NVIDIA Announces the GeForce RTX 20 Series | RTX 2080 Ti 2080 2070 | Discussion

Discussion in 'Video Cards & Monitors' started by DiGiTaL MoNkEY, Aug 21, 2018.

  1. weak beta male

    weak beta male Member

    Joined:
    Aug 21, 2018
    Messages:
    56
    does anyone know how popular the new rtx 2060 is? is it selling well or is it not moving like the higher RTX? are there still tons of 2nd hand miner cards on the market? (i want to buy their shares so this info could be helpful)
     
  2. havabeer

    havabeer Member

    Joined:
    Dec 12, 2010
    Messages:
    4,322
    When people complained why the gpu's where so expensive. It's because they make bugger all sales on the top tier cards

    Especially when you have one company that spends a couple hundred million on one contract

    https://www.datacenterknowledge.com...d-world-s-fastest-supercomputer-us-government

    They'll make a shit load more money from those type of contracts then they will selling the occasional 2080ti to us gaming plebs
     
  3. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    35,050
    Location:
    Brisbane
    Moore's Law isn't dead. Moore's Law simply states that capacity will double every 18 months. And that is still true.

    Moore's Law says nothing about visual quality. What Nvidia can do with double the transistors isn't giving them double the visual quality. That has nothing to do with Moore's Law. That has to do with the upper bounds of how current generation lighting engines in video games and real time 3D engines work.

    Likewise, people often incorrectly assume that Moore's Law means things get "twice as fast" every 18 months. That's not true either. Previously with twice the capacity of transistors, CPUs "felt" faster for our day to day usage. Right now your Windows 10 desktop doesn't feel twice as fast when you upgrade your CPU because of other limits in asymmetric computing architecture. Ditto for your games.

    Nvidia (and others) are trying to mitigate customer expectations here, which in turn are incorrectly set through a misunderstanding of how technology progresses. Moore's Law is still alive and well, but video game engine technology, even with all the advancements in DirectX12 and Vulkan, has been largely stagnant for the better part of a decade with the way it "cheats" something as fundamental as how light bounces from a source, off an object, into the eye/camera.

    Go outside for a moment and stare at a real world object. Anything. A single brick, a blank bit of wall, the side of a car, whatever. Really concentrate, and think about what you are seeing and how that works at a physics level. You're not seeing the object. You're seeing energy that came from some distance source, travelled through space, lost energy along the way (atmospheric interference, refraction, absorption of materials it bounced off) and results in your eye interpreting colour and luminance to give the illusion of shape and texture.

    Now imagine how small a photon is (hint: really small) how fast it travels (hint: fast) and how many of those you need to have enough information to build an image (hint: a lot), and how many billion transistors you'd need to calculate and sample that information ever few milliseconds (hint: a lot squared).

    In human history, we've been nowhere near that, so we bullshitted it up until now with hacks and approximations. Right now, we're closer than ever, but still a whiles off making it realistic for what the customer base expects a video game to work at. Easy to comprehend, hard to explain to shareholders that they have to wait a little while for shit to catch up. So as a CEO, how do you mitigate that? You announce "Moore's Law is dead". It isn't. But it's a double-edged sword of demonstrating a technology that's 5-10 years off mainstream use now, and keeping people patient enough for the technology to catch up.

    Don't confuse marketing bullshit with science and engineering. We're still progressing as always. But right now this is like the Nvidia Riva 128 just got released, and everyone wants GeForce level technology and graphics (remember when "Hardware TnL" appeared, and we all were amazed?). Shit takes time, yo.

    I hope this also explains a little better why, after this long working in professional 3D, I'm not so fussed about all this progress. Improvement is inevitable, as are forums full of people pissing and moaning that improvement isn't happening fast enough, that the sky is falling, and that we've apparently reached an plateau. :)
     
    Last edited: Feb 6, 2019
  4. Brett240

    Brett240 Member

    Joined:
    May 11, 2017
    Messages:
    65
    best game I ever played was the secret of monkey island on my 386. the remake on xbox live is fantastic
     
    dzajroo likes this.
  5. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    35,050
    Location:
    Brisbane
    Brett240 likes this.
  6. mAJORD

    mAJORD Member

    Joined:
    Jun 4, 2002
    Messages:
    9,406
    Location:
    Griffin , Brisbane



    Key words were "as we knew it" and by that I mean as it applied to the end technology / products.

    You can increase # of transistors on an IC right up until you take up the entire wafer for one chip - Don't need a law for that - and if one simply wanted to continue 'moore's law' in it's most simplistic definition for the hell of it, they could. But it would be pointless.

    Moores law in the real world though is closely tied to Transistor density, performance, and cost - Both to tape out, and on a per chip basis.. All complex issues in there own right (taking into account yield, logic scaling of different types of cell etc) , but until recently, that has followed a predicable trend over time - it hasn't been perfect, but Manufacturers could continue producing faster products, at similar cost to predecessors.. and End users could rely on this Cadance for the future.

    That trend is now broken - That's why manufacturers (including nvidia as above) are no longer relying on it / are informing people not to rely on it going foward as an indicator of future expectations- at least for now. There is light at the end of the process scaling tunnel, with emerging technologies for beyond 3nm, but there's too much uncertainty about it still.
     
