AIDA64 adds preliminary support for GV104 and GV104M and Hardware IDs

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Aug 7, 2018.

  1. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    38,529
    Likes Received:
    7,120
    GPU:
    AMD | NVIDIA
  2. cryohellinc

    cryohellinc Ancient Guru

    Messages:
    3,215
    Likes Received:
    2,448
    GPU:
    RX 5700 XT/GTX 1060
    Well, GV - V stands for Volta, so it won't be a refresh (that's great). And we know it will be 1100 series, not 2000.

    I guess that's finally some interesting bit of info as of late. Thank you @Hilbert Hagedoorn
     
  3. oxidized

    oxidized Master Guru

    Messages:
    233
    Likes Received:
    34
    GPU:
    GTX 1060 6G
    Good job Hilbert. And of course thanks to EL1TE.
     
    Last edited: Aug 7, 2018
  4. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    38,529
    Likes Received:
    7,120
    GPU:
    AMD | NVIDIA
    Props go out to EL1TE here on the forums for this spot.
     
    cryohellinc likes this.

  5. HardwareCaps

    HardwareCaps Master Guru

    Messages:
    452
    Likes Received:
    154
    GPU:
    x
    Volta was designed for AI with the tensor cores architecture. that won't be the case here. we know it is gonna use a refined 16nm process node and not a completely new one.
    Hopefully we will see a big leap like Pascal brought though I fear that won't be the case. also my guess would be higher pricing as well...
     
  6. cryohellinc

    cryohellinc Ancient Guru

    Messages:
    3,215
    Likes Received:
    2,448
    GPU:
    RX 5700 XT/GTX 1060
    I don't see why they wouldn't use the same tech, scaled down for the consumer market.
     
  7. HardwareCaps

    HardwareCaps Master Guru

    Messages:
    452
    Likes Received:
    154
    GPU:
    x
    Tensor cores are optimized for AI workloads. are you suggesting it would be used in consumer graphics as well? Have you seen any support for the tensor core in game engines or games that could suggest that?

    sounds like a waste of silicon for me. using that space for regular CUDA cores sounds like a better idea, maybe I'm wrong and Nvidia is planning to use tensor cores for its geforce series as well but it sounds very very fishy
     
    Last edited: Aug 7, 2018
    schmidtbag likes this.
  8. HardwareCaps

    HardwareCaps Master Guru

    Messages:
    452
    Likes Received:
    154
    GPU:
    x
    We should expect the expected which should be support for Ray-Tracing, new GDDR6 memory and small optimizations and improvements.
    nothing huge, anything more would be a surprise for me
     
  9. Denial

    Denial Ancient Guru

    Messages:
    12,830
    Likes Received:
    2,115
    GPU:
    EVGA 1080Ti
    Tensor cores are optimized for matrix operations - AI workloads utilize matrix operations but other things do as well. Also RTX is accelerated by Tensor - https://devblogs.nvidia.com/nvidia-optix-ray-tracing-powered-rtx/ but fwiw AMD has implemented it's own denoiser on mixed math FP16 - which next generation consumer Nvidia hardware might also get.
     
    cryohellinc likes this.
  10. HardwareCaps

    HardwareCaps Master Guru

    Messages:
    452
    Likes Received:
    154
    GPU:
    x
    I would be surprised to see tensor cores used for gaming[RTX I already suspected myself for support] maybe it is a refreshing architecture after all.
     

  11. Noisiv

    Noisiv Ancient Guru

    Messages:
    6,990
    Likes Received:
    674
    GPU:
    2070 Super
    Nvidia does not let anything go to waste.
    Even though Volta will be 15 months old by the time 1180 debuts, they're still gonna (re)use Volta SM (they claim 50% more efficient than Pascal). A bit disappointing that they're saving themselves for 7nm, but what can you do...

    It seems to me that the biggest challenge for Nvidia is not R&D, but how to use the existing IP to create the most optimum sequence of products and launches.
     
  12. HardwareCaps

    HardwareCaps Master Guru

    Messages:
    452
    Likes Received:
    154
    GPU:
    x
    I think you missread the 50%..
     

Share This Page