New Upcoming ATI/AMD GPU's Thread: Leaks, Hopes & Aftermarket GPU's

Discussion in 'Videocards - AMD Radeon' started by OnnA, Jul 9, 2016.

  1. Redemption80

    Redemption80 Ancient Guru

    Messages:
    17,802
    Likes Received:
    5
    GPU:
    GALAX 970/ASUS 970 STRIX
    I know, but it's not uncommon to get just min/avg/max.

    It's not a figure you can just discard, well maybe you can but I can't and without more detail the average is is something I would discard before the minimum.
     
  2. Tuga

    Tuga Member

    Messages:
    48
    Likes Received:
    0
    GPU:
    nvidia 670 gtx
    Last edited: Mar 23, 2017
  3. Rich_Guy

    Rich_Guy Ancient Guru

    Messages:
    11,548
    Likes Received:
    17
    GPU:
    Sapphire Fury Tri-X OC x2
  4. chispy

    chispy Ancient Guru

    Messages:
    7,846
    Likes Received:
    59
    GPU:
    GTX1080Ti @ 2126/6156 H2O
    Any ETA for big Vega yet ? I'm on the fence if i should keep on waiting and waiting and waiting :infinity: ... for big Vega or just go ahead and step up with the evga program to a gtx1080Ti
     

  5. Maddness

    Maddness Master Guru

    Messages:
    431
    Likes Received:
    4
    GPU:
    Asus RX480 Strix
    I have heard rumors around May release. Still nothing confirmed at this stage though. AMD is very tight lipped.
     
  6. Fox2232

    Fox2232 Ancient Guru

    Messages:
    5,583
    Likes Received:
    5
    GPU:
    Fury X - XL2420T(Z)@144Hz
    In that "presentation" yesterday, they did show some "illustrative" images.
    Originally Vega was rumored to have 2x HBM2. But on those images there were 4x HBM2. Or they were taken from Fiji.
     
  7. Maddness

    Maddness Master Guru

    Messages:
    431
    Likes Received:
    4
    GPU:
    Asus RX480 Strix
    That is interesting as so far all I have seen is Vega with 2 HBM stacks. AMD really have kept the lid on this GPU tight.
     
  8. OnnA

    OnnA Ancient Guru

    Messages:
    3,310
    Likes Received:
    17
    GPU:
    Nitro Fiji-X HBM 1150/570
    Some wonders if this Vega will be 3xTimes faster than Fiji?
    IMO it's possible, but now we need new Software (Game Engines) etc. to fully benefit from 8/16 and those BIG GPUs around there ;)

    We need brakethrough now to make Gaming Great again :flip2:
    PS. Where is Crysis 4 Vulkan/DX12 ?
     
  9. Ryu5uzaku

    Ryu5uzaku Ancient Guru

    Messages:
    6,661
    Likes Received:
    2
    GPU:
    980ti gaming 1490/7700
    Not possible. Or the thing would be quite a bit faster then 1080ti that manages to be 30-100% faster then Fury X depending on title. I can see it being equal to 1080ti tho if things have been done right.
     
  10. Rafnagud

    Rafnagud Member Guru

    Messages:
    165
    Likes Received:
    0
    GPU:
    Gigabyte GTX 1080 Ti OC
    Some 4K, DX12 or Vulkan Benchmarks maybe but i would be surprised if it can handle the Ti in DX11 1440p. Of course i hope it will because it gives gamers choices.
     

  11. Ryu5uzaku

    Ryu5uzaku Ancient Guru

    Messages:
    6,661
    Likes Received:
    2
    GPU:
    980ti gaming 1490/7700
    If they have done what they have said they have done. It should be way better in dx11 then gcn cards were.
     
  12. Rafnagud

    Rafnagud Member Guru

    Messages:
    165
    Likes Received:
    0
    GPU:
    Gigabyte GTX 1080 Ti OC
    Let's hope so :thumbup:
     
  13. OnnA

    OnnA Ancient Guru

    Messages:
    3,310
    Likes Received:
    17
    GPU:
    Nitro Fiji-X HBM 1150/570
    :thumbup:
    So as i say long time ago, this is not overhead
    It's ATI H/W Sheduler, when Game is payd by N it's build that way :bang:

    End of story, VEGA will be OK for DX11 also :banana:
    (BTW im happy with my Fury)

     
    Last edited: Apr 3, 2017
  14. Chastity

    Chastity Master Guru

    Messages:
    663
    Likes Received:
    9
    GPU:
    Sapphire Nitro 390 BP
    Very interesting vid read, Onna, thx.
     
  15. OnnA

    OnnA Ancient Guru

    Messages:
    3,310
    Likes Received:
    17
    GPU:
    Nitro Fiji-X HBM 1150/570
    THX, that guy knows stuff I know, but he can explain this in easy way :)
     

  16. Only Intruder

    Only Intruder Master Guru

    Messages:
    822
    Likes Received:
    1
    GPU:
    Sapphire R9 Fury Nitro
    So really, our thoughts of driver overhead these past couple of years has been while not incorrect but rather inaccurate in that there isn't driver overhead but rather the exposure of poor thread handling in game engines that show the primary thread being saturated with game logic while also that same thread is responsible for issuing the draw calls/rendering commands.

    Is this also why nVidia decided to abandon support for Fermi? Because Fermi's performance could have been equal to Kepler when exposed to the low level APIs? (Obviously though Fermi is a dead horse now, it's several generations old now and doesn't have the raw performance to compensate).

    Further more, does this also mean that in games utilising cores efficiently, the general advisement for AMD CPU owners to buy an nVidia GPU is also a poor recommendation as it turns out nVidia cards require more CPU cycles to perform it's scheduling, now while in single threaded applications, sure enough the recommendation is true but far from it in multi-threaded.

    Whatever the case, we have interesting times ahead of us. :D
     
  17. pharma

    pharma Master Guru

    Messages:
    714
    Likes Received:
    4
    GPU:
    Asus Strix GTX 1080
    https://forum.beyond3d.com/posts/1974379/
     
  18. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    6,857
    Likes Received:
    3
    GPU:
    Sapphire 7970 Quadrobake
    It's obviously both a hardware and a software issue. NVIDIA has a much more "DX11-friendly" architecture, and their graphics part of the pipeline is by far their strongest point. Their GPUs (even Pascal) are more difficult to be used by a lot of programs in parallel, but that's also a result of them keeping busy all the time.

    AMD seems to be doing seriously great under the new APIs, their driver and general architecture seems to be a mile away from NVIDIA's (see how much better they scale with multicores). The interesting battle will be Volta vs Vega, as AMD seems to have taken a lot of lessons from NVIDIA's designs with Volta (more efficient SIMDs, tile-based rendering), and Volta seems to be having much larger compute units and support HBM.
     
  19. OnnA

    OnnA Ancient Guru

    Messages:
    3,310
    Likes Received:
    17
    GPU:
    Nitro Fiji-X HBM 1150/570
  20. chispy

    chispy Ancient Guru

    Messages:
    7,846
    Likes Received:
    59
    GPU:
    GTX1080Ti @ 2126/6156 H2O
    Well i have decided to wait for AMD Big Vega for my next GPU upgrade :) , i keep hearing from everywhere it will be released next month in May ~ June around computex , so that's very close. We are almost there !
     

Share This Page