Review: Nvidia GeForce GTX 1080 Founders edition graphics card

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, May 17, 2016.

  1. GeniusPr0

    GeniusPr0 Maha Guru

    Messages:
    1,439
    Likes Received:
    108
    GPU:
    Surpim LiquidX 4090
    I consider the gtx 680 the only true nondisclosed midrange gpu. Just look at its crap performance now...


    As far as 980 being midrange, that can't be true since sli performance doesn't make a big enough difference going to 980ti. I have a 3dmark firestrike total system score of 30k with 3way 980s. Single card is only about 25% more on avg both overclocked.

    980 is almost twice as fast as a 960.

    This is also an insult to AMD, but they're catching up.

    Since the marker has always been relative family gpu performance to determine a class of card, the 1080 might actually be a "midrange" card due to gp102 having over 3000 cores and hbm2 if we only consider 4k as the new 1080p. Otherwise the term is now meaningless since 1080 60fps is easily attainable for most games now.
     
    Last edited: May 17, 2016
  2. Denial

    Denial Ancient Guru

    Messages:
    14,206
    Likes Received:
    4,118
    GPU:
    EVGA RTX 3080
    Why would you assume it should gain more by switching to DX12? The whole reason why AMD gains so much is because it's DX11 performance wasn't what it should be.
     
  3. Aelders

    Aelders Guest

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
    Which gigabyte one? The g1 gaming or just reference gigabyte 980 ?

    Anyway novemeber'14 is after launch by quite a bit no? launch was in September ?

    MSRP was $549, now it's $599.

    I'm saying, when it first launched, just the reference models, you couldn't buy one for less than 700 euros, no way.

    It really isn't that different to the 980 launch, the only major difference is that the inflation due to demand is accounted for in this "founder's edition"

    Trust me, speaking out of experience, you're fighting a losing battle. I've been trying for close to six months.

    I posted my AotS benches on reddit and the result was I was hounded and harassed by people until I closed the account.
     
  4. BahamutxD

    BahamutxD Master Guru

    Messages:
    269
    Likes Received:
    15
    GPU:
    Asus TUF 4070Ti
    I'll wait until prices for Eu go down, they are absurd right now.
     

  5. KFBR392

    KFBR392 Member

    Messages:
    34
    Likes Received:
    0
    GPU:
    RTX 3090
    Why would you assume that the gains in DX12 are purely in API overhead reduction? Yes it makes a difference in AMD's cards, but look at the Hitman bench there. The FuryX gains 4 fps from the dx11 to dx12 switch, pretty much the same as the 1080. So the current dx11 performance with AMD is totally fine. However, looking at AOTS, which benefits MASSIVELY from asynch computing you see the AMD cards take a huge jump in performance, yet the 1080 only sees a marginal gain. It seems like the async compute ability is improved from Maxwell, but it doesn't seem as strong as it is for AMD. This points to Pascal having an inferior implementation of async compute compared to AMD. Async is basically what hyper-threading was for CPU's. Its weight is only going to increase over time as it did with Intel v. AMD.
     
  6. Aelders

    Aelders Guest

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
    No.
    Async is not a big boost in AotS. It's DX12 that is the big boost.

    [​IMG]

    I'm so damned sick of hearing the same unfounded BS over and over and over and over.
     
    Last edited: May 17, 2016
  7. Denial

    Denial Ancient Guru

    Messages:
    14,206
    Likes Received:
    4,118
    GPU:
    EVGA RTX 3080
    You're falsely attributing Hitman's performance to Ashes. Ashes the difference between DX11 and DX12 is huge even without Async, turn Async on and it only gains about 4fps, 8-10%. Which is exactly what most of the DX11/12 benchmarks for the 1080 are showing.
     
  8. KFBR392

    KFBR392 Member

    Messages:
    34
    Likes Received:
    0
    GPU:
    RTX 3090
    10% is not a big boost?
     
  9. KFBR392

    KFBR392 Member

    Messages:
    34
    Likes Received:
    0
    GPU:
    RTX 3090
    1. I'm not attributing Hitman's pef to AOTS, I'm pointing to the fact that you said that AMD's performance in DX11 is lacking (this was true, but not anymore really). Showing that the performance gains in the Hitman bench were similar for both AMD and nvidia says something there.

    2. The gains for AMD are in the 5-10%, and for the 1080 it seems like 4-6%. Which again points to a weaker implementation.
     
    Last edited: May 17, 2016
  10. WhiteLightning

    WhiteLightning Don Illuminati Staff Member

    Messages:
    30,768
    Likes Received:
    3,937
    GPU:
    Inno3d RTX4070
    Yeah its the G1 one. dont know why the graph didnt go to the review date. nothing i can adjust about it either.

