SLI AMD 480 performance.worth than 1080?

Discussion in 'Videocards - AMD Radeon' started by sunnyp_343, Jun 1, 2016.

Thread Status:
Not open for further replies.
  1. pharma

    pharma Ancient Guru

    Messages:
    1,682
    Likes Received:
    491
    GPU:
    Asus Strix GTX 1080
    Best performance can be seen on the Hall of Fame:
    http://www.3dmark.com/hall-of-fame-2/fire+strike+3dmark+score+performance+preset/version+1.1/1+gpu

    For GTX 1080 I expect to see much higher scores once the custom cards become available. Your comparison is above is LN2 vs air.
     
    Last edited: Jun 4, 2016
  2. Aelders

    Aelders Banned

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
    It's a fair comparison if your end goal is make the 1080 look like a POS :D
     
  3. eclap

    eclap Banned

    Messages:
    31,495
    Likes Received:
    3
    GPU:
    Palit GR 1080 2000/11000
    I still don't understand why they are so obsessed with beating the 1080. 480 is a mid range card, they should be showing off how it compares against it's competitors, not the 1080.
     
  4. sammarbella

    sammarbella Ancient Guru

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    They can't compare it yet with the Nvidia product it will compete (1060) because is still not in the market or even paper launched.

    So AMD decided to show his 480 as a good deal in CFX(!?!) against a high end GPU (1080) in a somewhat bogus slideshow. :3eyes:

    I guess a proper (and logic) single GPU vs single GPU will be available between Vega and 1080 on Vega launch...when 1080Ti will beat both. :D
     

  5. Aelders

    Aelders Banned

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
  6. the9quad

    the9quad Master Guru

    Messages:
    855
    Likes Received:
    4
    GPU:
    1080ti MSI Gaming-X
    I feel sorry for anyone buying two of these thinking AMD is going to provide quality crossfire support. They better get used to, " X game does Y while in AMD Crossfire, AMD recommends disabling Crossfire in game X" in every single release note.

    Or get used to developer responses like, "AMD is aware of the problem, and we are working close with them to resolve the issue." Followed by a hot fix with the above patch notes.

    My advice to anyone thinking of two of these, would be to spend the extra $300 and just get a 1080.
     
    Last edited: Jun 4, 2016
  7. The Mac

    The Mac Ancient Guru

    Messages:
    4,404
    Likes Received:
    0
    GPU:
    Sapphire R9-290 Vapor-X
    or $400 less for the rumored rx490...
     
  8. Fox2232

    Fox2232 Ancient Guru

    Messages:
    11,803
    Likes Received:
    3,357
    GPU:
    6900XT+AW@240Hz
    What people did not get is that nVidia positioned those new GPUs in way that their sales of older generation will not suffer.
    While AMD intends to deliver around r9-390(x) performance for less than 2/3rds of price.

    Two RX-480 may or may not be competitive solution against GTX 1080. It really does not matter since not many even think about buying CF/SLi.

    But those benchmarks did show something. CF RX-480 delivered 6.5% higher fps than GTX 1080. While AMD PR tried to make it look like GPUs were actually not utilized, they were. But how much? My guess is that average utilization for both GPUs was around 85%, but let's presume it was 90% and that there was not much potential untapped.

    So, if fully used one of those RX-480 would deliver around 59% (1.065 / 0.9 / 2) of GTX 1080 performance while costing $200.
    GTX 1070 delivers 80% of GTX 1080 performance while costing $400.
    And that makes RX-480 ~25% slower than GTX 1070 while costing 1/2.

    Then we have economical impact. Cards priced as GTX 1080 are practically invisible in steam stats, not enough people buy them. GTX 1070 will make enough of people to think, so it will end up being in steam stats as it will get some sales.
    But $200 card with performance similar to previous generation priced ~ $350... This is where people will spend their money.

    And that original comparison of GTX 1080 vs 2x RX-480... Think about 2x RX-480 vs GTX 1070. Price about same. Worst case scenario RX-480 delivers 25% lower fps. Best case scenario CF RX-480 delivers 150% performance of GTX 1070.

    And thing is, CF usually does not work in some small indie project which does not need a lot of GPU power anyway. GPU intensive games rarely end up without CF/SLi support as developer himself realizes that maximum details are not exactly playable on single GPU.
     
