1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Anti-Lag Thread

Discussion in 'Videocards - AMD Radeon Drivers Section' started by Eastcoasthandle, Jul 10, 2019.

  1. HiAlgoBoost

    HiAlgoBoost Member

    Messages:
    21
    Likes Received:
    22
    GPU:
    Navi XT
    BTW - you don't have to make your game GPU-bound "on purpose". When CPU-bound - anti-lag or no anti-lag - you will be getting the optimal latency (no vsync, or vsync + enhanced sync - but best FreeSync). But if your game becomes GPU-bound at some point during gameplay - having anti-lag enabled will keep the lag at minimal, whereas without it you will get increased lag.
     
    Eastcoasthandle likes this.
  2. HiAlgoBoost

    HiAlgoBoost Member

    Messages:
    21
    Likes Received:
    22
    GPU:
    Navi XT
    There is an open-source tool called OCAT (https://gpuopen.com/gaming-product/ocat/) that shows a scrolling graph (only frametimes for now) - but it can be updated (opensource!) to show the latency graph as well...
     
    signex and Eastcoasthandle like this.
  3. sherpagoodness

    sherpagoodness Member

    Messages:
    10
    Likes Received:
    8
    GPU:
    Nitro64/Morpheus56
    holy christ man!

    i've not played pc2 much because it was so frustrating to try and drive, i chalked it up to physics being different or maybe that it was geared towards sim racers with a full wheel and pedal setup (esp after using a decent setup recently and experiencing forcefeedback just with forza 7 on console)

    but i just tried this with your settings and it is night and day difference, i cannot wait to try with my wmr headset.....i have to go adult and buy supplies at the store for the week first though....but seriously, god damn!
     
    HiAlgoBoost likes this.
  4. signex

    signex Ancient Guru

    Messages:
    8,685
    Likes Received:
    120
    GPU:
    Aorus RTX 2060 S
    Nice, when you activate it you hear a sound, still not a big difference, maybe it's cause my monitor already has such low latency.
    Anthem is the only game i have that struggles with FPS, but it seems they improved the overal smoothness of the game as 50fps looks incredibly smooth. :S
     
    spajdrik likes this.

  5. Alessio1989

    Alessio1989 Maha Guru

    Messages:
    1,345
    Likes Received:
    222
    GPU:
    .
    Not sure, I didn't test it since I do not have input lag problems in CSGO... However even if running in D3D9Ex mode the application need to use flip swap effect in the frame buffer queue, otherwise DWM/DXGI is not directly involved.
     
  6. OnnA

    OnnA Ancient Guru

    Messages:
    9,150
    Likes Received:
    1,830
    GPU:
    Vega 64 XTX LiQuiD
    Which one is better?
    "FlipQueueSize"=hex:31,00 (1 Frame) or Deleted Value.
    What happends when we del. this string (it's absent in driver BTW)

    Played BFV & ME:A no difference, same smooth experience :D
    I have now "FlipQueueSize"=hex:31,00 Deleted, no string.
     
    Last edited: Jul 17, 2019
  7. Alessio1989

    Alessio1989 Maha Guru

    Messages:
    1,345
    Likes Received:
    222
    GPU:
    .
    I am not sure what value really mean, since DXGI maximum queue allowed is 16 (0x10), 0x31 means 49.. Usually there is no real reason to use more then 3 frame buffer on a swapchain (some exception may be dual-gpu + vr o just an EVR rendering), but that value is outside legit bonds... So that probably means something-else in the engine (custom constant values or a bitmask)
     
  8. OnnA

    OnnA Ancient Guru

    Messages:
    9,150
    Likes Received:
    1,830
    GPU:
    Vega 64 XTX LiQuiD
    hex:31,00, 32,00 33,00 means in Radeon Reg. 1,2,3
     
  9. Eastcoasthandle

    Eastcoasthandle Ancient Guru

    Messages:
    2,112
    Likes Received:
    200
    GPU:
    R9 Fury
    OnnA did you disable Freesync?
     
  10. HiAlgoBoost

    HiAlgoBoost Member

    Messages:
    21
    Likes Received:
    22
    GPU:
    Navi XT
    Hex 0x31 is the ascii code for “1”... duh. 0x3100 is the Unicode for “1”.
     
    A2Razor and Kulagin like this.

