about the VIA Nano too, in which they find out PCMark gives dodgy results for the Nano
copper: the most I got from them is to include a link to Ars Technica's article, but no further research on PCMark's integrity was done
Milyardo: Thats not cool
Milyardo: Well thats proboly why I don't visit Tom's Hardware
copper: they also published wrong results for the Phenom's Sandra memory bandwidth benchmark, and they stayed online for months, despite Legit Reviews' findings that the outdated version of Sandra's software put the Phenom at a major disadvantage
copper: now Tom's Hardware has the correct results, placing the Phenoms above the most expensive Intel CPUs in that benchmark, but that's only because they recently decided to make a major update to their charts, and in that process a more recent (and corrected) version of the Sandra software was used
Milyardo: Average: 43.02 Signs Per Second
Milyardo: I like how my Sempron LE gets more sign persecond than Tom's X3 Phenom :)
copper: I'm not into conspiracy theories according to which Tom's Hardware would be accepting paychecks from Intel, I just think they're flaky and no better than any two bit fan site
Milyardo: that or they're just incompetent
copper: Milyardo: my Phenom 9600 scores almost *4* times more signs per second than theirs - and even the Intel results are way too low
copper: michaellarabel: btw the guy at Tom's Hardware US who told me they would look into it, turned out to be no less than the chief redactor (not sure what the correct term is in english)
copper: Chris Angelini
copper: Anyway, gotta go (afk)
Milyardo: WHy such poor performance on OSX?
GNU\colossus: it's called OSuX for a reason
copper: michaellarabel: now the TH staffer is being condescending: "Those results are real. Try testing 54 processors under 43 applications and then come and tell us how we should do it, et why no-one else does it."
Milyardo: ./phoronix-test-suite benchmark all
Milyardo: sudo shutdown
Milyardo: change processor
Milyardo: omfg I forgot the now
Milyardo: I guess I really am a nub who could never do what they do
Ivanovic: michaellarabel: uhm, you mixed some things up
Ivanovic: 8.54 will become 8.10
Kano: Ivanovic: well he just counts +.1 from one driver to the next
Ivanovic: Kano: and 8.9 is already at 8.53.2
Kano: but even that beta or whatever does not show pointsprints
Ivanovic: that is the 8.54.x series is 8.10
Kano: label="ATI Proprietary Linux Driver-8.532"
Kano: thats correct
Kano: that was 8-9
Ivanovic: and +1, which should be the 8.10 driver, would be 8.54.something
michaellarabel: Yeah, there was a miscommunication. It's 8.10.
Ivanovic: waits for a recent fglrx to install on his gentoo unstable system
Ivanovic: that is: for a driver that works nicely with xorg-server 1.5.2 and on my system...
Ivanovic: hmm, maybe it will be faster to just wait for video acceleration for r6xx chips in radeon(hd)...
Kano: it is just without installer,but everything needed is there
Ivanovic: you know, i prefer building the stuff myself, especially the kernel module
Ivanovic: build myself including the extra progs
Milyardo: michaellarabel: is there a test profile aimed at headless machines?
michaellarabel: Milyardo: a test suite? universe-cli
Milyardo: I've been wanting benchmark Performance under ESX
Milyardo: compare Debian running as a VM versus an native install
Milyardo: I'm getting a 404 on pts-gui when I sync git FYI michaellarabel
Milyardo: error: Couldn't get http://www.phorogit.com/repo/phoronix-test-suite.git/refs/heads/pts-gui for heads/pts-gui
Milyardo: The requested URL returned error: 404
Milyardo: *errrr when I clone not sync
maligor: oh, nice, they have the orig tarball too
maligor: just not very useful
maligor: files without installer
maligor: surprisingly fast action from ati with that, heh
Deanjo: Hmmmm, found a patent ALSA violates..
Deanjo: Quite possibly pulseaudio as well http://www.patentstorm.us/patents/5630175/fulltext.html
maligor: lalalala... I can't hear you
maligor: I guess they'll just have to disable dual card use
maligor: triple card use should be fine :P
Deanjo: Heh, disabling dual card use would mean a complete revamp of ALSA
maligor: Deanjo, actually I expect that has prior art
maligor: like early multichannel audio cards
maligor: hard to say if they used general purpose processors tho
Deanjo: Well it's the routing that could potentially violate the patent
maligor: "and passing the surround sound channel data to an additional audio card."
maligor: alsa doesn't do that
maligor: neither does pulse
maligor: beats me what it means either
Deanjo: Ya alsa does that
Deanjo: it can anyways
maligor: only some alsa plugins would infringe on that anyway
maligor: alsa doesn't really have any means to actively reroute data
Deanjo: Sure it does
maligor: I always thought inorder to copy datastreams back to hardware, you need a external app
maligor: funnily that patent doesn't mention the most problematic part of that setup anyway
maligor: which is: frequency shift
maligor: err.. frequency drift
copper: I'm routing S/PDIF input from my Terratec USB sound card to my Edirol UA-4FX USB sound card right now
Deanjo: In theory, AISO would violate that patent as well
maligor: hmm.. that patent seems to have some address routing stuff that alsa can't do too
copper: screw patents
copper: we're all criminals anyway
maligor: part 5
maligor: what's a surround sound algorithm?
maligor: surround sound algorithms mentioned are dolby stuff
Deanjo: ie: Dolby, SRS, QMSS, etc
Deanjo: And Alsa is fully capable of doing part 5
maligor: part 2 is a big vague, not sure what it means
maligor: how familiar are you with alsa structure?
maligor: mm.. yes, I suppose
maligor: silly patent
Deanjo: I'm fairly familiar with the alsa structure
maligor: the plugins use mmap?
