Search Unity

  1. Unity 6 Preview is now available. To find out what's new, have a look at our Unity 6 Preview blog post.
    Dismiss Notice
  2. Unity is excited to announce that we will be collaborating with TheXPlace for a summer game jam from June 13 - June 19. Learn more.
    Dismiss Notice
  3. Dismiss Notice

HDRP GPU Big Overhead

Discussion in 'Graphics Experimental Previews' started by Dinamytes, Nov 26, 2019.

  1. Dinamytes

    Dinamytes

    Joined:
    Nov 3, 2016
    Posts:
    51
    Why does hdrp has a big overhead and is overall slow?

    230fps without anything - gtx 750 ti
     
  2. AcidArrow

    AcidArrow

    Joined:
    May 20, 2010
    Posts:
    11,989
    Please check in builds, checking in editor is useless.

    Also performance of an empty scene is also of limited usefulness.

    With all that said HDRP is supposed to have some overhead but scales really well when you throw lots of stuff at it, so...
     
    hippocoder likes this.
  3. Dinamytes

    Dinamytes

    Joined:
    Nov 3, 2016
    Posts:
    51
    1-In builds it always gives me lower FPS using 1F / Time.unscaledDeltaTime.
    2-You are right but why would use so much gpu without even a camera?
    3-I hope so :) i don't have a good scene to test.
     
  4. hippocoder

    hippocoder

    Digital Ape

    Joined:
    Apr 11, 2010
    Posts:
    29,723
    Bandwidth is the cap and the 750ti has none to speak of. So this says it'll top out at 200 ish for that GPU. But under severe load it doesn't slow down as rapidly as built-in would.

    Sad these common posts popping up from people with blank scenes. It's very frustrating because the blank scene "test" never meant anything, never did mean anything, and never will. It's utterly meaningless.

    I can see why people imagine things but they really don't know what is happening under the hood or have much experience where the bottlenecks are (usually it'll be bandwidth). HDRP renders several things ahead of time, even on a blank screen, so when you want to push it with serious heavy workloads, it can carry it when built in will fall over and die, basically.

    The real test is a game under load, and it's here HDRP proves to be much faster, for much better (AAA) visual quality. I also know a high end premium indie developer working on HDRP with 750 ti right now actually, and they are consistently shocked at how well it runs. It completely sailed past their previous built-in effort.

    Of course they target monitor refresh rate like any sensible dev will. Not numbers that will never be used. You cannot imagine 500FPS or 400FPS as actually being wiggle room. That's 100% incorrect since FPS is not a linear measurement!

    "My application was running at 900FPS, then I added rendering of .... and my frame rate dropped to 450FPS. Why is this feature so slow?? Why is it cutting my performance in half?!??"

    Answer: http://www.mvps.org/directx/articles/fps_versus_frame_time.htm


    Use ms and you realise a very different story. It only takes a small amount of ms to dramatically change 500FPS to 200FPS.

    Sadly even an artificial test like lots of spheres and post effects still don't really push it. To push it you need a lot of lights, materials, mostly different but same shader, with SRP batching on, volumetric lights, shadows set up properly for performance etc, using all the great features such as low resolution transparencies and more.

    It is difficult to test so the blank scene thing is often tried, and I can't blame people for doing that, but it's not indicative...
     
    Dinamytes and AcidArrow like this.
  5. Dinamytes

    Dinamytes

    Joined:
    Nov 3, 2016
    Posts:
    51
    This is the answer I was looking for. Thank you.