Intel SNA vs. Modesetting GLAMOR - DDX Benchmarks

Following all of the Intel 3D graphics tests this week for DRM-Next code for Linux 4.7, Mesa 11.3-devel, and more, there's been a few readers requesting a fresh xf86-video-intel vs. xf86-video-modesetting comparison.

The Intel DDX driver continues to be what nearly all Linux distributions use as the default X.Org driver when Intel hardware is detected, but xf86-video-modesetting can alternatively be used thanks to it universally targeting DRM/KMS drivers. The Intel driver continues to be focused on its performant but huge SNA acceleration architecture that's written for each generation of Intel graphics while the xf86-video-modesetting driver just uses GLAMOR to accelerate 2D over OpenGL in a generic manner. With xf86-video-intel 3.0 still not happening and that release cycle being dragged on for years while SNA being a big maintenance burden, some have questioned whether xf86-video-modesetting is a better future for Intel hardware.

Similar to my recent Radeon vs. Modesetting DDX comparison, I ran an Intel vs. Modesetting DDX comparison on a Haswell and Broadwell laptop/ultrabook.


Changing out the DDX driver on this Intel hardware didn't have much of an impact on the 3D/OpenGL tests, contrary to what some Phoronix readers have been reporting...



For 2D, the Intel DDX driver with its custom SNA architecture was faster than the generic xf86-video-modesetting with GLAMOR.

You can dig through this Friday benchmarking fun via this OpenBenchmarking.org result file. Via the forums you can share your own experiences with the Intel and Modesetting X.Org drivers.