Selling my Radeon VII GOLD edition for 2080ti FE?

Discussion in 'Videocards - AMD Radeon' started by AnnekeFloorever, Jun 17, 2019.

  1. AnnekeFloorever

    AnnekeFloorever Member

    Messages:
    10
    Likes Received:
    0
    GPU:
    HD7970
    I wanted to ask.

    What was the point of AMD putting 16GB of VRAM on the GPU?

    Did they have a good reason for this?

    The most I’ve seen a game use is 10GB VRAM on a card. And from the reviews I’m seeing, the amount of VRAM doesn’t even matter.

    Please correct me if I’m wrong.

    The other thing is the difference between the 1080ti and 2080ti is the memory type and RT cores. They have the same amount but the 2080ti is still faster than the 1080ti. Which automatically means VRAM means nothing. Right ?
     
  2. The_Amazing_X

    The_Amazing_X Master Guru

    Messages:
    367
    Likes Received:
    128
    GPU:
    Red Devil V64
    Wel it depends, you dont see any review with res scale to the max, I can tell you that far cry 5 with HD and res to 2 eats 8g thats at 1080p, now put that in 1440p and then 4k.

    The vram will go up and in the near future even the 6g card will be no good.
     
  3. AnnekeFloorever

    AnnekeFloorever Member

    Messages:
    10
    Likes Received:
    0
    GPU:
    HD7970
    That’s so weird, then why would Nvidia keep 11GB of VRAM on their latest enthusiast card after 2 years still ?

    Wouldn’t it make sense if they made 2080ti 16GB too...?
     
  4. The_Amazing_X

    The_Amazing_X Master Guru

    Messages:
    367
    Likes Received:
    128
    GPU:
    Red Devil V64
    Well going above 11g was expensive Gddr6 still new.

    Look at the price of the Titan with 24g
     

  5. Shadowxaero

    Shadowxaero Master Guru

    Messages:
    200
    Likes Received:
    21
    GPU:
    Radeon VII
    Memory bandwidth. More HBM stacks means more bandwidth, AMD needed 4 stacks of HBM2 to reach the required bandwidth that Vega 20 (Radeon VII) needs.

    Once you start using your card you will notice that Vega scales better with memory clocks more so than core. Had AMD only gone with say 8GB of HBM2 they would effectively be cutting the total bandwidth in half.

    You can't really compare GDDR to HBM, nor can you compare Pascal and Turing to GCN. Just different architectures that behave differently especially when it comes to vRAM. GCN has always favored huge memory bus widths for example.
     
    Last edited: Jun 20, 2019

Share This Page