Nvidia GeForce 364.47 WHQL driver download & discussion

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by maur0, Mar 7, 2016.

  1. RzrTrek

    RzrTrek Ancient Guru

    Messages:
    2,527
    Likes Received:
    722
    GPU:
    RX 580 ♥ MESA 21.1
    On my second restart my screen went all black; will try if 364.51 solves it.

    Edit: problem solved, installed 364.51 and it's working again.
     
    Last edited: Mar 8, 2016
  2. Denial

    Denial Ancient Guru

    Messages:
    13,528
    Likes Received:
    3,073
    GPU:
    EVGA RTX 3080
    Nvidia cards were always lesser equipped than their AMD equivalents though. People just ignored it because they only seem to care about benchmark numbers instead of the actual hardware.

    Take a 680 vs a 7970. Both cards were meant to compete. The 680 is a 2.5Tflop 2GB card vs the 7970's 3.7Tflops 3GB. At launch they were competitive, now the 7970 crushes the 680 in recent titles, why? Probably because it had ~50% more performance that was completely untapped.

    All the games coming out in the last year or so are also the first games/engines developed entirely on the new generation of consoles. All those consoles have GCN. All those programmers are now learning how to extract every piece of performance out of GCN. AMD's PC cards are benefiting from it. What's amusing is that people act like this is surprising, but there were entire threads dedicated to it, especially when 680/7970 came out, that talked about the long term performance.

    Is Nvidia putting Kepler/Maxwell on the backseat? Probably. But I think the vast majority of the performance differences we are seeing is simply from GCN finally being fully utilized. AMD had a long term strategy, with both its architecture and mantle and it's starting to pay off for them.
     
  3. alanm

    alanm Ancient Guru

    Messages:
    10,561
    Likes Received:
    2,687
    GPU:
    Asus 2080 Dual OC
    ^ Yep. And my future continuity with Nvidia is now in question.
     
  4. Denial

    Denial Ancient Guru

    Messages:
    13,528
    Likes Received:
    3,073
    GPU:
    EVGA RTX 3080
    Yeah, I mean I can see how it would turn people off. For me it didn't really matter as I upgraded my 690 just as the 2GB started becoming an issue and I'm upgrading my 980 this year.. most likely to Polaris. So with both cards I basically got better performance than their AMD equivs during the time they were out. Anyway, I like what AMD are doing with their driver software and I have a feeling their cards are just going to be better overall this generation due to their early push with HBM and direction with DX12 and stuff. But anything can happen, need to wait till more info is available.
     
    Last edited: Mar 8, 2016

  5. -Tj-

    -Tj- Ancient Guru

    Messages:
    17,237
    Likes Received:
    1,958
    GPU:
    Zotac GTX980Ti OC
    Well 290X still had higher shader peak Tflops as e.g. 780TI or 980GTX, but big Maxwell kinda changed that vs FuryX. Although most of the time when we see a shift in NV favor by 780TI & 980GTX vs 290/390X its because of higher raster Gtris peak performance.

    Although FuryX has theoretical max shader Tflops ~ 8.6, rasterization is much lower ~ 4.2 Gtris.


    [​IMG]
    http://techreport.com/review/28513/amd-radeon-r9-fury-x-graphics-card-reviewed/4



    Polaris full chip apparently has 10-12Tflops, e.g. 980TI @ 1400MHz has ~8.4Tflops, that's why I doubt we won't see such a drastic boost GM200 vs full Pascal. Mid-range Pascal GP104? will be probably be just a little higher then 980GTX..
    Imo Volta will have a bigger perf. difference, I will upgrade then, Pascal looks just a test phase atm.
     
  6. Denial

    Denial Ancient Guru

    Messages:
    13,528
    Likes Received:
    3,073
    GPU:
    EVGA RTX 3080
    Yeah, I mean it really gets so complicated that it's really not possible to definitively say one thing or another.

    Like for example, people are comparing the 770 to the 370 because technically the 770 is a 680 and the 370 is a 7850 fine. But look at the a 770 vs a 370 in Luxmark:

    http://www.hardwareluxx.com/index.p...nd-inno3d-gtx-950-and-msi-r7-370.html?start=8

    The 370 destroys the 770 by more than double. Now consider that more and more games are utilizing directcompute in games to offload CPU stuff onto the GPU -- especially in consoles due to the relatively weak jaguar cores.

    If a game is going to heavily make use of compute, it's going to bottleneck the 770 way worse than the 370, regardless to the tflop count, memory bandwidth, etc.
     
  7. LimitbreakOr

    LimitbreakOr Master Guru

    Messages:
    392
    Likes Received:
    16
    GPU:
    2070 Super
    Thanks denial, i often don't agree with your point of view, but you're spot on here and i give you a +1 for it.
     
  8. Arctucas

    Arctucas Ancient Guru

    Messages:
    2,137
    Likes Received:
    44
    GPU:
    eVGA RTX2080 FTW3
    So... I just installed the 364.51 BETA.

