• Vizra@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    11 months ago

    If this can help with Shader Compilation in games that would be amazing. I still can’t help but feel like I’m beta testing an architecture that’s over 1 year old though…

    Really frustrating from someone who came over from NVIDIA

  • kiffmet@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    11 months ago

    It’s kinda weird, but actually GCN up to and including RDNA2 did work scheduling in hardware, including automatic context switches in case of a bubble, hazard or stall.

    With RDNA3, this was changed to a compiler/driver based approach in favor of a simpler and smaller CU design. The tradeoff is increased driver complexity and higher CPU overhead - this is also how Nvidia has been handling things for ages now.

    HAGS seems to mitigate the overhead to some extent, this probably works through the HW providing hints to the OS about when there’s a good moment to switch tasks, such that the latter can set up memory transfers and other things needed slightly ahead of time and also doesn’t need to mindlessly wait for the GPU to finish whatever it’s doing, since in-flight Wavefronts aren’t preemptible.

    Anyhow, last time I looked at tests, the performance benefits were quite minimal (around 5% on a 2080ti in CPU limit)

    • Vizra@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      Gonna be honest… I didn’t understand a single word I just ready but word man that’s sick lol

  • Firefox72@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    11 months ago

    Great so now we have these beta drivers that don’t show up on the actual GPU page, the actual release branch that does show up and the Fluid Motion drivers which are another separate branch.

    I understand these will become the release version once the game ships but still. Just an unnecesary mess of different branches.

  • nawfhtx@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    11 months ago

    Anyone compared this to the 23.30.01.02 AFMF driver yet? It’s still been the best performance wise on my 7900xt, newer ones have had a negative effect on my clock speeds so I’m curious to see. Will check after work if not

    • AreYouAWiiizard@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      I have slightly lower performance watching videos in MPV with heavy shaders vs 23.30.01.03 AFMF driver but that one actually improved it vs 23.30.01.02 so I guess it’s about even. Idk about in games since I always ran them with Chill enabled.

    • theSurgeonOfDeath_@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      Did you enable hags in windows settings before running test?

      Just curious. I don’t expect gains i didn’t had much on nvidis too when I started using

  • Melodias3@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    11 months ago

    I believe it once hags lands into the next technical preview drivers and 23.12.1 or 24.1.1 whichever driver adds avatar frontiers of pandora optimizations first

    • I9Qnl@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      That used to happen often during 5000 series, you get HAGS one update and it’s immediately removed in the next one.

      • ms--lane@alien.topB
        link
        fedilink
        English
        arrow-up
        1
        ·
        11 months ago

        comment section on videocardz I suggest you go there

        I’d rather not.

        No one would. There isn’t anything to be gleaned there but the crappiest of insults and faboyism.

        • Opteron170@alien.topB
          link
          fedilink
          English
          arrow-up
          1
          ·
          11 months ago

          lol there is a direct answer to his question there last I checked you are not B1UE_H4WK

        • sequentious@alien.topB
          link
          fedilink
          English
          arrow-up
          1
          ·
          11 months ago

          It’s not even in the comments, it’s in TFA:

          AMD HAGS is not supported on Radeon RX 6000 series, and it appears that AMD has no plans to enable this feature. In 2021, AMD confirmed that HAGS does not provide expected performance improvements for RDNA2 GPUs, which is why it was omitted in the further driver releases.

          • rocketchatb@alien.topB
            link
            fedilink
            English
            arrow-up
            1
            ·
            11 months ago

            I believe in this context they were talking about the drivers from 2021. There’s a 39003 HAGS driver on Guru3D for 6000 series and people notice better frametimes.

    • FDSTCKS@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      I’m enjoying driver level frame generation on the 6800xt, it’s pretty awesome not gonna lie

      • lovely_sombrero@alien.topB
        link
        fedilink
        English
        arrow-up
        1
        ·
        11 months ago

        Some beta drivers enabled HAGS for the 6000 series and my 6800XT had improved performance in some cases (Horizon: ZD being one of them), more smooth frametimes (AC: Valhalla) or mostly just no improvement (everything else).

    • NewestAccount2023@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      HAGS is “hardware accelerated GPU scheduling” it’s not like antilag+ that was reaching into game engines from the outside and fiddling with them. It’s just replacing an older crossing guard with a newer one that’s usually faster

    • Evonos@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      Hags is a Windows feature.

      Nvidia got it since years and sometimes it increases performance minimally and sometimes it lowers it.

    • Opteron170@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      What does that have anything to do with Avatar: Frontiers of Pandora and HAGS?

      Seems like a totally off-topic question.

      • mayhem911@alien.topB
        link
        fedilink
        English
        arrow-up
        1
        ·
        11 months ago

        He’s asking a question. Is HAGS, a new to AMD tech, going to get VAC bans. Like the last time AMD introduced a new feature. Seems like a valid question, and its funny.

        • shalol@alien.topB
          link
          fedilink
          English
          arrow-up
          1
          ·
          11 months ago

          Seems like a snarky comment unrelated to a singleplayer game feature introduced in support by the game devs, not AMD alone, rather than a question they want an answer for, coming from a user named nvidia_rtx5000.

          • mayhem911@alien.topB
            link
            fedilink
            English
            arrow-up
            1
            ·
            11 months ago

            you know full well if this situation were reversed you’d think it was hilarious. I thought it was lol

    • Proliator@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      When running something like a game, a number of tasks need to be run on the GPU to render the scene. Scheduling of those tasks is typically done on the CPU by the display driver.

      HAGS offloads this scheduling to dedicated hardware on the GPU. In the case of AMD, it leverages a custom chip on the GPU designed to do that scheduling. Doing it right on the GPU reduces CPU overhead and can offer performance improvements.

    • RealThanny@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      In Windows, access to the GPU by multiple processes is scheduled, much like access to the CPU is. Without HAGS, all of this scheduling is performed by the CPU. With HAGS, some of the scheduling in particular situations can be offloaded to the GPU.

      It has nothing whatsoever to do with how work on the GPU is scheduled across the compute resources.

      It will have no measurable impact on gaming performance, and it’s not supposed to, if done correctly. It’s only going to potentially affect performance when multiple applications are trying to use the GPU at the same time to a significant degree.

      So if you’re trying to play a game at the same time as you’re using your GPU to render something, then HAGS might slightly reduce the overhead of both tasks sharing the GPU. You still won’t actually notice a difference.