In testing of OpenBenchmarking.org and preparations for the release of Phoronix Test Suite 3.0-Iveland at the end of February from SCALE, a lot of benchmarks have been happening to test the various analytical features and other new capabilities of this open benchmarking platform. In fact, it is really an overwhelming amount of benchmarks; the power capacity in my office is maxed out as benchmark after benchmark and system after system there is all sorts of test scenarios being looked upon. The benchmarks coming out on Phoronix.com over the past two months have just been barely scratching the surface of what has been going into the OpenBenchmarking.org system. Recently a lot of OpenCL compute benchmarks were pumped in, and since we have only published a few OpenCL Linux benchmarks -- OpenCL on Linux vs. Mac OS X and OpenCL NVIDIA vs. ATI on Linux -- here's some more in this article.
Now that the kernel mode-setting page-flipping for the ATI Radeon DRM kernel module has been merged into the Linux 2.6.38 kernel and the respective bits have been set in the xf86-video-ati DDX, we're in the process of running new open-source ATI graphics benchmarks under Linux. Our initial results (included in this article) show these latest improvements to cause some major performance boosts for the open-source ATI driver as it nears the level of performance of the proprietary Catalyst driver.
With our big AMD Linux GPU / driver comparison we found its open-source Gallium3D driver to be noticeably faster than the classic Mesa DRI driver across an array of Radeon hardware from multiple generations. However, the official Catalyst driver was multiple times faster (roughly 5.18x faster) than the Gallium3D driver, not to mention its lack of proper support for OpenGL 3/4, VA-API/VDPAU/XvBA video playback, and many other features only found within the proprietary Catalyst driver. Now though it is time to see how the Gallium3D Nouveau performance compares to that of NVIDIA's proprietary Linux driver across different GeForce graphics cards.
As was alluded to in our New Year greeting, we have been working on a massive graphics card / driver comparison under Linux. Beginning with ATI/AMD hardware, we have tested a series of graphics cards spanning the Radeon X1000, HD 2000, HD 3000, HD 4000, and HD 5000 generations using the very latest drivers. These drivers include the official Catalyst 10.12 Linux release as well as the very latest development code for the open-source Mesa and Gallium3D drivers. The results for seven ATI GPUs spanning four generations with three drivers are quite interesting.
Earlier this week we published our annual look at AMD's Catalyst driver releases from the past year. Not only did the Catalyst Linux driver this year picked up a couple new features, its driver performance had improved slightly over the past twelve months. In building up some initial test data for OpenBenchmarking.org we decided not only to do these tests on the latest consumer-grade graphics card this year, but expand it to cover the workstation performance too and to go back nearly two years in time. These results for an AMD FirePro V8700 graphics card with the monthly driver updates going back to Catalyst 9.2 are quite interesting. AMD announced twice this year optimizations to their FirePro driver software, but in reality these "optimizations" were largely unsustainable and not optimizations as much as they were attempting to address driver regressions from the past.
Earlier this month we delivered our annual performance look at NVIDIA's 2010 Linux graphics drivers and now the tables have turned to do our annual examination of the ATI/AMD Catalyst graphics drivers for the Radeon graphics processors. This was certainly an interesting year -- both good and bad -- for AMD with their Catalyst Linux driver.
At the end of each year for the past five years we have delivered "year in review" articles looking at the performance of NVIDIA's (and ATI/AMD's) proprietary Linux drivers. Both in terms of new features introduced during the year in their driver updates and benchmarking the driver releases to see how the performance has evolved over twelve months. With 2010 coming to an end, it is time for this year's driver reviews. We are starting this year seeing how the NVIDIA performance has matured in 2010.
Yesterday we shared benchmarks of the ATI R600 Gallium3D driver compared against the classic Mesa R600 driver and then the proprietary AMD Catalyst driver. The proprietary driver was much faster than the open-source drivers were, but the Gallium3D driver did possess higher performance in most of the tests than with the classic Mesa driver. This is similar to the R300 Gallium3D driver being faster than its now-deprecated R300 classic driver. Meanwhile though Intel continues to back only their classic Mesa DRI driver and there are no signs of them switching over to the Gallium3D architecture anytime soon. It is not as if Intel's current Mesa driver is feature-complete and performance-optimized as our tests from earlier this year show Intel's Linux graphics performance being far behind their Windows driver. In this article though we are seeing where the Intel Mesa performance is at when using the very latest DRM and Mesa code.
