Nvidia's 2nd generation DX11 GPU's?

I've heard much about ATI's southern and northern islands, but does anyone have any idea what Nvidia's next-gen GPU's will be called, and when they're going to be released?

I hope they aren't a power-hungry, overheating flop like Fermi was.
34 answers Last reply
More about nvidia generation dx11
  1. ambam said:
    I've heard much about ATI's southern and northern islands, but does anyone have any idea what Nvidia's next-gen GPU's will be called, and when they're going to be released?

    I hope they aren't a power-hungry, overheating flop like Fermi was.


    Well it depends what you could as next gen. Their real "Next gen" is Kepler, Which seems to be coming out around the end of 2011. Q3 2011-Q1 2012. However they also have apparently some cards coming in between now and then. Fully unlocked GF104, which has a myriad of possible names, most being GTX 560 or GTX 475. Then theres the fully unlocked GF100, called GF100b. This may be called the 485. Then theirs the GTX 580, which is a new design apparently, however not far away from fermi. This should be paper launched around december, and im guessin really launched around february if there are no delays. Then of course scaled down 580, which would be the 570. All these are rumors, but most are likely, just not necessarily anytime soon.
  2. ares1214 said:
    Well it depends what you could as next gen. Their real "Next gen" is Kepler, Which seems to be coming out around the end of 2011. Q3 2011-Q1 2012. However they also have apparently some cards coming in between now and then. Fully unlocked GF104, which has a myriad of possible names, most being GTX 560 or GTX 475. Then theres the fully unlocked GF100, called GF100b. This may be called the 485. Then theirs the GTX 580, which is a new design apparently, however not far away from fermi. This should be paper launched around december, and im guessin really launched around february if there are no delays. Then of course scaled down 580, which would be the 570. All these are rumors, but most are likely, just not necessarily anytime soon.


    Doesn't Nvidia's "Maxwell" cards have Sixteen times the performance of the Fermi's?
  3. no they dont have 16x the performance, they speculate that they predict 16x performace. probably need a 2kw psu and watercooling to run them though.
  4. ambam said:
    Doesn't Nvidia's "Maxwell" cards have Sixteen times the performance of the Fermi's?


    NO!!!!! They have Speculated it will have 16x "CUDA Performance Per Watt" and thats not even against fermi. CUDA performance doesnt eqaul real performance, and per watts is easy to get better on shrinking MP.
  5. iam2thecrowe said:
    no they dont have 16x the performance, they speculate that they predict 16x performace. probably need a 2kw psu and watercooling to run them though.


    lol, is that a joke? :D . btw if i'm not mistaken when they say 16 times performance its not the game will run 16 times faster but you can get 16 times more DP compared to current Fermi architecture.
  6. "We came across some interesting information regarding Nvidia's response to Cayman. The new chip from Nvidia, codenamed GF110 is going to launch in November, most likely before Cayman. AMD set its launch date around November 20 and Nvidia is likely to launch the GF110 a bit earlier.

    It is not only that the chip will have more than 480 cores, we dare not say 512, but all these cores will be improved. Nvidia had more than a year to work on this mid-life kicker, something that can compete with AMD's Cayman and could likely end up as the fastest GPU on the market.

    First reports indicate that it will get available for around $499 but this will very much depend on Cayman performance and pricing.

    The cards are in production as we speak, almost ready to ship at launch. Now its all a matter of politics and deciding when to launch them. As the winter in the northern hemisphere draws near the Cayman and the GF110 will come in quite handy, as both will generate enough heat to keep gamers cozy at sub-zero temperatures."
    http://fudzilla.com/graphics/item/20676-geforce-gtx-580-to-launch-in-november
  7. THE DESPERATION AT Nvidia has reached silly proportions, with the paper launch of the GTX580 pulled in from 'Cayman Day' to 'Investor Conference Call Day', both holidays in November. I wonder how many reviewers will overlook the fact that there won't be cards this year?

    We told you about the GTX580 a few weeks ago and now a few more details have come out. The short story is that the chip behind the GTX580, the GF110, is nothing more than a bug fixed GF100. If you think this should be called the GF100b, you are right. If you think performance will underwelm, you are right too.

    Sources deep within Nvidia tell SemiAccurate that the clock speeds targeted by GTX580 are 750/1500, or the exact same frequencies that Nvidia was targeting before the GF100/GTX480 missed. Where have we heard that number before? Oh yeah, we have been saying that that was the original clock target for GF100 for over a year, and Nvidia has been denying it for just as long. They denied many other problems too, but that's not important now that we have an imminent launch.

    It is pretty obvious that GF100b/GF110 was supposed to be a shrink to 32nm, but when TSMC canceled that node, Nvidia had to jump into action on Plan B. That meant there was time to do a base layer respin on GF100 to fix some of the bugs.

