Announcement

Collapse
No announcement yet.

Linux "RADV" Radeon Driver Gets A Big Speed-Up For 16-bit FidelityFX Super Resolution

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • dimko
    replied
    Originally posted by qarium View Post

    if you see the AI/DLSS hardware in nvidia hardware it is 4bit/8bit/16bit/32bit mix. (nvidia hardware can also do 64bi5 of course)
    amd RX580 only has 32bit and 64bit and my vega64 has 16bit and 32bit and 64bit

    it does not mean that the hardware represent 24bit color in only 16bit with less quality.

    No it means some part of the AI agloritmus need to be calculated but it does not need to be calculated at 32/64bit. some parameters in DLSS only need 4bit precision. of 4bit is enough why do it in 64bit ?

    and these vector/simd units are made in a way that you can do the double amounts of single calculations if you do 16bit instead of 32bit... if you only use 8bit you can do the double amount as 16bit
    and if dome calculation only need 4bit precision you can do the double amount than 8bit.

    just go back in history of the first intel cpu it was only 4bit... many calculations do not need more than 8bit.

    this "benefit of 35%" by using 16bit floading point really means it is used in a part of the algorytm who does not need higher precision and the result is the same quality with 35% higher performance.

    on the same hardware DLSS2.3 is 6% faster at the same result than FSR2.0 at 16bit FP because it use 4bit and 8bit calculation in moments no higher precision is needed.

    You have to unterstand the performance does not come out of a black-hole vaccum instead it comes from more calculations per second and on the same transistor amount you can only do this with lower precision like 4bit or 8 bit or 16bit. and in many moments in these AI and algoritms there is no need for higher precision.

    just imagine this: theoretical you would even get more performance if you add 2bit compute to with only 2 bit you can only represent 4 numbers but in many cases you do not need higher precision...

    2BIT=4 numbers
    4bit=16 numbers
    8bit=246 numbers
    and so one and so one. many calculations do not need more than this..
    This is why I love open source communities...

    Leave a comment:


  • qarium
    replied
    Originally posted by dimko View Post
    What is this "16-bit FSR" referring to?
    If I have screen/GPU that work on 24bit colour, does benefit of 35% in any way touch me?
    if you see the AI/DLSS hardware in nvidia hardware it is 4bit/8bit/16bit/32bit mix. (nvidia hardware can also do 64bi5 of course)
    amd RX580 only has 32bit and 64bit and my vega64 has 16bit and 32bit and 64bit

    it does not mean that the hardware represent 24bit color in only 16bit with less quality.

    No it means some part of the AI agloritmus need to be calculated but it does not need to be calculated at 32/64bit. some parameters in DLSS only need 4bit precision. of 4bit is enough why do it in 64bit ?

    and these vector/simd units are made in a way that you can do the double amounts of single calculations if you do 16bit instead of 32bit... if you only use 8bit you can do the double amount as 16bit
    and if dome calculation only need 4bit precision you can do the double amount than 8bit.

    just go back in history of the first intel cpu it was only 4bit... many calculations do not need more than 8bit.

    this "benefit of 35%" by using 16bit floading point really means it is used in a part of the algorytm who does not need higher precision and the result is the same quality with 35% higher performance.

    on the same hardware DLSS2.3 is 6% faster at the same result than FSR2.0 at 16bit FP because it use 4bit and 8bit calculation in moments no higher precision is needed.

    You have to unterstand the performance does not come out of a black-hole vaccum instead it comes from more calculations per second and on the same transistor amount you can only do this with lower precision like 4bit or 8 bit or 16bit. and in many moments in these AI and algoritms there is no need for higher precision.

    just imagine this: theoretical you would even get more performance if you add 2bit compute to with only 2 bit you can only represent 4 numbers but in many cases you do not need higher precision...

    2BIT=4 numbers
    4bit=16 numbers
    8bit=246 numbers
    and so one and so one. many calculations do not need more than this..

    Leave a comment:


  • dimko
    replied
    What is this "16-bit FSR" referring to?
    If I have screen/GPU that work on 24bit colour, does benefit of 35% in any way touch me?

    Leave a comment:


  • simburde
    replied
    I remember that initial Wine FSR patch had only 32-bit path implemented. Was it extended to support a 16-bit path later?

    Leave a comment:


  • treba
    replied
    Originally posted by shmerl View Post

    I wonder if some plugin for Wine Wayland driver can be added to use it, since it has an idea similar to "fullscreen hack".
    If I'm not mistaken it uses Wayland API to do that, the wp_viewporter protocol. So for that to work, Wayland compositors would need to implement FSR in their renderers. Could be a good idea, also for upscaling non-HIDPI apps.

    Leave a comment:


  • shmerl
    replied
    Originally posted by skeevy420 View Post

    In addition to Gamescope you can use Wine-GE
    I wonder if some plugin for Wine Wayland driver can be added to use it, since it has an idea similar to "fullscreen hack".

    Leave a comment:


  • skeevy420
    replied
    Originally posted by shmerl View Post
    What's a good way to use FSR for GOG games? Is there some set up guide?
    In addition to Gamescope you can use Wine-GE and use the following environment variables to enable it and set its strength. Strength varies between 0 (best) and 5 (worst), 2 is default.

    Code:
    WINE_FULLSCREEN_FSR=1 WINE_FULLSCREEN_FSR_STRENGTH=2
    It also has "WINE_FULLSCREEN_FAKE_CURRENT_RES=$SOME_RES" to define the resolution you're upscaling from. See the latest release notes for full details.

    Leave a comment:


  • JacekJagosz
    replied
    Originally posted by shmerl View Post
    What's a good way to use FSR for GOG games? Is there some set up guide?
    Also, now that FSR 2.0 is open source there should be game-specific mods popping up. It should be easy to replace a game's DLSS with FSR 2.0, according to this mod author: https://github.com/PotatoOfDoom/CyberFSR2
    It is noticeably better than 1.0, but it can't be enabled to just any game. So using gamescope is a great solution everywhere else.

    Leave a comment:


  • Quackdoc
    replied
    Originally posted by JacekJagosz View Post
    So it reduces the hit of FSR by ~35%. Really cool, but I thought the hit was really small already. So I wonder how much difference it actually makes.
    it all has a hit, FSR has two layers that can be surprising intensive, but still much less then native rendering, historically AMD 16bit preformance is pretty bad on radv, no idea what the patch does, but if it makes it better that is really nice.

    Originally posted by shmerl View Post
    What's a good way to use FSR for GOG games? Is there some set up guide?
    I use gamescope myself and it's not embedded into the game, no guide on it, but since FSR is post processing it's not an issue.

    it is it's quite simple to use, https://github.com/Plagman/gamescope#options

    Leave a comment:


  • jntesteves
    replied
    Originally posted by shmerl View Post
    What's a good way to use FSR for GOG games? Is there some set up guide?
    Code:
    gamescope -w 1280 -h 720 -U -- %command%
    https://github.com/Plagman/gamescope

    Leave a comment:

Working...
X