Forgot your password?
typodupeerror
AMD Open Source Games Linux

NVIDIA Is Better For Closed-Source Linux GPU Drivers, AMD Wins For Open-Source 185

Posted by Soulskill
from the best-of-different-worlds dept.
An anonymous reader writes "Phoronix last week tested 65 graphics cards on open source drivers under Linux and the best result was generally with the open source AMD Radeon drivers. This week they put out a 35-graphics-card comparison using the proprietary AMD/NVIDIA drivers (with the other 30 cards being too old for the latest main drivers) under Ubuntu 14.04. The winner for proprietary GPU driver support on Linux was NVIDIA, which shouldn't come as much of a surprise given that Valve and other Linux game developers are frequently recommending NVIDIA graphics for their game titles while AMD Catalyst support doesn't usually come to games until later. The Radeon OpenGL performance with Catalyst had some problems, but at least its performance per Watt was respectable. Open-source fans are encouraged to use AMD hardware on Linux while those just wanting the best performance and overall experience should see NVIDIA with their binary driver."
This discussion has been archived. No new comments can be posted.

NVIDIA Is Better For Closed-Source Linux GPU Drivers, AMD Wins For Open-Source

Comments Filter:
  • by Anonymous Coward

    ...for you guys who like closed source stuff:

    funny.exe
    boobies.exe
    yourprize.doc

    Have fun!

  • by Anonymous Coward

    I can't quite put my finger on it, but something tells me this is a job for a captain. Captain something or other. Anyone care to help me out?

  • Intel seems to have the only graphics that doesn't suck horribly on Linux for normal day to day use.

    • by Anonymous Coward

      Not that odd. AMD and NVidia are both trying to find ways to tweak the standards for tiny increases in framerate, Intel is honoring the standards and improving performance by boosting the power of the hardware underneath. That means Intel will continue to lag when it comes to all the different GPU metrics (but they did close the gap a lot recently), but since they're not trying to do shortcuts with the OpenGL and DirectX standards, they are much more straightforward to use.

    • by LWATCDR (28044)

      Intel graphics just plain suck... Actually they are not too bad these days but really do not complete with AMD or nVidia.

      • And yet I'm looking at lots of pixels driven by Intel graphics, with no obvious inability to support the applications I use day to day.

        Sure I have Nvidia in my windows gaming rig. But Linux is the topic here.

        • by LWATCDR (28044)

          OpenCL, Cuda, CAD, POVRay, Blender..... Not everyone uses Linux the way you do. There are many linux systems that do not have any graphics cards at all and just use a UART.

    • by jedidiah (1196)

      > Intel seems to have the only graphics that doesn't suck horribly on Linux for normal day to day use.

      Are you kidding? Their hardware sucks horribly for normal day to day use regardless of what OS you're talking about.

      • by haruchai (17472)

        What do you consider "normal day to day use"?
        In my experience, starting with Sandy Bridge & HD3000, it's been acceptable for Windows office desktop stuff, Office apps, web browsing, online streaming, etc.

        • Re:That's Odd. (Score:5, Insightful)

          by jones_supa (887896) on Friday June 13, 2014 @12:48PM (#47231313)
          Even a GMA950 can easily perform all those tasks.
          • by haruchai (17472)

            Probably not on Windows 7 and not with very recent versions of Office, FF, Chrome and Internet sites.

            • Yes, it can. Very easily.
              • by haruchai (17472)

                I decided to check to see if it would support my programs. It didn't take long to hit a roadblock.

                Requirements for Office 2013 - http://office.microsoft.com/en... [microsoft.com]

                Hardware acceleration Graphics hardware acceleration with DirectX10 graphics card

                According to http://www.intel.com/products/... [intel.com] , there's no Directx10 support from this board.

                • I see. You are correct, that would indeed mean that you cannot get full graphics acceleration in Office 2013, as the GMA950 is only DX9-compatible.
        • Well, you managed to mention not one thing where video power truly matters.

