GeForce 456.71 WHQL drivers: download & discussion

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by Hilbert Hagedoorn, Oct 7, 2020.

  1. ReaperXL2

    ReaperXL2 Member Guru

    Messages:
    179
    Likes Received:
    69
    GPU:
    EVGA 2080ti Hybrid
    OK peeps, I have posted a couple of times regarding my 1080ti being stuck at 1556mhz at desktop idle using this or any of the 456.38 drivers onwards. Today I discovered a very confusing thing. If I disable HAGS my 1080ti will not downclock lower than 1556mhz idle on the desktop. You will find rocking horse sh*t before my card will go any lower (regardless of 456.38 onwards). However, if I enable HAGS my card suddenly wants to behave rationally & down clocks to the correct 139mhz @ desktop idle. What the hell is this?

    I have tested this with over 30 different windows power/NVCP settings/Inspector settings/with GFE/without GFE and the behaviour is consistent, regardless of settings...

    Anybody at all got any ideas what's going on here? Is this how HAGS is designed to improve min frames? by holding the gpu at a higher clock in preperation for a load demand that may or may not come or is this an Nvidia thing?

    Don't mind leaving HAGS on as NVidia say they fixed their sh*t & I dont seem to be experiencing any pain from its enablement but is this really the way they fixed the perf regression with HAGS?
     
  2. wintermutex

    wintermutex Member

    Messages:
    38
    Likes Received:
    1
    GPU:
    EVGA 3080 XC3 Ultra

    ^^ I have had a similar issue on every driver since 30** support. It looks like setting "max performance" no longer allows clocks to settle back to 2d values in desktop reliably. For now I just set to Optimal, but I maintain that this is not the appropriate fix and that you should be able to achieve 2d idle clocks in desktop with "max performance" flag set. I have HAGS enabled and still see this behavior. Sometimes, it will play nice and settle into 2d clocks regardless of HAGS setting, but it's rare. And no.. I am not running anything in my desktop idle that has any gpu utilization, it's <1%
     
  3. Obsidian_Oasis

    Obsidian_Oasis Active Member

    Messages:
    57
    Likes Received:
    21
    GPU:
    ASUS RTX 2070s
    Sad to say but I knew this would happen with these cards. So many unsure results with these RTX 30 series cards from Bin 0 overly stock units that most consumers are purchasing to questionable drivers.

    I'm definitely glad that I'm waiting for Oct. 28.

    Refund if you can and wait for better yields or...
     
  4. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    17,564
    Likes Received:
    2,961
    GPU:
    XFX 7900XTX M'310
    Could just be something with Vulkan or how No Man's Sky utilizes Vulkan, GTA V and Red Dead 2 in particular which offers both Vulkan and D3D12 being stable would indicate some issue with the game.
    Best to do is probably report it to NVIDIA and see if they can't replicate it after doing the usual like updating the Vulkan loader/runtime and clearing the games local cache of shaders.

    https://vulkan.lunarg.com/sdk/home
    (Not the full SDK just the barely a megabyte 1.2.154.1 runtime itself.)

    Game and driver implementation and support is important but the runtime gets it's fixes too and not always for just the SDK part.

    Guessing the current driver runtime and supported Vulkan API is 1.2.140 somewhere now so it should be pretty robust overall but there's still some fixes in the newer versions outside of newer SDK and extensions getting supported. :D
    What the game is doing with the API for No Man's Sky is something the developers would have to look into or NVIDIA possibly together with the developers if they can replicate issues with the game in their test environment confirming it as a software error.


    EDIT: Though if KX11's issue is in general and not just with this one game unlike some earlier posts specifically about No Man's Sky then yeah that's different.


    EDIT: Hardware's a possibility of course, I'm on a 5700 type GPU myself after all it didn't quite work out as smoothly as I was hoping.

    Detroit Became Human and Division 2 from posts in the NVIDIA GPU section.
    Hmm well Detroit is Vulkan too but Division 2 is D3D11/D3D12 here so that's not just something with Vulkan then.

    Speaking of GPU power and that stuff wonder if Afterburner would work to trim the clock speeds or power draw down seeing how the GPU operates at a lower value if there's a issue with that.