While we are still waiting on open-source support for the AMD Radeon HD 6000 series of graphics cards that were released last month, today AMD is releasing their initial open-source support for their Ontario hardware. AMD's Ontario is their low-powered Fusion processor designed for use in netbooks and other such devices. This dual-core chip with integrated Radeon HD 6250 graphics is only starting to ship now, but the open-source support for this first AMD Fusion chip is now available to Linux users, complete with 3D support.
While we have already published two exciting articles today looking at the native ZFS file-system for Linux and also new benchmarks of OpenSolaris / BSD / Linux, here's a third article for the day. We might as well test our new Phoronix serving infrastructure while already having excess load today due to Slashdot, etc (it's good practice for OpenBenchmarking.org), so here are benchmarks of AMD's newest Gallium3D driver compared to their classic open-source Mesa driver and also their proprietary Catalyst driver. Oh yeah, a fourth article is also in the queue for today or the very near future when AMD has a major Linux driver announcement to share.
In August we reviewed Apple's Enhanced OpenGL Stack that was introduced as an update to Mac OS X 10.6.4. This Snow Leopard Graphics Update 1.0 came out following Valve bringing Steam and its games to Mac OS X and collaboration via Apple, NVIDIA, and ATI/AMD to better the Mac OS X graphics stack with supporting new OpenGL extensions and making other optimizations, such as adding OpenGL occlusion queries support. While we tested this major Apple graphics update under a NVIDIA MCP79-based Mac Mini, which is a GPU that did not receive as many performance optimizations as those higher-end NVIDIA ASICs in newer Macs, the performance improvements were noticeable in some cases. In select games, however, Ubuntu Linux with NVIDIA's proprietary driver was still faster than Mac OS X 10.6.4 + SLGU. With Apple having released Mac OS X 10.6.5 this week that "addresses stability and performance of graphics applications and games," among other changes, we have carried out some new Mac OS X benchmarks seeing how the performance has changed, if at all.
After years of development work by Tungsten Graphics (now VMware) and the open-source community at large, the Gallium3D driver architecture is finally getting ready to really enter the spotlight of the normal Linux desktop user. With the recent Mesa 7.9 release, the open-source ATI developers switched their R300 driver (that supports up through R500 ASICs, the Radeon X1000 series) from the classic Mesa to their newer Gallium3D driver as the default choice. Vendors are now preparing to do the same as well within Fedora and other distributions, and it was just agreed upon this week Ubuntu 11.04 will use R300g. There will finally be a real, common hardware driver that is based upon Gallium3D and is used by mass amounts of people on a daily basis in a production environment.
Last month we carried out our fourth annual Linux Graphics Survey in which we sought feedback from the Linux community about the most common graphics drivers and hardware in use, what display/GPU-related features desktop users are most interested in, and collect other metrics to aide developers. Here are the results from this year's survey.
While LLVM 2.8 was just released, we have been curious to see how the latest Low-Level Virtual Machine compiler code affects the performance of the LLVMpipe driver. This is the Gallium3D graphics driver that lives in Mesa and leverages the unique modular LLVM compiler to efficiently handle processing the graphics rendering workload on a modern CPU as a much faster alternative to that of their legacy software rasterizer. To see how much of a performance impact - for better or worse - that LLVM 2.8 has on this open-source software driver we tested it when being built with LLVM 2.6, 2.7, and the 2.8 SVN code.
Not only have we been busy testing Mesa 7.9 with the Intel and ATI/AMD drivers along with the Gallium3D drivers (including LLVMpipe), but the Nouveau driver that continues to be developed by the open-source community for NVIDIA GPUs received a fresh round of tests too. Our first published benchmarks of the Nouveau Gallium3D driver were back in February when it was nearing a decent state in terms of supported features and stability. Its DRM also finally entered the mainline Linux kernel earlier this year thereby allowing many Linux distributions to now use the Nouveau KMS driver even though not many have yet adopted the Gallium3D driver for OpenGL acceleration. We delivered updated Gallium3D benchmarks in June with the latest Mesa code at that point, but since then there was the integration of a new GLSL compiler into Mesa and many Nouveau changes, so here are our most recent OpenGL benchmarks from this open-source NVIDIA driver.
