This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Mali 628 floating point precision

Note: This was originally posted on 17th August 2013 at http://forums.arm.com

Could you please elaborate on a floating point precision of Mali-628?
What part of the IEEE754-2008 is supported in GPU? I.e. how many bits of precision we can get (medump) ?

Thank you in advance
Maxim

http://blogs.arm.com/multimedia/318-arm-mali-t604-new-gpu-architecture-for-highest-performance-flexibility/
Parents
  • Note: This was originally posted on 22nd August 2013 at http://forums.arm.com

    I'd suggest sticking to the Khronos precision minimums when picking qualifiers; it is very easy to write code which happens to work on one architecture which only works because that implementation provides more bits than technically necessary, which then fails on another device in weirdly wonderful and hard to debug ways.

    FWIW on all of the Mali GPUs:


    • highp is fp32
    • mediump is fp16
    • lowp is fp16
Reply
  • Note: This was originally posted on 22nd August 2013 at http://forums.arm.com

    I'd suggest sticking to the Khronos precision minimums when picking qualifiers; it is very easy to write code which happens to work on one architecture which only works because that implementation provides more bits than technically necessary, which then fails on another device in weirdly wonderful and hard to debug ways.

    FWIW on all of the Mali GPUs:


    • highp is fp32
    • mediump is fp16
    • lowp is fp16
Children
No data