  11. OnnA

    OnnA Ancient Guru

    Messages:
    9,150
    Likes Received:
    1,830
    GPU:
    Vega 64 XTX LiQuiD
    Never :p
     
    HiAlgoBoost likes this.
  12. Kulagin

    Kulagin Member

    Messages:
    43
    Likes Received:
    3
    GPU:
    Nvidia MSi
    Ok, so what should we set in games with locked FPS? For example, Tekken 7. It's locked on 60 FPS and we can't change it. Should we set flip queue size to 0x3000 in registry and enable RAL in AMD settings? FlipQueueSize 0x3000 and disable RAL? Some other combo of these 2 settings? To get the least amount of lag possible. Anything else you can recommend?
    My PC can sustain 60 FPS in Tekken all the time. CPU and GPU are used only by ~20%, so I shouldn't have any FPS drops no matter what settings I use.
     
    Last edited: Jul 17, 2019
  13. Kulagin

    Kulagin Member

    Messages:
    43
    Likes Received:
    3
    GPU:
    Nvidia MSi
    What do you mean it's absent in driver? You mean it's absent in AMD Settings manager?
     
  14. Alessio1989

    Alessio1989 Maha Guru

    Messages:
    1,345
    Likes Received:
    222
    GPU:
    .
    Why storing a UINT value as character value in a dword value? That's stupid... Moreover you need at least 2 buffers in a swap-chain.
     
  15. A2Razor

    A2Razor Master Guru

    Messages:
    477
    Likes Received:
    50
    GPU:
    Vega FE Liquid
    In the brief testing that I've done (old camera + LED clicker method), I've not found changing the flip-queue size to matter for games where Radeon Anti-Lag is working. (e.g, meaning the end latency was identical with a flip-queue of 1 or 3 set) I've also not found flip-queue size of 1 to beat Anti-Lag either.

    FPS limit should be whatever the maximum value the game allows is. FRTC, RTSS' limiter, etc, should be disabled where possible. I believe AMD disables FRTC when anti-lag is enabled, that said.

    -- If a game can really churn out frames and if freesync is active, then it should be capped under the monitor's limit of course.


    I usually do obscure things like this in development whenever I don't want people monkeying with the settings by hand. My hunch is it's the same thought process by AMD's driver-team.
     
    Last edited: Jul 17, 2019

  16. ClockClocker

    ClockClocker Master Guru

    Messages:
    408
    Likes Received:
    1
    GPU:
    970 gtx
    I tested it with Overwatch. There might be a small difference however that could be placebo. But especially when you use an old CPU (still on sandybridge) i notice that it draws on performance. I guess my cpu is just too slow. So far i'm more happy with it disabled.
     
    A2Razor likes this.
  17. A2Razor

    A2Razor Master Guru

    Messages:
    477
    Likes Received:
    50
    GPU:
    Vega FE Liquid
    Definitely, driver frame queueing, frame-pacing, etc, is there for a reason afterall. If FPS significantly drops, or stuttering becomes a problem ... that can be just as much a detriment to gameplay (in enjoyment or competitive gaming) as any increase in latency may be. Consistency and smooth gameplay is more important than anything, IMO. It's possible to compensate for static lag, yet it's not possible to compensate for erratic spikes of lag.

    In the end it's best to play around with it given it can be turned on and off on the fly.
     
  18. Eastcoasthandle

    Eastcoasthandle Ancient Guru

    Messages:
    2,112
    Likes Received:
    200
    GPU:
    R9 Fury
    So I'm not the only one then. Good to read you've noticed a difference. The reason why I found Project Cars 2 so important is that PC2 is very anti Radeon game. Even though they use AVX an equivalent Radeon card against nv card will show a pretty large deficit in FPS.

    Have you tried a 2 cycle weather race going from Cloudy to rainy with 20+ AI cars?
     
    Last edited: Jul 17, 2019
  19. BAZOKA1942

    BAZOKA1942 Member Guru

    Messages:
    171
    Likes Received:
    3
    GPU:
    sapphire r9 390
    so can you give us flip-queuesize 1 vs antilag results by latency ms numbers and which games tested
     
  20. Eastcoasthandle

    Eastcoasthandle Ancient Guru

    Messages:
    2,112
    Likes Received:
    200
    GPU:
    R9 Fury
    I tried what you've asked with a previous driver (19.5.1) to remove any variables that this driver may have with RAL on and off. But it doesn't matter as that was normally the case with any driver prior to 19.7.1 with my setup.

    There was a difference that I could notice. But I kept FQS at 1 as shown in my video. I didn't change it. The results are written in the OP.
     
    BAZOKA1942 likes this.

Share This Page