The Stutter is REAL: Driver, Game or MSI A/B Problem? (Now Corrected/Fix)

Discussion in 'Videocards - AMD Radeon Drivers Section' started by Eastcoasthandle, Feb 4, 2020.

  1. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    1944?
    What is the name of the file? Mine is called teedriver... The name itself makes me want to delete it.
     
  2. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,044
    Likes Received:
    7,380
    GPU:
    GTX 1080ti
    teedriver is legit.
     
  3. MerolaC

    MerolaC Ancient Guru

    Messages:
    4,385
    Likes Received:
    1,097
    GPU:
    AsRock RX 6700XT
    Eastcoasthandle likes this.
  4. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    Last edited: Feb 9, 2020

  5. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    I'm finally getting this card to act right :eek:
    2 matches of @ MW 1440p undervolted averaging 2000Mhz
    Those temps, Stable Frequencies, High Frame Rates, Smooth GamePlay...This is what Nitro is all about.


    [​IMG]
     
    SpajdrEX and MerolaC like this.
  6. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT


    Here is some game play now that everything is working like it should (1440p available shortly).


    [​IMG]

    Keep in mind that the above was provided while still using Relive to record the gameplay. :eek:
     
    PhazDelta likes this.
  7. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    New Update: Finally fixed frametime spikes in Hardware Monitor of MSI AB.

    I finally figured out why I was seeing frametime spikes in Modern Warfare using MSI Afterburner Hardware Monitor and not in OSD.
    It was due to a tweak: bcdedit /set useplatformtick yes
    This tweak was used to get the Current Timer resolution to .5 instead of .4996 or 1.0 instead of .9973 when using timer resolution app, etc. Although that did work in win10 I was unaware of it's negative effects on MSI A/B. Deleting it from bcdedit resolved the issue..

    What's also strange is that it didn't happen as bad in other games. So I'm not sure what the trigger is. Modern Warfare was the biggest offender of this tweak. In other games not so much.

    Keep in mind that framerates are capped at 120 FPS.
    [​IMG]
     
    Angantyr and Jackalito like this.
  8. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    17,564
    Likes Received:
    2,962
    GPU:
    XFX 7900XTX M'310
    Maybe some games are more sensitive to timing tweaks, not that I know too much about it but isn't the newer builds of Windows10 generally fairly good at utilizing this effectively long as the hardware itself supports it which I guess most of them now do. :D

    EDIT: Sure is a ton of supposed latency tweaks involving modifying the various timers.

    Sure are a lot of different timers too and other stuff. :p
    https://docs.microsoft.com/en-us/windows-hardware/drivers/devtest/bcdedit--set
     
    Eastcoasthandle likes this.
  9. Alessio1989

    Alessio1989 Ancient Guru

    Messages:
    2,959
    Likes Received:
    1,246
    GPU:
    .
    Thumb rule with OS timing: do not mess around with timers!
     
    Caesar and Krteq like this.
  10. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    To be clear, there was no stuttering issues in games. It only showed up in MSI Afterburner HW frametime.
    Which is demo'd 4 posts earlier in MW gameplay. Even in MSI AB's OSD there were no spikes. It's only in the Hardware Monitor it shows up.

    I more or less wanted to know why it was happening. I forgot I made that tweak and should have known in hind sight. As it was suppose to reduce frame rates in favor of latency.

    But as @JonasBeckman suggested it seems that even with the multiple timers being used in win10 seems to mange it well. However, when I use a timer resolution applications to clamp it to .5 or 1.0 the difference with it on vs off in MW is very specific. I can "snap to" a players head while on the move from a 90 degree angle with my cross hairs easier...is the best way to describe the difference.
     
    Last edited: Feb 19, 2020

  11. someonelikeyou

    someonelikeyou Member Guru

    Messages:
    132
    Likes Received:
    17
    GPU:
    RX 5500 XT
    Oh boy, HPET tweaks... Are you sure you want to go down this rabbit hole?
     
  12. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    LOL, I know right. I guess you can say I had the "experience".