    GeForce Experience and the experimental features as a alternative dragging the sliders to (slightly) reduce power draw and such seeing if that changes anything.
    That would just work around whatever the problem itself is assuming it works not actually fix the problem itself.
     
    Last edited: Oct 13, 2020

  5. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,016
    Likes Received:
    7,355
    GPU:
    GTX 1080ti
    https://www.nomanssky.com/2020/10/origins-patch-3-03/

    might be the game itself.

    certain utilities will lock the clocks while the gpu is using HAGS

    1.2.154 wait, no, thats the vulkan dev driver.

    1.2.142
     
    JonasBeckman likes this.
  6. windrunnerxj

    windrunnerxj Master Guru

    Messages:
    487
    Likes Received:
    127
    GPU:
    RTX 4060
    No such issue on my 1066 with HAGS on or off, must be something exclusive to your card or setup.
    The only HAGS issue I have is random rare driver crashing outside of games. At least better than a few drivers before (I had it disabled for a while prior to this driver) where crashing was very frequent (like a few times a day) and would straight up BSOD the entire system even during gaming.
     
    Last edited: Oct 13, 2020
  7. ReaperXL2

    ReaperXL2 Member Guru

    Messages:
    179
    Likes Received:
    69
    GPU:
    EVGA 2080ti Hybrid
    I would think that the locked clock would apply with HAGS 'On' not HAGS 'Off'. HAGS 'Off' & the card should behave based on bios/drivers exclusively without Win10 interference. I have uninstalled all 3rd party apps, utils etc. including NVInspector/AB/RTSS/Aida64 monitoring & cleaned the crap left behind. End result is HAGS 'On'/'Off' is the only game changer (@Astyanax ;)). I must also stress that this behaviour is only happening since 456.38 onwards.

    Due to the way this issue seems to apply itself, I'm inclined to think the issue lies within Win10 not Nvidia. A clean install of Win10 is on the cards I believe. However with HAGS 'On' the card behaves correctly & I'm not getting a perf hit in PhysX games so not a big deal atm.

    Just wanted to know if this problem was apparent to anyone else & if it is by design or a bug!
     
    Last edited: Oct 14, 2020
  8. AntiSnipe

    AntiSnipe Master Guru

    Messages:
    442
    Likes Received:
    147
    GPU:
    Radeon RX 6600
    Tried these for a few days. On the new GTX 1660 Super, same as my old 1060 3GB, getting lots of the error below + blank white screen in Firefox with no error shown...again. Back to 451.85hf

    Event ID: 4101
    System: Display driver nvlddmkm stopped responding and has successfully recovered.
     
    Last edited: Oct 15, 2020
  9. X7007

    X7007 Ancient Guru

    Messages:
    1,874
    Likes Received:
    72
    GPU:
    ZOTAC 4090 EXT AMP
    is HAGS On fixed? No lower performance/reduction?
     
  10. BingoBob

    BingoBob Member

    Messages:
    17
    Likes Received:
    8
    GPU:
    ASUS ROG 4090 OC
    Is the 2080ti performing well with this driver? With the release of the 30 series I don't want my card to start slowing down.
     

  11. garyb

    garyb Master Guru

    Messages:
    255
    Likes Received:
    78
    GPU:
    Gigabyte RTX 4070
    This also happens with Microsoft edge(new version) and chrome. I have tested both and they have the same issue, try disabling hardware acceleration seems to stop this.

    Is this a driver issue ManuelG
     
  12. Nastya

    Nastya Member Guru

    Messages:
    185
    Likes Received:
    86
    GPU:
    GB 4090 Gaming OC
    Your card won't slow down. In fact, I'm getting a tiny performance increase in synthetics with this driver compared to 446.14. Might just be margin of error, though.
     
    BingoBob and fear029 like this.
  13. BingoBob

    BingoBob Member

    Messages:
    17
    Likes Received:
    8
    GPU:
    ASUS ROG 4090 OC
    No harm in trying I guess.
     
  14. vmanuelgm

    vmanuelgm Active Member

    Messages:
    83
    Likes Received:
    20
    GPU:
    Giga 4090 Master

    Have u checked the temperatures on the backside of the pcb (capacitors under the core, memories and backside of the VRM's)???

    They get pretty hot after some time gaming, specially if not cooled properly.
     
