AMD: Navi Speculation, Rumours and Discussion [2019]

Discussion in 'Architecture and Products' started by Kaotik, Jan 2, 2019.

  1. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    9,130
    Likes Received:
    3,020
    Location:
    Finland
    Like I said, it's apparently supposed to be their relationship to each other, not in performance or any such.
     
  2. Leovinus

    Newcomer

    Joined:
    May 31, 2019
    Messages:
    116
    Likes Received:
    48
    Location:
    Sweden
    This always irks me. There is a lot of hate hurled at AMD if they don't keep prices low and performance up, if they raise prices it' because they're "turning into that which they were foretold to destroy". And if they fail to compete on performance on the merger budget they've had they're inept. It's a heck of a damned if you do, damned if you don't situation. Though hopefully the recent windfalls for AMD have given them the budgetary space they need. I don't think people understand the comparative shoestring budget AMD's graphics division has been working with compared to nVidia.

    Personally I can only hope that we see progress on the power efficiency side of the equation besides whatever ray tracing technology they introduce. Do that and remain somewhat competitive with the low to mid-end of whatever nVidia comes up with next and I'd be rather content. Of course, should they hit it out of the park with a high end SKU, better still.
     
    w0lfram, Alexko, ethernity and 6 others like this.
  3. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,603
    Likes Received:
    644
    Location:
    New York
    From a pure engineering standpoint the criticism isn't unwarranted. Navi has a feature deficit and a node advantage over Turing and still didn't make much of a dent. Also, AMD's CPU division has proven that it is possible to topple goliath will a well timed shot.
     
    Cuthalu, Lightman and DegustatoR like this.
  4. JoeJ

    Regular Newcomer

    Joined:
    Apr 1, 2018
    Messages:
    981
    Likes Received:
    1,108
    ... but maybe that's not much worse than being the 'self appointed innovator, master of marketing lies, and price increasing evil' like some certain competitor :D
     
    Malo likes this.
  5. Remij

    Newcomer

    Joined:
    May 3, 2008
    Messages:
    114
    Likes Received:
    138
    Luckily Nvidia has access to the same process node as AMD.. so it's probably not going to be quite as easy as that when it comes to GPUs... but we'll just have to wait and see the difference between Ampere and RDNA2 GPUs. AMD has claimed that they would disrupt the 4K gaming space, so I'm ready either way. Light the fire under the butt of Nvidia and lets get a good high end GPU war going on between them!
     
    w0lfram and Lightman like this.
  6. Kaotik

    Kaotik Drunk Member
    Legend

    Joined:
    Apr 16, 2003
    Messages:
    9,130
    Likes Received:
    3,020
    Location:
    Finland
    And when you make a miss on big bet you're out of competition for years, as seen with Bulldozer and it's variants, can AMD take such risks on two fronts at the same time?
     
    sir doris and BRiT like this.
  7. Leovinus

    Newcomer

    Joined:
    May 31, 2019
    Messages:
    116
    Likes Received:
    48
    Location:
    Sweden
    Certainly. And that's the level of discussion I enjoy on this forum. Levelheaded, honest, and upfront with an appreciation for the complexities involved. But fanboyism and its ilk paint the world in much simpler terms. And it's just... silly.

    But with regards to the feature deficit and node advantage AMD currently has against nVidia: That's precisely why I think efficiency improvements will have to be premiered in RDNA2 aside from ray tracing tech. And I hope the Vega improvement in Renoir is a hint as to how seriously AMD is focusing on it. Because nVidia will come out swinging. God knows they haven't been sleeping on their laurels. And as it stands nVidia wouldn't have to do much more than port Turing to 7nm to outpace RDNA1 with comparatively modest efforts involved.

    At the same time I remain cautious. AMD could have been spread rather thin working with Sony and Microsoft as well as readying RDNA2 for desktop deployment. Without any insight we can't know how that dynamic shifted resources and development focus, or if there was meaningful cross-pollination. It's paramount that there was though. Unless their ray tracing implementation is so staggeringly competitive as to make other shortcomings forgivable as an alternative, I don't think AMD would be able to face nVidia on even nearly equal terms with a comparatively sizeable efficiency deficit. And that would still assume that ray tracing proves useable on these cards. I'd argue that the ray tracing capabilities of the current RTX crop is of limited usefulness now, and even less so for the next generation of ray-tracing titles.
     
    w0lfram likes this.
  8. dorf

    Newcomer

    Joined:
    Dec 21, 2019
    Messages:
    43
    Likes Received:
    114
    A similar naming scheme is probably the reason for the comparison. The Navi 10 product names help clarifiy what the XTX, XT, XL & XE tags signify.

    Also maybe the similar naming scheme means they are both the top dies of their generation.
     
  9. yuri

    Newcomer

    Joined:
    Jun 2, 2010
    Messages:
    211
    Likes Received:
    184
    TL;DR AMD has to be the cheaper solution since recent generations have comparatively worse drivers, less features, lacking compute ecosystem, and worse TDP rating. Raising the prices but fixing none of the traditional flaws makes ppl complain.

    AMD surely has a shotestring budget and it simply shows.