          The moment you go into games, game development, image processing, rendering and modeling, perhaps HD video playback (and processing?), or working with very high resolutions, your video card sure does matter, so does the quality of the drivers and its acceptance of standards (specially OpenGL).
          I found nVidia to be the safest bet in both those tasks I mentioned, as well as support for dual-booting while keeping the same capabilities in

          • by haruchai (17472)

            Did you read jedididah's comment above mine?
            I did specifically ask what he considers "normal day to day use".
            What I specified is the case for 80% of the 10,000 users that my organisation supports. And even so, there are fewer than 500 that have anything beyond a stock, onboard Intel graphics card.

            At home, I have a 9600GT but it's only now after perhaps 4 years that I think it's becoming the bottleneck in my main system despite 2 CPU & RAM upgrades in that time.

    • by MBGMorden (803437)

      I've been using a Nvidia card in Linux (with binary drivers) for years now with no issues whatsoever. Not sure how it would "suck". System boots, runs fine, and does what I need it to :S.

      • It sucks because it doesn't work until you download the drivers, install them and mess with the X configuration.

        With Intel graphics, it just works, because the open source driver code is integrated into Linux.

        • by MBGMorden (803437)

          Didn't have to do any of that. After install I got a little message saying "proprietary drivers available". I clicked and they installed. Its been that way for years now.

    • by OneAhead (1495535)
      That's only true if you don't need support for newer versions of the OpenGL standard. If you do, the Nvidia proprietary driver is pretty much the only viable option in Linux. The Intel driver is perpetually a few generations behind, and the AMD/ATI drivers are perpetually buggy (though I hear they're getting better).
  • Is that why, on my HD6450, Linux Mint setup boots into a garbled screen, PCLinuxOS skips the desktop installation, and Desktop BSD just gives me a black screen when any setup method is selected?
    • Is this something you tried recently? Phoronix did their testing with the 3.13 kernel.

    • by higuita (129722)

      CURRENT open source drivers should work well on that card, get a more update distro or manually update the kernel, libdrm, mesa and possibly the xorg-ati driver. Also, this open drivers status is for Linux, for BSD the open status may not be as good due the missing/incomplete lower level support in the kernel

      If it is failing on a recent distro, with recent kernel and mesa , you should open a bug (sometime fixing a bug on new cards can create another on older cards due the different features available)

  • by jones_supa (887896) on Friday June 13, 2014 @11:39AM (#47230717)

    In last week's testing of 65 GPUs on the open-source Linux drivers, the winner overall was the AMD Radeon graphics cards: they were the least problematic (though several Radeon GPUs still ran into different problems) and they delivered the best performance (including generally the performance-per-Watt).

    Can confirm. The open source Radeon driver has been improving greatly. A bit surprisingly, Radeon hardware is actually starting to become a quite good choice for a Linux user.

  • Yeah, as long as you ignore "NVidia Optimus". I have a AMD-A6 based desktop, it works fine with the occasional glitches, but the only thing that is truly stable and works for everything is called "Intel". There simply is no contest.
    • Depends on what you're willing to compromise. I have Nvidia Optimus, everything works perfectly for me. I had to use bumblebee and not Nvidia's own hacks, since NV's don't work yet, and bumblebee does, but it's pretty close to the Windows experience. While it doesn't auto-detect apps and select Intel/Nvidia automatically for me, it does allow me to manually force Nvidia usage much more simply, so I score that a wash.

      Intel for 2D/desktop - works great.
      Nvidia for performance 3D - works great.
      Auto-power-off of

  • by Anonymous Coward on Friday June 13, 2014 @11:45AM (#47230767)

    One note:
    AMD OpenSource drivers are best OpenSource drivers out there, but shitty drivers per se.
    NVIDIA drivers are great drivers, but not OpenSource.
    This is the real difference and conclusion. Don't try to hide it.

    • This!
      I don't care much if AMD's drivers are open when they are mediocre at best. Everything else seems to boil down to zealotic anti-binary-blob commentaries.

      I don't care if it's closed as long as it works. And nVidia works both in windows and linux, so that's where my money will go.

