Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
AMD Graphics Stats Hardware Linux

28-Way Radeon GPU Comparison Under Linux 85

Posted by timothy from the in-case-you're-a-bit-compulsive dept.
An anonymous reader writes "Phoronix has conducted a 28-way Radeon graphics card comparison under Linux that illustrates the differences among the open-source and closed-source graphics drivers."
This discussion has been archived. No new comments can be posted.

28-Way Radeon GPU Comparison Under Linux More

28-Way Radeon GPU Comparison Under Linux

Comments Filter:

  • I like... (Score:3)

    by elsurexiste ( 1758620 ) on Tuesday September 20, 2011 @08:24AM (#37454970) Journal

    the fact that in TFA they say it's AMD comparison on Oktoberfest. These people don't take breaks!

    • Re:I like... (Score:4, Insightful)

      by robthebloke ( 1308483 ) on Tuesday September 20, 2011 @08:36AM (#37455082)
      I like the way they decided to spread the article across 38 pages that each take a minute to load. Oh. Spoke too soon. Now it seems to be slashdotted. I'm guessing this means no one will be reading that article. It's phoronix though, so I'm guessing the rest of article will be something along the lines of:

      * Here's a test of lots of 3D graphics cards from AMD.
      * Now let's conduct a large series of 2D tests, all of which are entirely irrelevant for users of 3D graphics cards....
      * Conclusion: Could this be the year of linux on the desktop?

      • Re: (Score:3, Informative)

        by jedidiah ( 1196 )

        Just use Nvidia.

        Better 3D performance.
        Better monitor support.
        Stuff like VA-API is actually supported.
        Stuff like CUDA and OpenCL if you tend to blither on about Beowulf...

        • Re: (Score:1)

          by Anonymous Coward

          Perhaps (though I highly doubt there's any difference in "monitor support", except maybe AMD supporting more of them at the same time on most cards), but with nVidia you also get more evil. So it's a tradeoff between 3D-performance and supporting evil.

          • Re: (Score:1)

            by m50d ( 797211 )
            I've got a radeon under my desk that lets me choose between two monitors or 3D rendering. I don't see any more evil in nvidia's support - with either nvidia or AMD, you get the choice of crappy open drivers, or better binary-only drivers. The only difference is nvidia's binary-only drivers are much better than AMD's.

            • Re: (Score:3, Informative)

              by Tomato42 ( 2416694 )

              Open AMD drivers are actually usable: KDE4 composing, XV acceleration, not to mention 2D performance.

              In fact, I'd say that the OSS drivers have (or at least had 8 months ago when I still struggled with fglrx) better 2D performance, If you have a 4096x1536 desktop you could actually count the seconds it took to move a window. So I'd say that 2D performance is actually more important than 3D performance. If I want to game I run the gaming OS anyway (for 3D intensive stuff).

              • Re: (Score:1)

                by m50d ( 797211 )

                In fact, I'd say that the OSS drivers have (or at least had 8 months ago when I still struggled with fglrx) better 2D performance, If you have a 4096x1536 desktop you could actually count the seconds it took to move a window.

                I'm still in that position, so I'm guessing your card is better supported than mine or something. (This is on my work machine; on my home machine with an nvidia card running freebsd everything moves instantly. So I'm in a position to compare)

                • I'm using HD 4850 together with Xorg 1.10.3.901, mesa 7.11 and Linux 3.0. YMMV, but I've been using this (or with a bit earlier versions) setup for the past few months with dual-head setup and have had absolutely no performance problems as long as I stick to 2D (even with KDE4 composing). On my work machine I have an integrated AMD card and have used it with OSS drivers for the past year or more, there were occasional performance problems in the beginning but nothing major, now it's practically flawless, bu

          • Re: (Score:1)

            by Anonymous Coward

            "except maybe AMD supporting more of them at the same time on most cards"
            Oh man, I just checked and you're right [phoronix.com]! Damn, guess I'll really have to consider retiring my current nVidia gpu + Matrox Dualhead2go setup...

          • Re: (Score:2)

            by jedidiah ( 1196 )

            Plenty of TVs throw off bogus EDID info that knarfle some Linux video drivers.

        • Re: (Score:1)

          by Anonymous Coward

          Soo.. you didn't even bother to read the keyword "open-source" here? The open source nvidia driver -- nouveau -- also does not support your stated features.

          • Re: (Score:2)

            by jedidiah ( 1196 )

            "the differences among the open-source and closed-source graphics drivers."

            I would love an alternative for nvidia on mini-itx boards. I would love an alternative in general.

            It sucks to only have one viable option for some things.

        • Comment removed based on user account deletion

          • Re: (Score:2)

            by marnues ( 906739 )
            You seem to like hyperbole. It might be fun to get your feathers in a ruffle, but it doesn't aid constructing a logical argument. Just a thought for you...how many people are knowledgeable in writing drivers, reading gpu specs, and have free time to give to this particular project. The number is clearly smaller than we all would like.

          • Some counters from the aricle.