Deanjo: They can
maligor: since that might comply with part 5 address stuff
maligor: but no default configuration actually can do that tho
maligor: if you sold a machine doing that, maybe... :P
Deanjo: I use to have 3 GUS's in a system that used the multicard surround sound trick on tracker modules :P
copper: what's the upside of using mmap()?
copper: with ALSA I mean
[TiZ]: Hi there. I'm having a little bit of a problem. After updating to kernel 2.24.21 in Ubuntu Hardy, and then updating to latest ATI drivers in tormod's repository and building current DRM modules... my X server's totally broken.
[TiZ]: I get the screen where the little X sits in the center and does nothing, and whiteness builds up from the bottom right corner of the screen.
copper: I don't think the state-owned TV network's news anchor knows what "live" means… he just said the US Presidential debate would be broadcast live at 5am CEST (the debate really starts at 3am CEST)
copper: *tap* *tap* *tap*… is this thing on?
michaellarabel: The Catalyst 8.10 is now out there w/o X Server 1.5.
maligor: no release notes th
maligor: amd's definitely improved their linux driver commitment :P
maligor: before it would've been a year late
Deanjo: Do you really need one? Probably looks near identical to the previous.
Roberth1: well the driver still needs a lot of bug fixing
maligor: Deanjo, bigger is better!
maligor: anyway, I'm waiting for xvideo sync fix :P
maligor: I don't really mind it that much but it would be nice
Deanjo: Ya, it would be nice to finally use that 690G board for it's intended purpose instead of having to pop in a nvidia card.
maligor: I guess the ubuntu release was final
maligor: the official drivers don't have x740
Deanjo: Then again the 690G had issues in windows as well. There seems to be a big circular area that has to be redrawn around the cursor when it is moved.
maligor: different driver in the ubuntu fglrx package too
maligor: or the md5 hashes don't match
maligor: oh.. you did say w/o :P
maligor: tempting to switch to 1.5 tho
Deanjo: thinks copper should by the 16 port IBM 8238 Tolkien Ring hub that he has sitting in his office.
Deanjo: $400 and it's yours. Cheap
Deanjo: lol finally a great security device. http://www.yankodesign.com/2008/10/14/this-cheese-is-not-for-you/
Deanjo: Damn is that one heavy heatsink. http://www.nordichardware.com/news,8242.html
Deanjo: " Due to the solid copper design, Thermalright Ultra-120 True Copper weighs 1.9kg (4.2 lbs), many times over what most of the high-end coolers of today weighs and more than twice as much as the sibling Thermalright Ultra-120 Extreme."
Deanjo: That's a hell of a lot of weight to stress the socket with.
Milyardo: I'm pretty sure you still have overclock very far with it
Milyardo: your FSB will melt before your CPU does
Deanjo: I really wouldn't even consider putting that kind of stess on the MB in a tower configuration. Desktop maybe
Deanjo: Not sure how much of an advantage it would give over the copper/alum setups. Copper is great for drawing out the heat but isn't as good as alum when it comes to dissipating it
maligor: weird changes in the ati drivers
maligor: pretty sure I haven't seen kcl_*.c files before
Kano: that final 8-10 driver is crap again, not even xserver 1.5 or 2.6.27 support. just the dri permission problem with etch is fixed
Kano: but that does not mean that it works, beryl is broken with etch and wine too
Kano: dont know why they needed to release that shit
Kano: did anybody see a _real_ improvement from 8-9 to 8-10?
maligor: there are 1.5 drivers tho
maligor: the ones in ubuntu's packages
maligor: I wonder why they dropped them from the real one
maligor: or why not release a real 'beta'
maligor: anyway, they seem to behave pretty much like 8.9
maligor: there's some weird new files in the drm module tho
maligor: unless I just didn't notice them before, the kcl_*
Kano: because ati is a synonym for silliness not cleverness
maligor: actually pretty possible I just didn't notice them, heh
Kano: the kernel module works with the same patch as 8-9,so not much could be changed there
maligor: yeah, definately the kcl_ files are new
maligor: typedef unsigned long KCL_TYPE_DevMajorMinor;
maligor: I wonder what it'd be short from
Kano: maybe it was used to fix the permission error on etch
maligor: I kinda doubt that
maligor: I suppose it could be possible tho
maligor: can't really make much of them
cxo: hey i get a silly error when i try the beta catalyst
cxo: Error: ./default_policy.sh does not support version
cxo: default:v2:x86_64:lib::none:2.6.27; make sure that the version is being
cxo: correctly set by --iscurrentdistro
cxo: anyone seen that before?