    A fully working GF100, a first for Nvidia's desktop Fermi/GF1x0 line, would have 512 shaders and get about a 7% speed increase over GTX480. Another 6% comes from the clock speed, a generous assessment would say bug fixes add a little more. All told, 20% net speed increase over GTX480 isn't out of the question, nor is a pretty decent lowering of power.

    http://www.semiaccurate.com/2010/10/30/nvidia-paper-launch-gtx580-week/

    If the above is true, do not be fooled, this will not be a 2nd gen, more a revision of Fermi
  8. Try to focus past his 'hate' posturing , to the last paragraph. If Charlie is already admitting there is a 20%+ plus performance improvement with a 'decent' lowering of power, this qualifies as next generation as much as Barts might, which is all we can compare it to right now, which took on x780 moniker of the past flagship gpu. Which is still controversial. Don't be fooled :)
  9. Barts has been described as the red haired stepchild of the 6xxx family, where the true innovations will be seen in the 69xx cards
    Hating, liking or neutral has little reflection on truth, overall

    My point is, is this really a 2nd gen?
    While it may be argued that the 6870-50 may or may not be new gen, tho theres many changes there, will we see this on the proposed 580?
    Or is it more like what we saw with the 4870 to the 4890? Or the 2900 to the 3870?
    Each of these either brought perf/power down, or just perf up
    The drastic reductions seen in the 2xxx to 3xxx series is considered a true new approach, whereas the 4870 to 4890 was seen as a pure upgrade path only, no new gen, with slight revisions
    If the 580 scales similarly with the 480 due to clocks only, plus additional shaders, which are found on 480, just unusable, then its more like a 4870 to 4890
    So, new gen? Im not so sure nVidia has explained that much
    Newer revised card on older series? This too is possible, and Im pointing that out, and therefor, it isnt a 2nd gen DX11 part IMO
  10. for me nvidia next generation will be kepler which is based on 28nm fabrication. however if you're talking about next series it will be still based on fermi which is much more refine version of it.
  11. Also, if this does occur, where we see a better Fermi, hopefully itll solve Fermis scaling problems, where weve seen the 450, 430 as underwhelming
  12. Well, in Nvidia's favor, they can do a lot with their arch if they wanted to. The problem is every step they take towards a more efficient arch is a step away from CUDA performance. Take the 460, or more specifically GF104 in example. It has 11% higher texture fill rate performance than the 470, and yet 25% less SP. Not to mention on a 37% smaller die. How did they do this? They cut out CUDA/GPGPU performance parts, some shaders, reducing several other things and the tesselation. So they can either have high efficiency and lower CUDA performance, or low efficieny and higher CUDA performance. Very troubling situation to be in IMO, either lose some of one of your best features, or continually lose in the efficiency race. BUT, in the event the 580 was built more so like GF104, it could do some serious damage. But no way is it releasing before Cayman. Id say we are lucky if we see it before the end of the year.
  13. You mean, were going to have to go thru this again?

    Hope it isnt wood screws n promises, as that reminds me of a country song
  14. I suggested a while ago that a fully enabled and sorted Fermi chip might use less power and got told in no uncertain terms that was an impossibility, so if it does come to pass it will just go to show how much (or little) people around here know.
  15. It all depends on when we will see the cards, doesnt it?
    I believe I said Id be impressed if nVidia did this in less time than the 2900 to 3870 timeframe
    From May to November, its already too late to have it done faster, and it should have been done last month, accordingly
    Later is better than never, tho, making a paper launch, if this happens, will have certain effects
    Hopefully nVidia has the cards ready, and the November rumors are real
    So, Im past being impressed, tho, throwing in the 40nm debacle, its still great for them to come soon, IF they do
  16. Mousemonkey said:
    I suggested a while ago that a fully enabled and sorted Fermi chip might use less power and got told in no uncertain terms that was an impossibility, so if it does come to pass it will just go to show how much (or little) people around here know.


    I dont recall saying it was impossible, but thats just the thing "fully enabled and sorted Fermi". This GF100B that was just going to add, not change anything, it would be almost impossible to lower power and increase performance. If this GF110 is a fairly changed arch, then no, it wouldnt be impossible.
  17. ares1214 said:
    I dont recall saying it was impossible, but thats just the thing "fully enabled and sorted Fermi". This GF100B that was just going to add, not change anything, it would be almost impossible to lower power and increase performance. If this GF110 is a fairly changed arch, then no, it wouldnt be impossible.

    Did I say it was you?
  18. Problem here is, nVidia needs this, and badly
    Not sure I believe Charlies take, or for that matter FUaDs either
    My earlier point should be taken seriously tho, especially by nVidia
    If they dicker around with the release times, and start the its coming soon, and it doesnt appear, then theyve hurt their user base in one huge aspect, trust
  19. Mousemonkey said:
    Did I say it was you?