            "While many users want this support so that they can move their video playback process to the GPU rather than taxing the CPU, chances are we will not see this open-source support either. While Advanced Micro Devices tries to be open-source friendly, they refuse to document or provide sample code for the UVD engine. The UVD area of the graphics processor is off-limits to official AMD open-source support over fears that exposing it could lead to compromising its Digital Rights Ma

            • Comment removed based on user account deletion

              • I was countering your point about "Wow, way to show undeniable proof that corporations should NOT support Linux! What did the community say? What were the words constantly used? "If you'll just give all your specs to the community why we'll not only do all the driver work FOR you, but we'll buy your product!""

                There are outstanding issues that prevent the above comment from being done which has nothing to do with the open source community and at least with the patents with AMD either. In essence "If you'll j

        • AFAIK, Radeon cards have the edge in performance at the moment, at any given price point, and run cooler. Nearly all Radeon here, and significantly better quality of life for me than in my NVidia days. My experience with NVideo Quadros, very strange behavior that can be quite frustrating. Like requiring a hard power cycle, not just a reset, to function properly in some situations. And not displaying text after switching to the text console. Mind you, I appreciate and respect NVidia, they are a worthy compet

          • But then there's those of us who have had the opposite experience. The ATI driver has never supported anything newer than a Rage Pro that I've tried it on; oh, it claims to support it, but then I get free reboots, or display trashing, or one of a million other problems.

            Mind you, my nVidia experience has not been trouble-free. I had a Quadro FX 1500M in a HP/Compaq with a Core Duo. nVidia admitted they had a die bonding problem. HPQ didn't. But after they sent a tech to my house twice and killed the laptop c

      • I clicked page 7 at random and it was literally a page long description of a single video card from 2006 with no performance information at all. I can't even imagine why someone would still own a Radeon 2900XT, the power bill alone would make it worth replacing at this point.

      • The tests seem to be 3D, but either one driver did not work, or the game used was CPU heavy and the difference are minimal

        The test suites used seems odd, they have the closed source driver using more CPU than the OpenSource? I suspect they are using the wrong tests?

      • What a dick response. Evidently, you opened the first page, and the only thing you looked at was the number of pages. Did you even read the last paragraph?

        As a note before proceeding, anyone who wishes to view these results ad-free (Please no AdBlock or similar; Phoronix.com is run single-handedly and funded entirely by advertisements and Phoronix Premium. This article alone required more than 310 hours to complete at great expense.) and view the entire article on a single page, can do so by subscribing to Phoronix Premium.

        If the multiple pages bothers you so much, then shell out for a subscription. If you don't want to do that, then just shut up and move on. Don't shit on someone else's hard work and complain that they don't give it to you totally for free.

        And why would anyone besides a total sociopath mod something like this up?

      • There haven't been even a single 2D performance test in The Fine Article. It all was about 3D performance, power draw and CPU usage.
  • well assuming there were a gallium or similar driver for them...
  • Is my internet sucking again or did the site already get slashdotted?

  • don't bother (Score:2, Informative)

    by Anonymous Coward

    30+ pages filled images or graphs. The gist of it: 3D capabilities of open source drivers still suck, and the newer cards suck worse.

    • Re: (Score:1)

      by antdude ( 79039 )

      And no single page to show all 38 pages!

    • Comment removed based on user account deletion

  • 38 page click through article (Score:3)

    by RichMan ( 8097 ) on Tuesday September 20, 2011 @09:14AM (#37455470)

    Yeah they deserve to be slashdotted for the design madness that multiplies readers by x38 or more.

    • Re:38 page click through article (Score:4)

      by Thelasko ( 1196535 ) on Tuesday September 20, 2011 @09:37AM (#37455744) Journal
      On the last page they mention the page layout was a response to all of the readers using Adblock. They note that you can pay a fee to see a single page article.

      • Re: (Score:1)

        by Anonymous Coward

        On the last page they mention the page layout was a response to all of the readers using Adblock.

        Well... I guess they F-ing showed us!

      • Yeah... a holy war against AdBlock will definitely turn out well for the guy that runs this site.

        I think he's had a few too many beers.

        • It doesn't help him that 3/4th's of the ads are flash only.

          I can only imagine the horror of browsing that site with flash enabled, and I only made it 3 pages in before getting bored.

      • Re: (Score:1)

        by Anonymous Coward

        Well they don't have to worry about me using Adblock on their site any more. What I mean is, they won't have to worry about me visiting their site anymore.

        • Re: (Score:2)

          by Atriqus ( 826899 )
          +1 on that; added this a while ago to my /etc/hosts.

          127.0.0.1 phoronix.com
          127.0.0.1 www.phoronix.com

  • fun little menage a trois.

  • An Open Response to Moronix (Score:5, Insightful)

    by drinkypoo ( 153816 ) <drink@hyperlogos.org> on Tuesday September 20, 2011 @09:45AM (#37455832) Homepage Journal

    If you did not make it through all of the pages, you can subscribe to Phoronix Premium so you can quickly scroll through all of the pages on a single page (and without any advertisements), so that you can draw your own conclusions about the state of open and closed-source Radeon Linux driver. (There also isn't much more of a conclusion since the large number of readers using AdBlock ruin it for the rest of you, since Phoronix.com is run single-handledly and funded entirely by advertisements and Phoronix Premium.)