As we have talked about in numerous articles now and delivered various benchmarks for different graphics processors from those using a classic Mesa DRI driver to the newer NVIDIA/ATI hardware with Gallium3D support, Mesa 7.9 brings a lot to the table. There are many new features to be found in Mesa 7.9 for all drivers, but in this article, we are specifically looking to see how the OpenGL performance of the classic R600 driver has changed compared to Mesa 7.7 and Mesa 7.8.
With Mesa 7.9 just around the corner and it sporting a new GLSL compiler, support for new OpenGL extensions, and months worth of other changes to core Mesa and its drivers, we decided to run some benchmarks of the latest Intel Arrandale graphics processor with the past few Mesa releases to see how the performance compares. We also have ATI and Nouveau Mesa benchmarks on the way.
Over the next few weeks there are a number of new Phoronix benchmarks to be published concerning the performance of Mesa 7.9 for both the Mesa classic and Gallium3D drivers from the different GPU vendors. Included in those tests will be new Intel Mesa benchmarks of their only officially supported 3D driver using one of the Arrandale processors, but for those currently missing out on the X Developers' Summit in Toulouse or PhoronixFest at Oktoberfest, here's a bonus article. For this extra round of benchmarking, we took one of the original Intel Atom benchmarks with i945 graphics and ran it with every major Mesa release since Mesa 7.4.
For the past three years we have hosted an annual Linux Graphics Survey in which we ask tens of thousands of users each time their video card preferences, driver information, and other questions about their view of the Linux graphics stack. This year we are hosting the survey once again to allow the development community to get a better understanding of the video hardware in use, what open-source and closed-source drivers are being used, and other relevant information that will help them and the Linux community.
AMD continues to abide by their commitment to provide open-source support for their graphics cards and as proof of that this afternoon they have released their initial hardware acceleration code that supports the ATI Radeon HD 5000 "Evergreen" family of consumer grade graphics processors. While this Evergreen support isn't yet finished and for the time being is targeted towards Linux developers and enthusiasts, you can now play around with your ATI Radeon HD 5000 graphics processor on an open-source driver while having 2D EXA, X-Video, and OpenGL acceleration.
With the imminent release of X.Org Server 1.9, last week we delivered benchmarks of Intel's 2D driver performance with X.Org Server 1.9. In those tests we found Intel's UXA (UMA Acceleration Architecture) performance only changed a bit -- for either better or worse -- with the updated X Server, but today we are looking at the 2D EXA performance using ATI Radeon hardware using this soon-to-be-released X Server.
X.Org Server 1.9 is set to be released as soon as next week, has already been pulled into Ubuntu 10.10, and is part of the X.Org 7.6 katamari. While X.Org Server 1.9 does not bring many exciting end-user changes like previously releases that introduced RandR 1.2, Multi-Pointer X / X Input 2.0, and other new technologies, there are plenty of bug fixes and other minor improvements throughout the X Server. In this article, we are looking at how the Intel DDX driver performance changes when upgrading from X.Org Server 1.8.2 to the latest X.Org Server 1.9 development code.
Last month we tested out Intel's new GLSL compiler for Mesa when running the ATI Radeon classic Mesa and Gallium3D drivers to see how this GL Shading Language compiler designed by Intel employee's for their hardware and open-source driver work for the other open-source drivers, since all of the Mesa drivers will be affected once this "GLSL2" compiler is merged into the Mesa code-base by month's end. The experience using Intel's new shader compiler with the ATI Radeon graphics driver worked fine except for Warsow where serious regressions were visible, but in the other games that are capable of running off Mesa, the experience was fine. What we have been curious to test since then with this new OpenGL shader compiler has been the LLVMpipe driver -- a Gallium3D driver we have been very excited about as it finally provides a better software rasterizer for Linux by leveraging Gallium3D and the Low-Level Virtual Machine (LLVM) compiler infrastructure for accelerating the Mesa state tracker atop a modern multi-core CPU that supports SSE4 instructions. We have now finished running tests of the Intel's GLSL2 branch with the most recent LLVMpipe driver code.
