Hardware-accelerated GPU scheduling - what is the trick to get this?

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by Teisco, Jun 25, 2020.

  1. PetritH

    PetritH Member Guru

    Messages:
    180
    Likes Received:
    19
    GPU:
    RX 7900 XT
    Do you still have to disable them via Properties - Compatibility tab? Or you just edit the registry, restart and you're set?
     
  2. Probably would have to disable all but two or one core and limit the frequency of your CPU
     
    angelgraves13 likes this.
  3. AsiJu

    AsiJu Ancient Guru

    Messages:
    8,938
    Likes Received:
    3,466
    GPU:
    KFA2 4070Ti EXG.v2
    Edit: meh...
     
    Last edited: Jul 15, 2020
  4. fry178

    fry178 Ancient Guru

    Messages:
    2,078
    Likes Received:
    379
    GPU:
    Aorus 2080S WB
    @hemla
    it makes no sense to even find out.
    all cards below performance of a 2080S or Ti WILL be gpu bottle necked.
    and the ppl that have a Ti, are usually NOT running a low end cpu, so for what user is it actually relevant to test this?
    its like testing a Ti @720p. irrelevant for real-world use comparison.
     

  5. I actually came from Commodores :D ... I was grown in a lab on a petri dish...

    EDIT: let me tell you about something called...
    [​IMG]
     
    fry178 likes this.
  6. hemla

    hemla Master Guru

    Messages:
    239
    Likes Received:
    27
    GPU:
    nvidia
    That is not true and not relevant to my question.

    If you want to test hardware professionally then you do it the way I described, by isolating components and testing them individually against various workloads in various conditions. So technically you would pick different applications that put different load on your PC components to see how HAGS operate in various bottleneck situations.

    And in this case "real-world" is just a meaningless buzzword, as most real-world scenario by your sense is fat 12-yr old kid playing GTAV and watching youtube. If something is slow then kid doesn't care about hardware specifics, he just asks his parents to get him faster computer.
     
  7. fry178

    fry178 Ancient Guru

    Messages:
    2,078
    Likes Received:
    379
    GPU:
    Aorus 2080S WB
    I doubt more than 10% own a 2080ti, thus are not really cpu bound, so how useful it is to know if there is or is not a difference, when "no one" actually experiences the bottleneck your testing for?

    Its like testing the full efficency of a car at its top speed. Sure, you would be able to tell that brand A is better than brand B when going +200mph, yet +90% will never reach that speed, not for long anyway), and usually ppl with the money to own +200mph cars, dont care if its getting 1 mpg, which makes that kind of test irrelevant for actual use.

    Same reason i found it stupid when sites were testing gpus above mid range at 720p (without testing 1080/1440p).
    Sure there are pvp ppl playing below 1080p to reduce lag, yet thats less than 10% of all gamers, thus making the test irrelevant for 90% of ppl.
     
    Last edited: Jul 15, 2020
  8. hemla

    hemla Master Guru

    Messages:
    239
    Likes Received:
    27
    GPU:
    nvidia
    You are correct with your calculations, I'd say even more than 90% of people. But the point of my question was to find out how HAGS act in CPU-bottlenecked situation. It's pure theory but could(or could not) produce some relevant data and is not related to "real-world" situations.
     
    fry178 likes this.
  9. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,040
    Likes Received:
    7,381
    GPU:
    GTX 1080ti
    cpu bottlenecks are gpu agnostic, you can get them on a 9700k and 1050ti if the game you're playing is a horrible single threaded opengl engine.

    just see how well Homeworld Remastered runs in AI vs Player combat with lots of ships
     
    BlindBison likes this.
  10. fry178

    fry178 Ancient Guru

    Messages:
    2,078
    Likes Received:
    379
    GPU:
    Aorus 2080S WB
    and thats the exception, not the rule, outside ppl unable to setup stuff and/or improperly matched hw (2080 on a 8y i7).
     

  11. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,040
    Likes Received:
    7,381
    GPU:
    GTX 1080ti
    Lol, no.
    It's very much a rule.
     
    BlindBison likes this.
  12. BlindBison

    BlindBison Ancient Guru

    Messages:
    2,419
    Likes Received:
    1,146
    GPU:
    RTX 3070
    @fry178 @Astyanax In some of Digital Foundry's GPU tests for the RTX Super line other than the 2080 Ti, they found that even at 1440p some stutter resulting from CPU limits still occurred in a variety of modern games (even ones that scale across multiple threads and on an 8700K iirc -- obviously this was worse at 1080p though). Only at 4K were they basically no longer running into CPU limits though to my recollection Kingdom Come Deliverance was still a mess pretty much in all test cases.

    I have a 3900X with a 2080S playing at 1440p and depending on the game, though "most of the time" i'm GPU limited, it still happens that I get CPU limited stuttering in some current gen games.
     
  13. fry178

    fry178 Ancient Guru

    Messages:
    2,078
    Likes Received:
    379
    GPU:
    Aorus 2080S WB
    Sure, but its neither "common for ppl to own say 2070S to play at 720p (outside some online/pvp), nor is everyone playing the same game as Astyanax is assuming (by naming one specific game), and its" sometimes" not "always".
    And i mean that in regards to the whole planet, as +60% are doing 1080p on something at/below 960/1060 etc, thus are, yes, gpu bound...
     
    Last edited: Jul 20, 2020
  14. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,040
    Likes Received:
    7,381
    GPU:
    GTX 1080ti
    You hit cpu limits regardless of resolution, it doesn't matter how new your stuff is, there are numerous cases where only faster clocks make a difference and those have limitations based in physics.
     

Share This Page