    Im only saying that IF the prices stated in the reviews are anything to go by the normal 1080 ones will be much higher then the 980 at lauch.
    Im anxious to see prices of the founders edition this week confirming me being worried about prices made sense or not.
     

  11. Aelders

    Aelders Guest

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
    not compared to the boost from DX11-DX12

    I don't know how much you researched this issue, but I've literally spent the last 5 months in open war with everyone about this.

    Any frequent forum users here probably associate the word async with me by now
     
    Last edited: May 17, 2016
  12. Aelders

    Aelders Guest

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
    I expect reference PCB + blower design to cost around $599, so accounting for retarded exchange rate + VAT I think it will not be so far off 980
     
  13. MainFrame Alpha

    MainFrame Alpha Guest

    Messages:
    139
    Likes Received:
    3
    GPU:
    Strix GTX 980 TI
    Nvidia should be going from "the way it's ment to be played' to" we are not the best we are the only ones" ecpeciallly with polaris 10& 11 being mid range and no details on higher end..so nvidia dose what nvidia dose best charge more!
     
  14. Denial

    Denial Ancient Guru

    Messages:
    14,206
    Likes Received:
    4,118
    GPU:
    EVGA RTX 3080
    Ok, then more like 6%. Except that 4 fps delta comes from the 290x in that graph. So I'm not sure what your point is.

    If you're trying to make the claim that AMD's Async implementation is better then Nvidia's, then yeah -- I'll agree, it is better. But they made a number tradeoffs to get it there. Not only that, but when Async is properly boosting performance, it basically means you screwed up somewhere else. Async essentially uses idle shaders to boost efficiency of the overall pipeline. If you have idle shaders then you aren't properly splitting the GPU tasks up over the pipeline in the first place.

    That's why Nvidia shows little to no gain, because their arch is already maxing out the efficiency of their pipeline. The problem with Maxwell is that the context switch to compute, after the graphics has been finished is slow for a number of reasons. That should be fixed with Pascal.
     
  15. Aelders

    Aelders Guest

    Messages:
    870
    Likes Received:
    0
    GPU:
    980Ti G1 @ 1490|8000
    Not to my understanding, the context switch is still needed to run a compute warp on an SM that was running a graphics warp.
    The difference is that on Pascal you can repartition the SMs (so each SM is doing EITHER compute OR graphics, not both ) dynamically without having to wait for draw call boundaries. At the GPC or GPU level, then it is indeed execution compute+graphics concurrently (and possibly in parallel), just not at the SM level.

    This is why AMD called their **** 'Async Shaders', because 'Async Shaders' refers specifically to their implementation in which multi-engine is done concurrently within the CU.
     
    Last edited: May 17, 2016

  16. Halfmead

    Halfmead Guest

    Messages:
    275
    Likes Received:
    50
    GPU:
    GigaByte 2070 Super
    6000+ Kroner (Danish) damnnnn...but, gonna get it anyways, when board partner cards surface :)

    Time to overhaul whole PC....
     
  17. crashburn162

    crashburn162 Active Member

    Messages:
    88
    Likes Received:
    10
    GPU:
    MSI 1080 Ti Wc
    So will we get SLI 1080 review in the coming days or you just got 1 demo card?
     
  18. Solfaur

    Solfaur Ancient Guru

    Messages:
    8,008
    Likes Received:
    1,530
    GPU:
    GB 3080Ti Gaming OC
    Great article Hilbert, enjoyed every bit of it!

    My conclusion is this:

    I will definitely buy a 1080, BUT I will definitely NOT buy an FE. Not only is it overpriced, but the cooling performance is just the same as usual (aka mediocre at best), nothing about the card's design warrants the extra 100 euros.

    However, an AIB like MSI Gaming, Gigabyte G1, ZOTAC xtreme etc. will likely be worth that extra cash. And by the time those are out, the OC software will be sorted... seriously, 2GHz on core! And with software issues... on a crap reference card?!? The potential in this department is huge, and I can't wait for the coming weeks to see the good stuff coming out.

    Also, I'm curious how this Nvidia FAST SYNC will work out, adaptive-sync was pretty meh, and I don't own a g-sync monitor yet. So maybe this new sync will finally replace good-ol' vsync? We'll see...
     
  19. zcess81

    zcess81 Guest

    Messages:
    8
    Likes Received:
    0
    GPU:
    EVGA GTX 980 Ti SC
    Paragraphs are your friend, Guru3D. I can guarantee you, almost no one will read through that conclusion wall of text.
     
  20. Aelders

    Aelders Guest

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

Share This Page