Guru3D Reviews: GeForce RTX 2080 and 2080 Ti Founder Edition

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Sep 19, 2018.

  1. alanm

    alanm Ancient Guru

    Messages:
    12,274
    Likes Received:
    4,478
    GPU:
    RTX 4080
    You fell for it. Thats the point of the 2080Ti being released at same time as 2080, to get peoples attention off the failure of the 2080. Lets face it, this launch could not have occurred with the 2080 by itself, a new card that is roughly the same performance point of a 1 1/2 year older card at about same price point (1080ti).

    A few weeks ago, many of us were wondering, how strange for Nvidia to move up a Ti cards release date to line up with a xx80 card, now we know.
     
    Valken and carnivore like this.
  2. RavenMaster

    RavenMaster Maha Guru

    Messages:
    1,360
    Likes Received:
    253
    GPU:
    1x RTX 3080 FE
    Nice one, looking forward to it :D
     
  3. Agent-A01

    Agent-A01 Ancient Guru

    Messages:
    11,640
    Likes Received:
    1,143
    GPU:
    4090 FE H20
    Sums it up well.

    I'd buy a 2080Ti if it was 800 USD.

    Considering I spent 600 USD flat for a 1080Ti really makes the price not worth it to upgrade.
     
    Elder III and chispy like this.
  4. metagamer

    metagamer Ancient Guru

    Messages:
    2,596
    Likes Received:
    1,165
    GPU:
    Asus Dual 4070 OC
    Those with 1080ti can stick with it, unless gaming at 4k. 2080ti performance is nice but I expected a little bit more from the 2080. 2070 no doubt will be slower than the 1080ti, by some 5-10%.
     

  5. yasamoka

    yasamoka Ancient Guru

    Messages:
    4,875
    Likes Received:
    259
    GPU:
    Zotac RTX 3090
    You didn't answer my question.

    Here's what your link has to say:

    For 16x SSAA:
    n = 16
    Each GPU: n / 2 = 16 / 2 = 8 color samples per pixel (i.e. 8x SSAA per GPU)

    For 32x SSAA:
    n = 32
    Each GPU: n / 2 = 32 / 2 = 16 color samples per pixel (i.e. 16x SSAA per GPU)

    Again, I ask, how do you think this could ever be feasible on modern GPUs for modern games at acceptable framerates?

    For 1080p, that would be akin to running the game, PER GPU, at ~ 5431x3055 for 16x SLIAA and 7680x4320 (8K) for 32x SLIAA.
     
  6. yasamoka

    yasamoka Ancient Guru

    Messages:
    4,875
    Likes Received:
    259
    GPU:
    Zotac RTX 3090
    In which games is this screen-space technique used?

    Those shaders offer lower precision per computation and are in no way related to render resolution, regardless of whether the fractions look the same. Reduced precision cannot magically allow you to render lower resolutions faster.
     
  7. Koniakki

    Koniakki Guest

    Messages:
    2,843
    Likes Received:
    452
    GPU:
    ZOTAC GTX 1080Ti FE
    Nuh, the RTX 2080 was of no interest whatsoever. Actually it's the other way around for me. The 2080 was released alongside the 2080 Ti and barely looked at it in the charts.

    I was focused on the 1080 Ti vs 2080 Ti. :)
     
  8. Yakk

    Yakk Guest

    Messages:
    144
    Likes Received:
    21
    GPU:
    Graphics card
    So basically only look at the highest priced 2080ti to be worth the performance increase, all the while massively jacking up prices across the board, nice marketing coup nvidia. And still having people buy into it.

    This is Apple grade marketing, +1 business respect.
     
  9. Robbo9999

    Robbo9999 Ancient Guru

    Messages:
    1,858
    Likes Received:
    442
    GPU:
    RTX 3080
    I'm not sure why there's a lot of people here saying that 2080ti performance was good but 2080 performance was disappointing. If you look at the game data that Hilbert tested, it's showing a bigger performance jump percentage wise going from GTX 1080 to RTX 2080 than the corresponding jump from GTX 1080ti to RTX 2080ti. I worked it out on each chart, the percentage difference. So actually the RTX 2080 is more impressive than the RTX 2080ti in this respect. What also makes this strange is that the RTX 2080ti has a huge percentage increase in cores (47% more cores!) over the GTX 2080 yet doesn't seem to be able to turn that into a large performance increase. I see the RTX 2080 as the "winning" card here if one HAD to be chosen from the two.
     
  10. metagamer

    metagamer Ancient Guru

    Messages:
    2,596
    Likes Received:
    1,165
    GPU:
    Asus Dual 4070 OC
    It's because last couple of generations, the 980 trounced the 780 and 1080 trounced the 980. x70 cards were on par with previous generation of x80ti cards (970=780ti, 1070 = 980ti). This time around, the 2070 will be slower than the 1080ti. By how much, we're soon to see.
     

  11. wavetrex

    wavetrex Ancient Guru

    Messages:
    2,465
    Likes Received:
    2,578
    GPU:
    ROG RTX 6090 Ultra
    Are you actually comparing a two-year-old $450 (now) card with a new $800 card and get surprised by the performance difference ?

    If NV would give a card AT 450 $ that would be 40% faster then I would be impressed. But that's not the case, isn't it ?

    Judging by shader count and clock speeds, it might even be a bit slower than 1080 without the Ti, or at least equal, give or take.
     
    carnivore likes this.
  12. HybOj

    HybOj Master Guru

    Messages:
    398
    Likes Received:
    327
    GPU:
    Gygabite RTX3080
    Last note... I wonder how the cooling will work, when the chip gets actually utilized? I mean, when you stress those 3 parts of the chip! As that is something we are not able to test right now. For that, you need a benchmark with a raytracing and DLSS, right? Well... its gonna be funny
     
  13. Endymion

    Endymion Member Guru

    Messages:
    125
    Likes Received:
    17
    GPU:
    4090 TUF
    Oh my god, what is this 2k performance? Not worth for switch from 1080 Ti.
     