    I am a proud AdBlock user. I think I speak for the rest of the AdBlock users when I say that if your publication cannot survive without annoying and intrusive advertisement, we would prefer that it go the fuck away and stop making an ugly hole in the internet.

    I hope Phoronix tanks, and soon. I prefer my articles in an essay format. For those of you who slept through elementary school, such a format involves an introduction, a body, and a conclusion. Crying about how much it cost you to develop an article that people aren't willing to pay for is what losers do. Winners go home and fuck the prom queen.

    • Re: (Score:1)

      by PhxBlue ( 562201 )

      Winners go home and fuck the prom queen.

      There were a lot of winners at my high school way back when.

    • I hope Phoronix tanks, and soon. I prefer my articles in an essay format.

      Yep, there's a reason why I pay a LWN subscription.

    • Re: (Score:2)

      by rdnetto ( 955205 )

      I don't normally use AdBlock. I only installed it on my netbook the other day because I my wifi connection was at modem speed. But this sort of behaviour guarantees I'm not going to be reading that site.

    • Dear Phoronix,
      You're like Toms Hardare bent over Eugenia in OSNews and had a retarded lovechild.
      Signed, Everyone

  • I realise it took a long time to do these tests and I realise that sites should be compensated in some way to justify their time and expenses. But 38 pages is taking the piss. There must be a better way of arranging an article that balances ad impressions against a visitor's patience. I doubt many people would bother to wade through that many pages and probably just skip to the end so the exercise is counterproductive.

    On the subject of AMD drivers, they stink. Both the closed source ones and the open sour

    • Re:38 pages (Score:5, Interesting)

      by Sloppy ( 14984 ) on Tuesday September 20, 2011 @10:13AM (#37456206) Homepage Journal

      I realise it took a long time to do these tests and I realise that sites should be compensated in some way to justify their time and expenses. But 38 pages is taking the piss.

      The saddest thing about Phoronix is that even if you pay him it's still 38 ad-laden pages. That is, unless you remember to load the forum page first, before you RTFA. And he has known this for years (that's not an exaggeration), received numerous complaints from his paying customers about it (doesn't have the "I didn't know you people were unhappy" excuse) and still hasn't done anything about it.

      And I really mean "sad." That is the best word to describe it. It's a (minor! I'll admit!) tragedy in the original sense of the word. The site is basically a good idea and otherwise handled well (for what it's intended to do), but is dooming itself simply because the central character refuses to see the elephant in the room that everyone else sees.

      I don't hate Phoronix and will be sorry to see it go (tables of numbers have their place). And yet I'm probably not going to renew my support. Maybe its loss will make new room for the same niche, but with someone else handling this basic common-sense in-your-face aspect sensibly.

      • I don't hate Phoronix either, but he's in a bad place trying to monetize eyeballs connected to brains capable of tracking more moving objects than the Chicago-Washington corridor after a threat level upgrade from orange to burnt lavender. The mental airspace is crowded enough without adding biplanes with banners.

        Why do the people crowing "use NVidia" put me in mind of Jack Nicholson shouting "You can't handle the documentation!" Damn straight, ya bunch of cowards.

        Apparently, with enough documentation, n

        • When the going gets tough, the tough get Binary Blobs. Not the banner I originally unfurled.

          ATI keeps promising to provide enough specs for proper support, and it keeps not happening until well after the fact. That, or the developers are a bunch of douches. I don't believe THAT, though, because the driver actually does work super-well for antique (well-known) hardware. Until the situation resolves itself (and ATI drivers have always been trash - I was getting crashes from ATI Mach32 drivers in Windows 3!) there is nVidia. Or maybe someday Intel will figure out how to make a GPU worth using for mor

  • Comment removed based on user account deletion

  • FTIW, I like Phoronix!

    They are the only ones testing a wide variety of hardware using Linux.

    There is no contender and I have based many buying decisions on their articles. I still read articles testing hardware on Windows, but their results not necessarily translate to Linux.

    So, FTIW, I like Phoronix!

  • Last 5 Pages (34-38) (Score:1)

    by Anonymous Coward

    With the very light workload of idling and video playback, the graphics cards when using the Radeon driver already tended to be warmer than the Catalyst driver, which properly supports PowerPlay and other power management technologies by default. The open-source power management options must be toggled manually and even then the mileage may vary. Some of these graphics cards were easily 8~10+ Celsius warmer when using the open-source driver! The Radeon HD 5830 was more than 20C warmer with the open-source d

  • Quote:
    * Here's a test of lots of 3D graphics cards from AMD.
    * Now let's conduct a large series of 2D tests, all of which are entirely irrelevant for users of 3D graphics cards....
    * Conclusion: Could this be the year of linux on the deskto?

Slashdot Top Deals

Think of it! With VLSI we can pack 100 ENIACs in 1 sq. cm.!

Close