No announcement yet.

New ARM X.Org Driver Promises Better Performance

  • Filter
  • Time
  • Show
Clear All
new posts

  • New ARM X.Org Driver Promises Better Performance

    Phoronix: New ARM X.Org Driver Promises Better Performance

    While xf86-video-sunxifb sounds like an old X.Org driver from the Sun Microsystems days for some obscure SPARC system, this driver is a fork of the xf86-video-mali DDX driver. What makes this ARM X.Org graphics driver interesting is that it promises better performance on the Allwinner A10/A13 SoC compared to the ARM vendor's official driver...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Like FLOSS.

    It make possible for some random programmer to wonder why something is not working the right way, and then to test and check it, and then come up with proper way.

    And rubbing nose of big company like ARM? Priceless

    Now, if only my BIOS was FLOSS....


    • #3
      I'm not quite a random programmer. Having spent a bit of effort optimizing ARM NEON code in pixman, trying to get the best out of it and treating even something like 10% improvement as a rather good speedup, I find it very disturbing that the modern drivers for ARM hardware can easily regress 2D performance by up to 10x factor and nobody cares. I think, I have a fairy good understanding about how ARM hardware can perform doing software rendering. So when some random guy may think "graphics is slow because it's a weak ARM hardware" or "it is not enough hardware accelerated yet", my reaction would be "wow, something is *really* broken here"

      Also I tried to make it clear that xf86-video-mali is not even advertised as a complete driver by the vendor. It is a kind of example code, which needs to be integrated with the display controller to make a complete driver. The problem is that the users and distro packagers may have (and sometimes do have) a wrong idea about what xf86-video-mali really is.