FFmpeg devs boast of up to 94x performance boost after implementing handwritten AVX-512 assembly code

Comments

• BrikoX@lemmy.zip ⁨1⁩ ⁨month⁩ ago

  Sadly, Intel takes another lose here.

  There is an issue, though: Intel disabled AVX-512 for its Core 12th, 13th, and 14th Generations of Core processors, leaving owners of these CPUs without them.

  source

  • NateSwift@lemmy.dbzer0.com ⁨1⁩ ⁨month⁩ ago

    Common Intel L recently. Shame it effects 12th gen

    source
  • cmnybo@discuss.tchncs.de ⁨1⁩ ⁨month⁩ ago

    I’m surprised Intel would remove a feature that AMD provides in their desktop CPUs.

    source

    • cubism_pitta@lemmy.world ⁨1⁩ ⁨month⁩ ago

      Then this is your reminder that ALL AMD CPUs are Unlocked and support overclocking…

      source

      • -> View More Comments
    • deegeese@sopuli.xyz ⁨1⁩ ⁨month⁩ ago

      Probably some BS market segmentation move.

      I imagine they noticed only certain server customers were using those extensions, so decided to limit them to high margin server SKUs.

      It would have been a smart move if there weren’t competitors putting that instruction in every CPU.

      source
  • shadow2@startrek.website ⁨1⁩ ⁨month⁩ ago

    My reactions: Ooooh… Awwww. 😮‍💨

    source
• Mike1576218@lemmy.ml ⁨1⁩ ⁨month⁩ ago

  Hand written assembly is pretty common in video, no matter what they say. All modern video codecs have hand written assembly for all modern SIMD extensions, even on ARM. They didn’t say anything about where these numbers come from. Likely compared against unoptimized C code. There will never be a case where having AVX-512 will give you that kind of speedup, because there will be fallbacks for more common extensions.

  source

  • xan1242@lemmy.dbzer0.com ⁨1⁩ ⁨month⁩ ago

    It’s mostly because AVX-512 doesn’t get used too well by compilers even today.

    However, what makes this impressive for me is that it is x86 after all. ARM is way easier to write assembly for.

    source

    • propter_hog@hexbear.net ⁨1⁩ ⁨month⁩ ago

      Word, major respect for writing x86 assembly. That shit’s trash.

      source
• cyborganism@lemmy.ca ⁨1⁩ ⁨month⁩ ago

  Holy shit.

  source
• AOCapitulator@hexbear.net ⁨1⁩ ⁨month⁩ ago

  Hand written is impressive? I thought all code was hand written until AI

  source

  • dev_null@lemmy.ml ⁨1⁩ ⁨month⁩ ago

    It’s handwritten assembly as opposed to bytecode generated by a compiler, from handwritten higher level language.

    source
  • BrikoX@lemmy.zip ⁨1⁩ ⁨month⁩ ago

    It’s about code optimization and efficiency. Most assembly code these days just relies on compilers for optimization as hand optimizing is extremely time-consuming work.

    source

    • propter_hog@hexbear.net ⁨1⁩ ⁨month⁩ ago

      To add to this, it’s also very easy and very likely to write assembly that has zero speedup or even significant slowdown versus what the compiler will write.

      source
  • possiblylinux127@lemmy.zip ⁨1⁩ ⁨month⁩ ago

    Well these days hardly anyone bothers with assembly

    source
• possiblylinux127@lemmy.zip ⁨1⁩ ⁨month⁩ ago

  Otherwise known as x86 64 bit assembly

  source