    Only UDP connections were initiated by the installer.

    nVIDIA is likely taking a page from the Microsoft playbook; spy on your customers.
     
  9. dr_rus

    dr_rus Ancient Guru

    Messages:
    3,075
    Likes Received:
    436
    GPU:
    RTX 3080
    Consider as well that Luxmark has nothing to do with DirectCompute as it's a raytracer running via OpenCL and this result is more relevant for the assessment of the state of OpenCL API performance on NV's h/w in general - where CUDA is obviously a better choice for a number of reasons.

    If you look at whatever DirectCompute benchmarks we have, it's not nearly as dramatic as you make it sound because contrary to a popular believe in-game compute doesn't really happen in pure peak flops numbers either.

    Your suggestion that a heavily compute oriented workload would somehow bottleneck a 770 way more than a 370 doesn't make any sense either as 370 has less than 2/3rds of peak compute performance of 770. If you expect to see something akin to Luxmark results in games anytime soon - prepare to be heavily disappointed as even in an AMD sponsored D3D12 game with a build running on unoptimized drivers on NV's side 680 (which isn't really a 770 as 770 is faster) doesn't perform anywhere close to 370.

    In all your assumptions above the only one which is definitely true is the assumption on GCN being "the default h/w" for most optimizations these days because it is what consoles are running on. This is the main reason we're seeing more games which favor GCN over Kepler/Maxwell, and this usually means that the engine of such game simply wasn't optimized for NV h/w at all. Because when it is properly optimized it usually looks like this - spot the 370 and 770 there:

    [​IMG]
     
    Last edited: Mar 8, 2016
  10. LimitbreakOr

    LimitbreakOr Master Guru

    Messages:
    392
    Likes Received:
    16
    GPU:
    2070 Super

    Or better yet, let's look at a very recent benchmark....

    [​IMG]

    As you can see, anything non Maxwell is suffering greatly compared to the competitors of their time. 290 vs 780 Ti? or 370 vs 770? I would be very sour if i purchased the 780ti instead of a budget R9 290 (non x).
     

  11. X7007

    X7007 Ancient Guru

    Messages:
    1,674
    Likes Received:
    42
    GPU:
    Sapphire 6900XT
    Can someone tell me why my Clock is becoming on IDLE max 540 or 640 or 708 max when IDLE ? if overclocking or not overclocking it still happens.

    I used DDU.

    Windows 10 x64

    What could effect it ? if it was overclock or something else using Nvidia Inspector to Apply button on the overclocking button wouldn't return the clocks back. they just stuck till I restart or clicking on the Apply button again and it goes back to full speed.

    No one know what can cause the issue EXPECT Overclocking ? which is not the case because it still happens.
     
  12. LimitbreakOr

    LimitbreakOr Master Guru

    Messages:
    392
    Likes Received:
    16
    GPU:
    2070 Super
    that usually happens if you have more than one monitor. is that it?
     
  13. dr_rus

    dr_rus Ancient Guru

    Messages:
    3,075
    Likes Received:
    436
    GPU:
    RTX 3080
    I can see that anything from NV is suffering in this engine be it Maxwell or Kepler which is a clear sign of a bad engine optimization for NV's h/w.
     
  14. khanmein

    khanmein Ancient Guru

    Messages:
    1,647
    Likes Received:
    72
    GPU:
    EVGA GTX 1070 SC
    ipong = nv while android = amd

    amd always released higher spec which is really worthwhile but their s/w side really :puke2::puke2::bang:

    go buy AMD don't make noise..
     
  15. dabrooks

    dabrooks Master Guru

    Messages:
    351
    Likes Received:
    0
    GPU:
    SLI EVGA GTX 980Ti Gaming
    What is your monitor refresh rate? 120Hz and above will cause that.
     

  16. telc

    telc Member

    Messages:
    16
    Likes Received:
    0
    GPU:
    2 X MSI LIGHTNING 980Ti
    Still having issues with these Beta ones. Black screen freeze after a few minutes. Did manage to get The Division going for a little while till it started stuttering at 10fps. I did see a frame rate improvement over 362 when it did work.

    Back to 362!!! sigh

    Running 980ti SLI on Predator X34.
     
  17. dsbig

    dsbig Ancient Guru

    Messages:
    4,056
    Likes Received:
    43
    GPU:
    EVGA RTX 2070
    :3eyes:

    I havent had any problems here with windows 7 and the drivers.
     
  18. Vidik

    Vidik Master Guru

    Messages:
    606
    Likes Received:
    196
    GPU:
    MSI 1070 Gaming Z
    Shouldn't this thread be merged with 364.51 discussion? 364.47 doesn't even exist anymore.
     
  19. CK the Greek

    CK the Greek Maha Guru

    Messages:
    1,298
    Likes Received:
    17
    GPU:
    GB 2060S,Gsync,3DV2
    It should be locked.
     
  20. darrensimmons

    darrensimmons Ancient Guru

    Messages:
    1,637
    Likes Received:
    70
    GPU:
    GIGABYTE RTX3090 OC
    wrong thread.
     

Share This Page