  15. Assirra

    Assirra Active Member

    Messages:
    73
    Likes Received:
    1
    GPU:
    Asus Strix 980 4GB
    So this might be a weird question but do you have the "hacked" version of RTX voice by any chance?
    What i found a couple days ago was my card (1080TI Strix) refusing to downclock while RTX voice was running. After i closed it it instantly went to actual idle mode.
     

  16. ReaperXL2

    ReaperXL2 Member Guru

    Messages:
    179
    Likes Received:
    69
    GPU:
    EVGA 2080ti Hybrid
    I do have RTX voice installed. It is the official version provided by NVidia, now patched (officially) to support 10 series gpu's. You have however provided some food for thought with your question. Seeing as RTX voice utilises the GPU for it's learning capabilities it would not be unreasonable to surmise that whilst running in the background continuously, it would cause the GPU to up-clock to handle the demand on its various cores (AI algorythms) even when not under any rendering load.

    I will test RTX voice On/Off, HAGS On/Off to flesh this out further & report my findings.

    Thanx for the suggestion buddy :cool:
     
  17. ManuelG

    ManuelG NVIDIA Rep

    Messages:
    1,121
    Likes Received:
    620
    GPU:
    Geforce RTX 2080 FE
    There are people here who are not seeing this problem:
    https://forums.evga.com/driver-crash-in-No-Man39s-Sky-3080-with-4567145638-drivers-m3116107.aspx

    Likely something very specific to your configuration. I have not received a driver feedback nor GeForce Experience feedback for this issue. Can you share anymore regression information you have tried? Can you also share your graphics settings used in game?
     
  18. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    17,564
    Likes Received:
    2,961
    GPU:
    XFX 7900XTX M'310
    Wonder if other monitoring utilities could be interfering, non-standard techniques for hooking certain API's or trying to read hardware Corsair iCUE or Asus AI Suite though individual GPU's will perform differently so that's a thing too and then there's other software and system settings like just testing without XMP seeing if it's some irregularity with RAM like what this GPU I have is overly sensitive towards. :D

    Troubleshooting without a good complement of log files for running processes and crash info like memory dumps or similar is down to a lot of testing though, earlier drivers might be stable too but could just be making a existing issue more prevalent separate from the driver software itself and hardware wise well something like OCCT might perhaps pick up some errors and that could pinpoint a more hardware related errors but it could also be system or software in general it can't really know after all.

    VLK and D3D12 pushing the hardware more overall too compared to D3D11 games like Shadow of the Tomb Raider or other games on these API's being really sensitive to even milder overclocks, guessing benchmarks using these could also be good but then driver profiles for these would likely be tuned and optimized well plus the benchmark itself might be stable compared to some game or game engine pushing RAM or VRAM in different ways like what Tomb Raider here often hits problems with.

    Not much of a solution or anything just the common suggestions and ideas if even that, hard to troubleshoot effectively after all.
    Driver issue you'd ideally have everyone or everyone on a specific GPU hit a issue and that would confirm it but it's not always that easy.
    (This particular GPU I'm using and a certain type of crash issue that's still being investigated using that as a example again ha ha.)


    EDIT: Well thread safety and questionable or weird ways of hooking or reading from these API's and sensor data and other problems, Fraps, various video recording software, Corsair, Asus, Razer, others and so on. Some of it's good other stuff is compatibility nightmare.

    One reason log files and such helps, knowing what crashed and how and optionally also what was running giving an idea of what to test against or what to try and shut down and test if that resolves it.
    (Again more like standard suggestions more than anything, guessing Ampere compatibility could also be a thing for some of these suites trying to get readings on the sensors here as another possibility.)
     
  19. fleggy

    fleggy Member Guru

    Messages:
    142
    Likes Received:
    9
    GPU:
    GB 4090 Gaming OC
    Hi Manuel,

    I had the same problem after upgrade from 2080ti to Gigabyte 3090 Gaming OC. Lowering the Power Limit to 95% solved this. No problems using stock setting in other games so far.
     
  20. Derko1

    Derko1 Guest

    Messages:
    408
    Likes Received:
    0
    GPU:
    2x1080s
    I checked all pages and no one has brought it up, but what is the SLI profile for Baldur's Gate 3? I can't find anything online about it and it isn't working. So maybe it was a profile to disable it?
     

Share This Page