  14. Robbo9999

    Robbo9999 Ancient Guru

    Messages:
    1,858
    Likes Received:
    442
    GPU:
    RTX 3080
    Well we can't get around the fact that the overall performance increase of Turing is disappointing in comparison to some previous generations increase in performance. At the moment it looks like the 2080ti is not making full use of it's potential, there's something up with that, I expected so much better performance from it in comparison to the RTX 2080, given that it has 47% more cores - I mean it's essentially 1.5 times the card! So, GTX 1080 owners upgrading to RTX 2080 are a lot more sensible than GTX 1080ti owners upgrading to RTX 2080ti - by a long shot! RTX 2080ti is a disappointment, the RTX 2080 less so.
     
  15. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    I do not think any game renders additional geometry outside of display area. Yet rendering few more pixels fixes those most visible problems and is by far cheaper than raytracing.
    (Because for 1080p this would render just 32% more pixels... 1440p has 78% more pixels than 1080p, so performance impact would be somewhere in between regular rendering for those 2 resolutions.)
    Actually, calculating one value for 2x1, or 1x2 pixels behaves exactly that way. Something in opposite direction would have one shader calculation per 4x4 matrix. And I am quite sure nVidia did introduce that for some serious speedup purposes.
    [​IMG]
    I think that while none of us really wants to see 4x4 reduction on display, we would not mind using it off-screen.
    (If it slipped your attention, text actually means: "Shade less where it is not needed." Because this is not meant to double or quadruple shading density where additional details can be gained.)
     

  16. jura11

    jura11 Guest

    Messages:
    2,640
    Likes Received:
    707
    GPU:
    RTX 3090 NvLink
    Thanks @Hilbert Hagedoorn for adding the V-RAY benchmark

    My GTX1080Ti with 2113MHz will do this benchmark in 1 minutes and 2 seconds, 3 GPUs will do this benchmark in 26-28 seconds

    Personally I would be waiting on Octane and Blender Cycles benchmarks and will see if RTX 2080Ti is worth it for me

    Hope this helps

    Thanks, Jura
     
  17. Robbo9999

    Robbo9999 Ancient Guru

    Messages:
    1,858
    Likes Received:
    442
    GPU:
    RTX 3080
    I think you missed the point of my post. I was comparing the performance jumps from GTX 1080 to RTX 2080 with the performance jump from GTX 1080ti to RTX 2080ti. Your points are not related to that. I'm not impressed with the performance of these cards and certainly not at the increased prices, I think you missed the point I was making.
     
  18. yasamoka

    yasamoka Ancient Guru

    Messages:
    4,875
    Likes Received:
    259
    GPU:
    Zotac RTX 3090
    So that extra 10% left and right and 5% top and bottom would be enough to include geometry that is involved in lighting and shadows in the active area to actually compete with the raytracing approach?

    Oh, I see. You meant variable rate shading. That would definitely work off-screen. See top question though.
     
  19. Fender178

    Fender178 Ancient Guru

    Messages:
    4,194
    Likes Received:
    213
    GPU:
    GTX 1070 | GTX 1060
    Thanks for the Work Boss.
    It's too much money for me considering I just purchased a 1070 and a Gaming laptop. Plus 2070 looks like it's going to be cost more than the 1070 and besides the 1070 is good enough for me any way for the time being.
     
  20. Darren Hodgson

    Darren Hodgson Ancient Guru

    Messages:
    17,222
    Likes Received:
    1,541
    GPU:
    NVIDIA RTX 4080 FE
    After watching three video reviews of the RTX 2080 series cards on YouTube, I have to say that I'm deeply underwhelmed by them, especially the RTX 2080 which seems to be a more expensive version of the GTX 1080 Ti at its current pricing, only you pay a premium for the RTX feature which no-one has really been able to benchmark yet because... well, there's no games out there yet that make use of it!!!

    As for the RTX 2080 Ti, if money wasn't a consideration then the 20-40% framerate improvements at 1440p and 4K would be very nice but at (currently) £400 over the cost of the GTX 1080 Ti it looks to be poor value. The performance improvement over the last generation of GPUs is what you'd expect anyway so that £400 is for the RTX feature, again of which there is zero to show for it yet besides a Star Wars tech demo and a FFXV benchmark for DLSS.

    So glad I resisted the urge to pre-order. As some sites have mentioned, there's a sense that this card has been rushed out unfinished and why...? There's no competition from AMD so NVIDIA could have held back the cards until more games had support added, the new Windows 10 update was released and the drivers were more mature. Quite why they felt this card was needed now is unknown as there really isn't any game out yet that really needs one. The GTX 1080 Ti is perfectly capable of 4K 60 fps gaming with a few tweaks so that £1,100-£1,200 is really hard to justify... unless you've got money to burn, of course. ;)

    This card for me is like the GTX 480, perhaps not quite as bad as that, but it wasn't until the GTX 580 and especially the GTX 680 that the technology matured enough to produce great cards at great prices. I think it's fair to say that the RTX 2180 Ti and RTX 2280 Ti will be significantly better cards and (hopefully) better value once AMD release something competitive. And by then there might actually be a decent number of games supporting DLSS and ray-tracing to make them worth buying.

    If you ask me, right now, I would rather have paid £700 for the RTX 2080 Ti *without* the Tensor cores and just had that traditional 30% bump in performance in rasterised games. The RTX feature at this point in time has no value and by the time there is stuff to showcase it there will be new cards out anyway.
     
    Elder III likes this.

Share This Page