  9. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,668
    Likes Received:
    591
    GPU:
    Inno3D RTX 3090
    There is a rumor about that? If it's so expensive it should thoroughly spank the 1070, and that means Vega. Unless they put a genie inside every driver package there is no way that Polaris could seriously pull that off. Sometimes I just wish my f*cking card was working, I wouldn't get anything now.
     
  10. sammarbella

    sammarbella Ancient Guru

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    Best scenario?Worst scenario?

    What we know for true is: AOTS is an AMD showcase, a huge effort was made on it (and in AMD drivers..) to show AMD performance in DX12 and how well AMD GPUs work on it.

    AOTS is the best scenario for AMD to bench their GPU: the game and drivers are tailor made for them.

    I think you extract too much conclusions from a flawed VS: wrong Nvidia driver version, different game version, wrong GPU usage (average) in the case of AMD GPUs.

    I can believe a 480 is around the performance of a 390X but we can't be sure of it's real performance from what AMD PR showed us until serious sites review the GPU (Guru3D, Anand,....)

    NDA lifts on 29...and the card will be on shelves on 29.

    Who is going to pre-order them on blind faith?
     

  11. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,668
    Likes Received:
    591
    GPU:
    Inno3D RTX 3090
    Only idiots.
     
  12. Fox2232

    Fox2232 Ancient Guru

    Messages:
    11,803
    Likes Received:
    3,357
    GPU:
    6900XT+AW@240Hz
    You know SP/TMU count and that it has reasonably higher clock than r9-390x. It will be very close in performance. That's one thing I am sure about. But final clock will decide of it is below or above it. Tho, I expect it to be faster in certain scenarios even if clock is not high enough to compensate for less SPs/TMUs.
     
  13. sammarbella

    sammarbella Ancient Guru

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    I have no problem to agree with you on that: As i said i can believe that is possible (around 390X performance).

    What i don't believe is in the AMD PR slide we see: it was flawed in many ways and we can't contrast it to reliable third party benches until NDA lifts on 29.
     
  14. Fox2232

    Fox2232 Ancient Guru

    Messages:
    11,803
    Likes Received:
    3,357
    GPU:
    6900XT+AW@240Hz
    They definitely did not make approximations easier :)
     
  15. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,668
    Likes Received:
    591
    GPU:
    Inno3D RTX 3090
    We are again in the same trap as the NVIDIA guys were with Pascal. They believed it would be different than an overclocked Maxwell. We have to see actual numbers. If the card performs better than a 390, then AMD has actually improved their IPC.
     

  16. sammarbella

    sammarbella Ancient Guru

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    I'm sure AMD improved his GPU IPC in Polaris.

    Maybe it's true that Pascal is nothing more than Maxwell "on steroids" (?!¿) but his higher base core clocks and achievable OC compensate and exceed any performance gained in AMD Polaris IPC.

    And i fall myself in the original AMD Polaris/Pascal trap set by AMD PR in the Polaris paper launch: we can't really compare 1080/1070 to Polaris cause they are in different price/performance segments. :D

    Comparing two red apples to a green watermelon is not fair.
     
  17. OnnA

    OnnA Ancient Guru

    Messages:
    12,810
    Likes Received:
    3,196
    GPU:
    Vega XTX LiQuiD
    All you can do is run:
    1100MHz in 3DMark:
    Fury-X and Maxwell and Pascal and then you'll see true IPC of those GPU's
    I Bet Fury will beat them with Large Margin :)
    But this is only "Test" and GPUs runs their own clocks regardless :grad:
     
  18. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,668
    Likes Received:
    591
    GPU:
    Inno3D RTX 3090
    That's not even a "trap". You can't compare a $199 card with a $799 one. If tradition holds, the 1060 will be a POS (and I can't see how it won't be, with 1500+ shader processors, no matter the frequency).
     
  19. sammarbella

    sammarbella Ancient Guru

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    Call it "mirage", "distraction" or "ilusion/delusion".

    Make your choice betwen them or a different one.

    We must do something to fill the time and Polaris threads until June 29. :D
     
  20. nateluthje

    nateluthje Master Guru

    Messages:
    266
    Likes Received:
    3
    GPU:
    Asus strix 1080 SLI
    I haven't had much trouble with xfire except in "The division" where windowed 1440p disables xfire. I would like a single card to run 4K @60fps but gonna have to wait till 2017 for that...hopefully.
     
Thread Status:
Not open for further replies.

Share This Page