Whether you are an owner of an ATI FirePro V3800 that retails for just over $100 USD, the proud owner of an ATI FirePro V8800 that goes for over $1,300 USD, or any of the FirePro products in-between, you will want to update your graphics driver when AMD puts out their next stable software update. Back in March AMD put out an amazing FirePro Linux driver that increased the performance of their workstation graphics cards already on the market (and the other Evergreen-based workstation cards that entered the market soon after) by an astonishing amount. Our independent tests of this proprietary Linux driver update found that the performance in some workstation applications had increased by up to 59% by simply installing this updated driver while other OpenGL tests had just improved rather modestly with 20%+ gains. AMD though is preparing to release another driver update for Microsoft Windows and Linux that ups their workstation graphics performance even more! We have run some tests of this new beta driver against their older driver with both their low-end and ultra-high-end FirePro products and have found the improvements again to be astonishing.
With Intel developers earlier this week expressing their plans to merge their new GLSL compiler into Mesa by the end of next month, which besides providing various shader compiler optimizations and being a better framework going forward is already set to correct 50+ bugs, we decided to try out this Mesa "GLSL2" compiler. However, as Intel explicitly stated they haven't tested this new GL Shading Language compiler that's been in development for months with any other hardware drivers (or even Gallium3D) besides their own Intel DRI driver, we decided to see how well it works with the open-source Radeon classic and Gallium3D drivers. It ended up being both good and bad.
Last quarter we compared the Catalyst and Mesa driver performance using an ATI Radeon HD 4830 graphics card, compared the Gallium3D and classic Mesa drivers for ATI Radeon X1000 series hardware, and ultimately found that even with the ATI R500 class graphics cards the open-source driver is still playing catch-up to AMD's proprietary Catalyst Linux driver. In this article we have similar tests to show the performance disparity with ATI's much older R300 class hardware. Even with Radeon hardware that has had open-source support much longer, their drivers are not nearly as mature as an outdated Catalyst driver in the same configuration.
Two months ago we published our initial benchmarks of LLVMpipe, the Gallium3D driver that accelerated commands on the CPU rather than any GPU and unlike other Linux software rasterizers is much faster due to leveraging LLVM (the Low-Level Virtual Machine) on the back-end. Since then we have published new ATI Gallium3D driver benchmarks and yesterday put out Nouveau Gallium3D driver benchmarks, so today we are providing updated LLVMpipe driver results to show how well Gallium3D's LLVMpipe driver can accelerate your OpenGL games with a modern processor.
In recent weeks we have published a number of benchmarks showcasing the ATI Gallium3D driver that supports the R300-R500 graphics processors as this open-source driver has been maturing at such an exciting rate with impressive changes and measurable performance gains over a short period of time. This ATI Gallium3D driver in most instances is outperforming the classic Radeon Mesa driver that supports up through the ATI Radeon X1000 series graphics cards. However, how is the Nouveau driver maturing that supports NVIDIA's wide-range of GeForce graphics cards? In February we published some Nouveau Gallium3D benchmarks, but now we have a fresh set of numbers from three different NVIDIA graphics cards and we also compare the Nouveau Gallium3D driver to NVIDIA's proprietary Linux driver.
In this month's AMD Catalyst 10.6 driver update for Linux they rolled out the ATI 2D Acceleration Architecture, which pleased many ATI Radeon customers, but they aren't the only ones working towards improved 2D support. Intel's open-source engineers have been working to optimize their xf86-video-intel DDX driver 2D performance with much of this work being clearly shown in the Intel 2.12 X.Org driver update. Here are some benchmarks showing the significant performance gains brought by this open-source Intel driver.
It has been two years since the ATI Radeon HD 4800 (RV770) series launched so we have gone back since that monumental hardware launch and have re-tested each Catalyst driver release since then to see how the performance has changed for the ATI Radeon HD 4850 graphics card. The Catalyst driver has certainly matured over the course of two years in speeding up the OpenGL performance with this hardware along with bringing new features to their proprietary driver, but it is not exactly smooth sailing.
784 display drivers articles published on Phoronix.