    • by MacTO (1161105)

      If we are going to be honest about things, we should also look at why: neither vendor is enthusiastic about providing complete documentation on the products.

      We should also be clear about some of the consequences. Better open source drivers provide a better long term solution under Linux. Yes, this is because Linux developers are somewhat hostile to closed source drivers. On the other hand, it is something that you should consider if you are using Linux.

      At the end of the day, the choice depends upon what

  • by UnknownSoldier (67820) on Friday June 13, 2014 @11:57AM (#47230843)

    The Truth on OpenGL Driver Quality [blogspot.hu]

    TL:DR;
    Vendor A nVidia - driver errs on the side of "make it work" vs GL spec
    Vendor B AMD - conforms to the OpenGL spec, but is buggy, inconsistent performance
    Vendor C Intel - best open source driver, but performance doesn't compete with nVidia or AMD

    Vendor A

    What most devs use because this vendor has the most capable GL devs in the industry and the best testing process. It's the "standard" driver, it's pretty fast, and when given the choice this vendor's driver devs choose sanity (to make things work) vs. absolute GL spec purity. Devs playing at home use this driver because it has the sexiest, most fun to play with extensions and GL support. Most of what you hear about the amazing things GL will be able to do in order to compete against D3D12/Mantle are by devs playing with this driver. Unfortunately, we can't just target this driver or we miss out on large amounts of market share.

    Even so, until Source1 was ported to Linux and Valve devs totally held the hands of this driver's devs they couldn't even update a buffer (via a Map or BufferSubData) the D3D9/11-style way without it constantly stalling the pipeline. We're talking "driver perf 101" stuff here, so it's not without its historical faults. Also, when you hit a bug in this driver it tends to just fall flat on its face and either crash the GPU or (on Windows) TDR your system. Still, it's a very reliable/solid driver.

    Vendor A supports a zillion extensions (some of them quite state of the art) that more or less work, but as soon as you start to use some of the most important ones you're off the driver's safe path and in a no man's land of crashing systems or TDR'ing at the slightest hickup.

    This vendor's tools historically completely suck, or only work for some period of time and then stop working, or only work if you beg the tools team for direct assistance. They have enormous, perhaps Dilbert-esque tools teams that do who knows what. Of course, these tools only work (when they do work) on their driver.

    This vendor is extremely savvy and strategic about embedding its devs directly into key game teams to make things happen. This is a double edged sword, because these devs will refuse to debug issues on other vendor's drivers, and they view GL only through the lens of how it's implemented by their driver. These embedded devs will purposely do things that they know are performant on their driver, with no idea how these things impact other drivers.

    Historically, this vendor will do things like internally replace entire shaders for key titles to make them perform better (sometimes much better). Most drivers probably do stuff like this occasionally, but this vendor will stop at nothing for performance. What does this mean to the PC game industry or graphics devs? It means you, as "Joe Graphics Developer", have little chance of achieving the same technical feats in your title (even if you use the exact same algorithms!) because you don't have an embedded vendor driver engineer working specifically on your title making sure the driver does exactly the right thing (using low-level optimized shaders) when your specific game or engine is running. It also means that, historically, some of the PC graphics legends you know about aren't quite as smart or capable as history paints them to be, because they had a lot of help.

    Vendor A is also jokingly known as the "Graphics Mafia". Be very careful if a dev from Vendor A gets embedded into your team. These guys are serious business.

    Vendor B

    A complete hodgepodge, inconsistent performance, very buggy, inconsistent regression testing, dysfunctional driver threading that is completely outside of the dev's official control. Unfortunately this vendor's GPU is pretty much standard and is quite capable hardware wise, so you can't ignore these guys even though as an organization they are i

    • by aliquis (678370)

      Also if you are a gamer why spend lots of money on a gaming PC and then live with shitty performance because you pick the open-source driver (even if it would be no more shitty than the AMD drivers)?

      AKA: If you play advanced games get an Nvidia card and run the proprietary drivers.

