AMD Radeon AntiLag

[jorimt]

As I see many don't understand that a game engine can overwrite or use own value and ignore pre-rendering setting at the nvidia control panel.

Yep. It's up in the air what games respect the MPRF override from the NVCP, and even for the ones that do, what minimum each accept; it's possible some games could accept higher value overrides, like "3" or up, but reject lower value overrides, like "1".

Someone would simply have to test a wide range of games in every combination with a high speed setup or the like, and to my knowledge, that has yet to be done for MPRF specifically.

[Simon95]

As I see many don't understand that a game engine can overwrite or use own value and ignore pre-rendering setting at the nvidia control panel.

Yep. It's up in the air what games respect the MPRF override from the NVCP, and even for the ones that do, what minimum each accept; it's possible some games could accept higher value overrides, like "3" or up, but reject lower value overrides, like "1".

Someone would simply have to test a wide range of games in every combination with a high speed setup or the like, and to my knowledge, that has yet to be done for MPRF specifically.

Is it possible to check the amount of prerendered frames with GPUview?

[jorimt]

I did know about that program, but I don't know if it's capable of that. Maybe someone else does here?

What I would also like to point out though, is that the max pre-rendered frames setting is just that: "max."

If you set it to "4" (and the override is respected by the test game), for instance, you don't all of a sudden have a constant 4 pre-rendered frames in the queue.

Instead, it simply allows the system to pre-rendered "up to" 4 frames; the queue can be filled with anywhere between 1-4 pre-rendered frames at any given time depending on what the system deems necessary to keep the game running smoothly (which further complicates attempts at testing this consistently).

Also, the pre-rendered frames queue typically only applies in GPU-limited situations (when the GPU usage is maxed), which is why we recommend a frame cap for the lowest input lag where possible, since whenever the framerate is limited by an FPS cap (which also indirectly prevents max GPU usage), the pre-rendered frames queue effectively no longer applies (effectively at "0").

[open]

AMD talked about it in more detail on their youtube. It is indeed limiting the ammount of work that the cpu can que up. Should not have a huge impact in all situations but it's good to see these things focused on and welcome as a user option. IMO every millisecond is important.

[Shrib]

Radeon AntiLag has been released with the AMD 19.7.1 drivers, what do people think?

[masneb]

Sadly it doesn't seem that reviewers got the memo that input delay matters and no one tests it, they all reference the earlier initial response Nvidia made after AMD talked about it at e3, which is they think it's what pre-rendered frames does, which has now somehow solidified itself as fact without a source.

This is the only look at it and the article definitely could be flushed out better with better testing methodology:

https://www.pcgameshardware.de/Radeon-R ... i-1293817/

Doesn't seem to be pre-rendered frames from what I can tell, but there is a lot of other testing that needs to be done there.

[Lolitro]

So is good for a competitive mode? I got a rx570 and i got this mode. I play mainly csgo.

[Notty_PT]

Best thing is seeimg for yourself

Altho from what I seen so far is just pre rendered frames equivalent, wich I use for years on my systems.

[RealNC]

Sadly it doesn't seem that reviewers got the memo that input delay matters and no one tests it, they all reference the earlier initial response Nvidia made after AMD talked about it at e3, which is they think it's what pre-rendered frames does, which has now somehow solidified itself as fact without a source.

AMD didn't respond? If not, then that looks like silent acknowledgement. If anti-lag isn't about reducing pre-render queue size then AMD surely would have corrected nvidia.

[masneb]

New vlog/techish review. Still nothing with really good methodology, but better then nothing.

https://www.youtube.com/watch?v=6KXQvzKace8