    GCN itself was a Fermi counterpart. It was so long overdue - a healthy unconstrained development would bifrucate the HPC line right after Hawaii. So gaming market would enjoy an actually working solution instead of Fiji, absent Polaris highend or utterly broken Vegas. Navi/RDNA1 came very late in this regard. Navi is still plagued by driver/firmware problems - downclocking inducing stuttering, weird 'blackscreen' problems, etc. I've stopped tracking this but Navi OpenCL driver seems to remain broken to this day. AMD's attempts to make something to counter CUDA are still more like hobbyist projects. The CDNA aka Arcturus seems to be just another iteration of Vega.

    A "Navi 10 Refresh" sounds faky since they apparently plan 3 RDNA2 models which should be able to cover the whole traditional market spectrum. Additionally, RDNA2 *should* target 50% perf/W improvement, which would make the Navi 10 not fitting well.

    RDNA2 has a potential - budgets might be a bit fatter, the design decision should be gaming oriented, etc. But Ampere will enjoy a full node jump.
     
  10. Samwell

    Newcomer

    Joined:
    Dec 23, 2011
    Messages:
    127
    Likes Received:
    154
    You're forgetting the timeframes for the different GPUs. A Navi refresh should happen in early Q3 this year. The only RDNA2 Chip this year should be N21 with N22, N23 coming later. At least i haven't heard many rumours of other RDNA2 GPUs beside Big Navi coming this year. I wouldn't expect N23 before Q2 21, so it's 9 months lifetime for the refresh. Sounds pretty ok for me.
     
    w0lfram likes this.
  11. yuri

    Newcomer

    Joined:
    Jun 2, 2010
    Messages:
    211
    Likes Received:
    184
    Good point. However, AMD recently managed to launch both 5500XT and 5600s just in a month. Maybe they could manage to launch all three RDNA2 models in like two months or so.
     
  12. DegustatoR

    Veteran

    Joined:
    Mar 12, 2002
    Messages:
    1,501
    Likes Received:
    242
    Location:
    msk.ru/spb.ru
    5600 isn't based on a new GPU though.
     
  13. DegustatoR

    Veteran

    Joined:
    Mar 12, 2002
    Messages:
    1,501
    Likes Received:
    242
    Location:
    msk.ru/spb.ru
    With or without RT? 8)
     
  14. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,603
    Likes Received:
    644
    Location:
    New York
    Only AMD knows what they meant but even without RT it’s a tall order. The 2080 Ti has about 40% more flops and bandwidth than the 5700xt but is 50% faster.

    Let’s say the bar for 4K disruption is 2080 Ti performance without RT for $500. That would be quite disruptive although the 2080 Ti doesn’t guarantee 4K 60fps in all titles. I doubt it’ll be capable of that in cyberpunk either.

    PS. mods please be mindful of the indiscriminate pruning of on-topic posts.
     
    w0lfram, pharma and PSman1700 like this.
  15. BRiT

    BRiT Verified (╯°□°)╯
    Moderator Legend Alpha

    Joined:
    Feb 7, 2002
    Messages:
    15,912
    Likes Received:
    14,827
    Location:
    Cleveland
    Everyone please be mindful of staying on topic and not getting into versus territory. If everyone stayed on topic there wouldnt be the need for moderators to get involved, so plead with your fellow posters.
     
  16. szatkus

    Newcomer

    Joined:
    Mar 17, 2020
    Messages:
    30
    Likes Received:
    23
    I recently did some math and got number like 20-30% over 2080Ti.
     
    w0lfram likes this.
  17. CarstenS

    Veteran Subscriber

    Joined:
    May 31, 2002
    Messages:
    4,943
    Likes Received:
    2,286
    Location:
    Germany
    After 2+ years and on a full node advantage, I'm really looking forward to your math coming true.
     
    pharma and PSman1700 like this.
  18. ToTTenTranz

    Legend Veteran Subscriber

    Joined:
    Jul 7, 2008
    Messages:
    11,076
    Likes Received:
    5,626
    He said "over the 2080 Ti", not over nvidia's next-gen offering.
    It's not hard to assume a 80 CU Navi 2x will be 20-30% over the 2080 Ti. Just do Navi 10 x2 that's where you stand.

    [​IMG]




    At the same time, assume AMD's own +50% power efficiency numbers and you get Navi 10's 225W * 2 / 1.5 = 300W.
    So a 300W graphics card that beats the 2080 Ti by ~25% is just within AMD's promise.

    Ampere is the real wild card here though. GA100's clocks aren't anything to write home about, when compared to GV100 and GP100, but it might not be representative of their consumer lineup.
     
  19. PSman1700

    Veteran Newcomer

    Joined:
    Mar 22, 2019
    Messages:
    2,686
    Likes Received:
    897
    What kind of TF performance numbers are we looking at with such a GPU, at higher clocks? Would give a raw idea, 2080Ti sits around 14/15TF at max boost?
     
  20. nAo

    nAo Nutella Nutellae
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    4,333
    Likes Received:
    128
    Location:
    San Francisco
    Sadly GPUs almost never scale 2x, especially when they get very big.
     
    CarstenS, pharma and DegustatoR like this.
Loading...

Share This Page

  • About Us

    Beyond3D has been around for over a decade and prides itself on being the best place on the web for in-depth, technically-driven discussion and analysis of 3D graphics hardware. If you love pixels and transistors, you've come to the right place!

    Beyond3D is proudly published by GPU Tools Ltd.
Loading...