Intel Sandy Bridge Acceleration On Non-SNB Hardware

Written by Michael Larabel in Display Drivers on 8 September 2011 at 02:00 PM EDT. Page 1 of 6. 8 Comments.

Last week benchmarks were published of Intel's New Sandy Bridge Acceleration architecture (SNA) that showed several performance improvements for 2D and 3D, but the new acceleration architecture still wasn't mature with a few regressions compared to the normal UXA back-end. While the focus of this SNA support is on speeding up operations for Sandy Bridge (SNB) and forthcoming Ivy Bridge (IVB) hardware, SNA is supported for older Intel graphics processors too. Here are some benchmarks of the Sandy Bridge New Acceleration architecture when using the Ironlake and Gen3 back-ends.

The back-ends for Sandy Bridge New Acceleration include Ivy Bridge, Sandy Bridge, Ironlake, Broadwater, Gen3, and Gen2. For none of these generations is SNA the default, but when building the xf86-video-intel DDX, the --enable-sna argument must be passed. For the older hardware support with SNA, it is still not regression free. There are a few performance regressions but also other problems, namely there is issues with rendering window borders at times; see the screenshots below.

Worth noting that after last week's SNA performance tests, Intel's Chris Wilson explored the test results and made two commits (9a563ea and 32fc0c8) to make performance improvements to the Gen6 / Sandy Bridge back-ends based upon the Phoronix results. For this benchmarking was done using xf86-video-intel DDX Git as of 5 August (695e711). Mesa 7.12-devel Git was also as of Monday morning.

The Ironlake testing was done from an ASRock HM55-HT NetTop with an Intel Core i3 330M "Arrandale" processor with 4GB of RAM. The Gen3 back-end testing was from an MSI NetTop with an Intel Atom 330 and Intel 945 graphics with 1GB of RAM. Ubuntu 11.04 x86_64 was the base operating system. Besides the Intel DDX and Mesa from Git, the Linux 3.1 kernel was also used from Git. The stock X.Org Server (1.10.1) was used, although an SNA patch for it is known to increase performance even greater.

All testing was done with stock setting aside from toggling the SNA support. A variety of 2D and 3D benchmarks -- mostly the same ones as last week's SNA testing on Sandy Bridge -- was run via the Phoronix Test Suite.


Related Articles