      Now if you don't play games do you really need a graphics card in the first place? Likely not. So get the integrated Intel or AMD graphics depending on your choice of processor. (And you could always leave AMD in the cold there to

  • by Anonymous Coward

    I had an AMD HD 6850 card that ran great on Windows, but could not run any game respectably in Linux. I was burned out waiting, so I bought an nvidia Geforce 750 ti, and now I can play games in Linux using the nvidia drivers from the website. This newer nvidia card is about the same performance as my old 6850 and it does not use any extra connectors from the power supply.

    • I was burned out waiting, so I bought an nvidia Geforce 750 ti, and now I can play games in Linux using the nvidia drivers from the website. This newer nvidia card is about the same performance as my old 6850

      Just getting facts straight: actually that NVIDIA card is 50% faster than your old AMD card. Still though, the GTX 750 Ti is a chip with reasonable price and fantastic performance/watt ratio, so congratulations on the upgrade. :)

  • I didn't go through every page so I might have missed it, but were there any tests done using the same game or benchmarks for both closed and open source drivers? It looked like the previous article was using a completely different set of games than this test.

    Anybody have links to actual apples to apples comparison? I'm using mostly amd cards for reasons that don't have anything to do with gaming but are opengl based. I'd like to get some idea just how far behind the open drivers are from the closed driv

    • Doing opensoure vs closedsource comparison has also being been done on a regular basis at phoronix.

      To sum things up:

      Current Mesa/Gallium3D stack is opengl 3.x only, proprietary drivers are 4.x (but work is being done, including by paid developers)

      AMD:
      except for the latest generation (where the opensource driver team is still debugging the support - but at least AMD does publish documentation and pays a few opensource developpers on their own, so I WILL EVENTUALLY end up supported), the opensource drivers ha

  • Open-source fans are encouraged to use AMD hardware on Linux while those just wanting the best performance and overall experience should see NVIDIA with their binary driver.

    You should definitely chose White. Or Black. Definitely.

  • This seems to always have been a "Chicken of the Egg" problem for Linux.
    We want major game titles to run on Linux, but vendors won't port because there isn't a large enough Linux user base; There isn't a large Linux user base because the quality of what is there is often inferior (dues to running in wine, bad/neglected drivers, etc) to Windows.

    • by BitZtream (692029)

      Steam has SEVERAL major games that run on Linux.

      • Of course there are a few here or there. The point I was trying to make was that the titles that are available cross-platform usually don't run as well on Linux as their windows counterpart due to the drivers.

        Companies don't put the money into better driver development for Linux because the user base isn't there.

      • by jxander (2605655)

        There are dozens of us!

        DOZENS!

  • These reviews are nice, but they always focus on gaming. There's very little information for media playback.

    How well do each of these drivers do with accelerated playback of MPEG2, MPEG4, and other formats? If given a 1080i source, can they produce a real 1080i stream to the display, or will the alternating fields get reversed? (I have an older CRT HDTV that is 1080i native. With newer displays, it's good to have the option of letting the display handle deinterlacing.)

    If I want to build a low-power medi

  • news at 11.
  • I'm pretty sure I could underclock my nvidia card to be as slow and shitty as the AMD card and the performance per watt would be comparable too.
  • At the very least, the AMD FOSS driver hasn't broken any systems for me. The Nouveau driver, however, has consistently booted up various systems with modes that didn't work on the display, causing it to blank shortly after booting or when starting X.

    I use a USB stick when dealing with client PC's. It's burned me enough times that I have memorized the need to put this on the kernel boot-line (basically, disable nouveau)
        nouveau.modeset=0

    • by JustNiz (692889)

      >> nouveau.modeset=0

      Thanks, this is great, just the information I've needed for a long while.

      I've never seen this parameter documented anywhere. How the heck did you find out about it?

  • I don't care if they are closed source or open source, as long as they are good.. I'd rather have excellent closed source drivers than crappy open source drivers...

The use of anthropomorphic terminology when dealing with computing systems is a symptom of professional immaturity. -- Edsger Dijkstra

Working...