1. Forum
  2. USENET
  3. comp.lang.asm.x86

  • Copying low 16 bytes of YMM register to the high 16 bytes

    From Robert Prins@robert@nospicedham.prino.org to comp.lang.asm.x86 on Sun Nov 12 22:50:17 2023
    From Newsgroup: comp.lang.asm.x86

    Or is there a way to immediately load a full YMM register with two copies of of
    a 16 byte variable?

    I've been looking at the various PSHUFxxx instructions, but they seem to stare me in the face like Edgar Allan Poe's "Stolen letter"...

    Thanks,

    Robert
    --
    Robert AH Prins
    robert(a)prino(d)org
    The hitchhiking grandfather - https://prino.neocities.org/
    Some REXX code for use on z/OS - https://prino.neocities.org/zOS/zOS-Tools.html

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Branimir Maksimovic@branimir.maksimovic@nospicedham.icloud.com to comp.lang.asm.x86 on Mon Nov 13 02:59:34 2023
    From Newsgroup: comp.lang.asm.x86

    On 2023-11-12, Robert Prins <robert@nospicedham.prino.org> wrote:
    Or is there a way to immediately load a full YMM register with two copies of of
    a 16 byte variable?

    I've been looking at the various PSHUFxxx instructions, but they seem to stare
    me in the face like Edgar Allan Poe's "Stolen letter"...

    Thanks,

    Robert
    VBROADCASTI128 is instuction you need.
    --

    7-77-777, Evil Sinner! https://www.linkedin.com/in/branimir-maksimovic-6762bbaa/

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Robert Prins@robert@nospicedham.prino.org to comp.lang.asm.x86 on Mon Nov 13 12:18:02 2023
    From Newsgroup: comp.lang.asm.x86

    On 2023-11-13 02:59, Branimir Maksimovic wrote:
    On 2023-11-12, Robert Prins <robert@nospicedham.prino.org> wrote:
    Or is there a way to immediately load a full YMM register with two copies of
    of a 16 byte variable?

    I've been looking at the various PSHUFxxx instructions, but they seem to
    stare me in the face like Edgar Allan Poe's "Stolen letter"...

    VBROADCASTI128 is instruction you need.

    Thanks! ...but given that my program must also run on an AMD Bulldozer CPU, I'm
    SOL, I'll just have to double the data I'm loading. C'est la vie.

    Robert
    --
    Robert AH Prins
    robert(a)prino(d)org
    The hitchhiking grandfather - https://prino.neocities.org/
    Some REXX code for use on z/OS - https://prino.neocities.org/zOS/zOS-Tools.html

    --- Synchronet 3.21d-Linux NewsLink 1.2