  7. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    35,050
    Location:
    Brisbane
    I think that's more people misunderstanding of what it means and where the bottlenecks are in systems, rather than the law being dead.

    Our desktop computing experience hasn't changed much in 10 years, even if Moore's Law has continued on. That, nor Nvidia's assertion, invalidate the law. All it does is demonstrate that the desktop has bottlenecks in other places.

    Ditto for gaming. Double the polygons and pixels, and the relative change to the gamer's eye is almost invisible today. That doesn't mean the change didn't happen, only that we're incapable of seeing it with the naked eye.

    Video game lighting, on the other hand, has sucked forever. That's going to be a huge break through, and much more visible to the unwashed masses.

    True, and "break throughs" come along and help. Look at the research HP is doing in their (stupidly named) "The Machine", and how they can dynamically assign memory to systems in another rack. Soon the limits of transistors on a single die will phase out, and we can build things that don't rely on everything being on one card. For problems like ray tracing, that's pretty exciting.
     
    Last edited: Feb 6, 2019
  8. power

    power Member

    Joined:
    Apr 20, 2002
    Messages:
    56,695
    Location:
    brisbane
    it's all getting chewed up in resolution and fidelity (models, meshes etc) right now.

    that's why we get billions of fps in old games.
     
  9. RnR

    RnR Member

    Joined:
    Oct 9, 2002
    Messages:
    12,217
    Location:
    Brisbane
  10. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    35,050
    Location:
    Brisbane
    Yeah, we hit a frequency wall years ago. You just can't push electrons through semiconductors much faster for a bunch of reasons (they tend to teleport around the place in weird ways that baffle particle physicists).

    That's why everything is multi core. And as I said above, the interconnect between cores and memory is going to change soon, and make life very interesting.
     
    Last edited: Feb 6, 2019
  11. weak beta male

    weak beta male Member

    Joined:
    Aug 21, 2018
    Messages:
    56
    optical computers will reach 1thz. fully gay traced real time 60fps avatar graphics in 10-15 years
     
  12. mesaoz

    mesaoz Member

    Joined:
    Jan 15, 2015
    Messages:
    4,621
    Location:
    Twitch / Brisbane
    I don't know why gay traced made me laugh, but it did.. so thanks.
     
  13. Court Jester

    Court Jester Member

    Joined:
    Jun 30, 2001
    Messages:
    2,333
    Location:
    Gold Coast
    I dont know about that I read somewhere that god hates frags

    here we go

    [​IMG]
     
    Last edited: Feb 6, 2019
    hvalac likes this.
  14. Perko

    Perko Member

    Joined:
    Aug 12, 2011
    Messages:
    3,102
    Location:
    NW Tasmania
    Dunno about you guys, but I'm voting yes to polyfrag.
     
  15. chainbolt

    chainbolt Member

    Joined:
    Jun 27, 2001
    Messages:
    103,015
    Location:
    Tokyo
    Does this make sense? If the mentioned issue with the wafers were the cause, then 2080 cards would have been also affected, as (I thought) the difference between 2080 and 2080ti is not the GPU itself, but the memory size and the clock / bus speed.

    The reason why some early samples of the 2080ti failed has been identified as the DDR6 memory, apparently in particular the one produced by MICRON.
     
  16. Mathuisella

    Mathuisella Member

    Joined:
    Dec 3, 2002
    Messages:
    5,990
    Location:
    in your gearbox...grindin

    I agree :)

    just like AMD put the memory controller on the CPU chip instead of under a northbridge heatsink on the motherboard, we'll see another game changer soon out of left field that nobody was expecting but will follow...

    just like Fosbury :)


     
  17. power

    power Member

    Joined:
    Apr 20, 2002
    Messages:
    56,695
    Location:
    brisbane
    well writing games off because of your pre-conceived notions i guess just means you miss out, and boy oh boy are you missing out.
     
  18. dzajroo

    dzajroo Member

    Joined:
    Dec 23, 2010
    Messages:
    791
    just because I like the old school games, doesn't mean I'm writing off new stuff, finding that indie games have much more fun in them that most of the AAA these days...

    now to the topic, looks like 11xx is really going to happen.

    [​IMG]
     
  19. Zenskas

    Zenskas Member

    Joined:
    Jan 27, 2010
    Messages:
    1,799
    Location:
    Tas
    The 1160Ti could be good - something new to compete with the currently much better value RX 580 8GB (when compared to the current 1060).

    While the RTX 2060 is much better value than the top end cards it is still hard to swallow a "mainstream" card for AU$600 or more - $700 for a Strix card! Hopefully this 1160 sits at around the same cost as current 1060 6GB cards but with better performance. Although knowing NVIDIA it will probably be a little more expensive...
     
    Last edited: Feb 8, 2019
  20. WRX_STi

    WRX_STi Member

    Joined:
    Sep 6, 2001
    Messages:
    1,693
    Location:
    Cairns
    I don't give a fuck about ray tracing or anything that gets in the way of performance of VR.
    The latest cards can't even run the shit properly, don't bring more garbage in to slow it down. Coming are massive FOV and 8k headsets.. Give me power bitches.
     

Share This Page