    No, i wasnt saying you said it was me, just saying I dont recall saying it. I suppose the appropriate answer would be to say "Did i say you said it was me?" :lol:
  20. Sumthin I enjoy, a real who dunnit
  21. Staying on topic here, who would consider a full Fermi a 2nd gen DX11 card?
    Wouldnt scaling in DX11 have to go up as well?
  22. I have sanded all my digital fingerprints off. :)
  23. As far as I can see this is just a tweak with a new number on the GPU just like the ATi cards are a tweak but with a different name.
  24. It depends what the 580 is. If it is GF100b, then no, its no where near 2nd gen DX11, but if its GF110, then its in the same position as the AMD 6xxx series.
  25. JAYDEEJOHN said:
    You mean, were going to have to go thru this again?
    http://i645.photobucket.com/albums/uu171/jaydeejohn/tesafilm.png
    Hope it isnt wood screws n promises, as that reminds me of a country song

    wood screws is for noise as the wooden screws don't make noise when they vibrate
  26. We dont know what the 69xxs will bring, but the 68xxs were a nice tweak, 1120 vs 1600, smaller die and all
    If the 69xxs are just the same, Id say it was at least a big tweak

    But, I dont think itll turn out that way

    As for nVidia, we have nothing but a driver detail thats been pulled, so I havnt a clue whether theres been any tweaking at all, or just a more mature process, on both ends
  27. @ Izzy, I meant that
    nVidia would be complete fools to attempt this again
    Wanting your product out is different than having it out

    We all want a full Fermi out, but its up to nVidia to deliver
  28. JAYDEEJOHN said:
    We dont know what the 69xxs will bring, but the 68xxs were a nice tweak, 1120 vs 1600, smaller die and all
    If the 69xxs are just the same, Id say it was at least a big tweak

    But, I dont think itll turn out that way

    As for nVidia, we have nothing but a driver detail thats been pulled, so I havnt a clue whether theres been any tweaking at all, or just a more mature process, on both ends


    Well if they waited for the process to mature just to release cards at release 480 like heat and power, ill be pretty pissed, but if they are a lot more powerful and only need a little more power, i wont mind, or if they are a decent bit more powerful, and require the same or less power, then thats fine too. But it will be a failiure if they keep fermi like efficiency in the fact the produced about 2x more heat and used 2x more power than the performance upgrade they offered over the 5xxx series. If thats the case again, its a large failiure. Same thing for AMD. If the 6970 is 30% faster, but needs 30% more power, 30% larger die, and so on, i sure wont be impressed.
  29. I agree
    I think we may see some progress from the 68xx series also on nVidia cards, as much of whats been done is due to process maturity, on the power side, allowing fow higher clocks, thus same power usage
    What Ive heard on Fermi is, the more shaders they add, the more hotter it gets, thus no 512
    Going with a few tweaks, process maturity, enabling 512 at this time can happen, with reasonable power/heat
    I also agree with dropping the DP thing, as it does nothing for gaming perf, yet increases Fermis ability
    Im truly hoping for a 104 like full fermi

    PS I do expect better DX11 capabilities with the 69xx series
  30. JAYDEEJOHN said:
    @ Izzy, I meant that
    nVidia would be complete fools to attempt this again
    Wanting your product out is different than having it out

    We all want a full Fermi out, but its up to nVidia to deliver

    i'm sure if they came out with the engineering sample they would have still be fire branded on the net but that would have passed them ******* up like this will last alot long T_T

    At the DP thing i guess the folding crowd gets hot and heavy over that stuff.
  31. What will be interesting is if people doubt the 69xxs as being a new gen
  32. JAYDEEJOHN said:
    What will be interesting is if people doubt the 69xxs as being a new gen


    It appears the 69xx is a lot more "new gen" than 68xx was, as 68xx was more of taking away inefficiencies. 69xx will actually be adding and changing to my knowledge.
  33. The 68xx seems to be a hodgepodge of previous componants of earlier chips, which can or cannot be considered new gen, depending.
    But changing the very structure of shaders hasnt been done in awhile from the red team, at least, not to this stature
  34. renz496 said:
    lol, is that a joke? :D . btw if i'm not mistaken when they say 16 times performance its not the game will run 16 times faster but you can get 16 times more DP compared to current Fermi architecture.


    Oh I wish it were a joke. honestly, judging by how nvidia cards have been going, every generations top end card is requiring more and more power. They sure know how to fit more transistors on a chip, but it costs in power. Any claim they make that their new gpu will require less power is bullplop. Ill believe it when I see it.....
Ask a new question

Read More

Graphics Cards GPUs Next Generation Nvidia Graphics