    But it's funny though for the most part it does work but there are "side effects" that you won't know about until you "know about it". I still wonder if those spikes from MSI AB Hardware Monitor were actualy spikes in the OS or something else? Because it wasn't in the game.
     
    Smough likes this.
  13. Smough

    Smough Master Guru

    Messages:
    984
    Likes Received:
    303
    GPU:
    GTX 1660
    MSI Afterburner graph at some games shows absolutely crazy frametimes that never reflect at the games I play, I have no idea why it does that and never got a meaningful answer, so I just trust the OSD, which seems to be completely accurate.

    Just my 2 cents.
     
    Eastcoasthandle likes this.
  14. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    Have you tried disabling useplatformtick (reboot) to see if frametimes return to normal on those games that exhibit the issues?
     
  15. Smough

    Smough Master Guru

    Messages:
    984
    Likes Received:
    303
    GPU:
    GTX 1660
    I will try it. By the way, do you consider your games run smoother with useplatformtick disabled or enabled? What has been your experience?

    An even number seems more sensible to me, but no idea if just our ignorance pretending to know better than MS and believing we can do better when we maybe are just making thing worse.
     

  16. Smough

    Smough Master Guru

    Messages:
    984
    Likes Received:
    303
    GPU:
    GTX 1660
    Crap. O.S spikes do seem far more worrying than any spike inside a game. Its such a shame that there isn't a lot of info on what useplatformtick really does and what it affects and what it doesn't affects. Googling doesn't yield a lot of results.

    Hmmmm, seems its time to investigate this even more.
     
  17. Eastcoasthandle

    Eastcoasthandle Guest

    Messages:
    3,365
    Likes Received:
    727
    GPU:
    Nitro 5700 XT
    In game no. But LatencyMon did show anomalies for win10.
    I haven't noticed any issues since disabling it. It seems that Hardware Monitor was showing some sort of spikes that weren't "in the game" but something else. So, I simply disabled it and haven't looked back. But remember, I'm on 1903.
     
  18. Smough

    Smough Master Guru

    Messages:
    984
    Likes Received:
    303
    GPU:
    GTX 1660
    So I tried the system without useplatformtick and now Afterburner shows perfect frametimes on the graph; so yeah, changing this setting messes the graphs, no idea why, but if you ask me, that looks like a bad sign: https://imgur.com/a/7AfUmhm

    That's from Deus Ex: Mankind Divided, the small spike at the end its because I saved and exited the game, but the rest of the graph is perfectly flat, with useplatformtick on, the graph shows CRAZY jumps for nothing, even if the game felt fine.

    Since I've been using bcdedit /set useplatformtick yes I haven't noticed any major benefits, at anything. I just was using that because it makes timer resolution an "even number" and because its "recommended". Coming to think about it, I think it may have done things worse for some games, so I will just do some more testing with the normal setting, which is having it disabled.

    Also, with useplatformtick, the game has some weird movement judder at random times, with the Windows stock setting seems to be fine, however, this could be placebo, so take my words with a grain of salt.

    I will do further testing by having my system with bcdedit /set useplatformtick yes and setting the Timer Resolution to 0.5 to see the behavior and see how games perform, if I see any improvements, I will let you know. If I don't see great results, i'll just stick to bcdedit /deletevalue useplatformtick. After all, that's how Windows is.

    If you need to see results from Latencymon from me, let me know.
     
    Eastcoasthandle likes this.
  19. gabsvm

    gabsvm Member Guru

    Messages:
    146
    Likes Received:
    26
    GPU:
    RX 580 8gb
    Thank you so much guys, reading this explained why i was getting random spikes in frametimes and stuttering every match in Warzone, i was thinking 8gb ram was the issue until now. Deleted this tweak bcdedit /set useplatformtick yes and played a match perfectly fine. W10 1709 btw.
     
  20. gabsvm

    gabsvm Member Guru

    Messages:
    146
    Likes Received:
    26
    GPU:
    RX 580 8gb
    Btw.. is recommended having memory compression enabled? I only have 8gb ram and disabled superfetch when i got my ssd. Some people say is better, some say is worst. Is fine having it enabled for memory compression to work?
     

Share This Page