Thursday, March 1st 2012
Kepler Unbeatable: NVIDIA
The tiresome wait for NVIDIA's next-generation GPU is drawing to a close. Or so suggests a Facebook wall post by NVIDIA Italy, which reads (in Italian, of course):
Source:
XtremeSystems Forums
Aspettando Kepler... pazienza, pazienza, pazienza che il momento giusto arriverà, e allora... non ce ne sarà più per nessuno! :-)That can be translated as "Waiting for Kepler ... patience, patience, patience, the right time will come, and then ... it will be unbeatable (sic)." From various sources we're hearing that there will be hectic activity surrounding the launch of NVIDIA's next-gen GPU in the weeks to come.
165 Comments on Kepler Unbeatable: NVIDIA
And we are talking about the part that is to compete with HD7970 obviously. GK104.
@claylomax
No I'm not joking, it's 2 months. Or are you talking about the "reaaaaally sloow" thing? Then yes it's irony, but not only necessarily pointing at you. Some people are saying that Kepler comes half a year later, so clearly time passes slowly for them.
EDIT: YOU were joking, using sarcasm, no? I think I get it now. Or not. Whatever it is, I blame british humour.
But I'm confused, I thought the Kepler high end single gpu was to be launched at the end of the year; so what's coming in April then? A stopgap solution.
And yeah british humour, no one gets it. :D
wccftech.com/nvidia-kepler-gk104-geforce-gtx-670ti-arrives-march-2012-performs-gtx-580-hd7950/
(* GeForce FX? HD2900? Bulldozer? etc etc) All of them, exceptions that just proof the rule, and i'm talking about nvidia.
And wtf? 3 months? GK110 comes in april? you sure? i don't think so.
(it's only 15% faster.) Cayman uses the same 40nm, so that was to be expected.
EDIT:(read latest posts), i was talking about GK100, which as you say comes in may - june "almost half a year"
GK104, "gtx680", will be "faster" (not that you can notice it), like 15% more than 7970, so it won't blow it out of the water, that's great because the price won't go up. (i based on the speculations that gtx670ti will be between 7970 and 7950, gtx680 won't be like MUCH faster than its sibling)
And then the question, will gk104 be as good at o/c as hd79XX?, seeing that 7950 can go as high as 1100 Mhz that's 40% give or take.....
For one i would be so much happy to see the gtx670 ti at U$299, and faster than 7950, that would mean hd7870 U$250 atm.
But GK100/110 is not the chip that will go against HD7970, that one is a lot faster and more heavily oriented for compute, so I don't know why anyone would want to compare them. The competition to Tahiti is GK104 (same die size) and again according to rumors it is significantly faster. 15% is not much? lol then I guess HD7970 is worthless to you and so must be HD7950, but you are apparently praising them, so I'm confused.
Regarding GTX670 or 680, etc. No one is sure of what the name for the highest binned GK104 SKU will be, but that's the one that is suposed to be 45-50% faster than GTX580, not any GK100/110 based card. So "Kepler" as in cards that will compete or beat HD7000 gen of cards will launch in March/April, so definitely NOT 6 months later.
GK104, two gpu's gtx670ti and gtx680(read the link) or whatever, do you really think there will be like a 40% distance between them? really?, at the most 30%(very unlikely), and hd7970 being faster(leaked info says gtx670ti is faster than gtx580 and 7950 NOT 7970) than gtx670ti, why would you think gtx680 would be a LOT faster than 7970?.
I'm not "praising" anything, i'm at the side of whoever offers me the best bang for the buck, i've had both.
You think 15% will make a difference? LOL, can you see differences between 30 and 34 fps?
If you take a look at my pc you'll see i have a 2.5 year old card, because i always get something that offers me at least 250% and the same or less price, but for some reason in the last 2.5 years nothing like that has came out.
I so don't care about who beats who, BUT i think you'll agree that they have to be close in order to keep prices in check. If one of the two is too far from the other, then prices skyrocket. no good for anyone of us but for nvidia or amd.
Take a look at this, they compare the highest binned gk104sku(gtx680) against gtx580, 7970 and 7950, guess what? 13.8% "faster"(not that you'll ever notice, but faster nonetheless) than 7970 at 1080p ;), will it widen or shrink at 2560x1536? dunno.
wccftech.com/nvidia-kepler-gk104-gaming-performance-figures-exposed-faster-gtx580-hd7900-series/
And if those are nvidia handpicked benchmarks.............go figure.
Oh, let me cite your source for you...
we.pcinlife.com/thread-1831876-1-1.html
Do you really trust a person posting on a forum without a source? I call shenanigans.
Anyway, based on the rumors of the 670ti performance, it seems, 680, will be slightly faster than 7970.
Why is it so many people want one of the to brands to be so superior to the other?
Don't you see that'll only benefit those companies? ffs....
Also you are using very very old info for your performance figures, and those charts are suposed to have been demostrated fake. I'm going by the recent news appearing here in TPU in the last couple days and a comment by Kyle Bennet from [H] who said it's (up to) 45-50% faster than GTX580. I had a hard time believing that it is so much faster, at first, but it's not imposible considering the specs (2x the Tflops, 2x the texturing power, etc.) so that's why I included that info in my post.
There's been so many rumours that at this point GK104 could be slower than HD7800 or up to 50% faster than GTX580. Since rumors tend to be more accurate as we come close to release, I'm going with the latest info, and that's 50% faster, but bear in mind I do not personally believe that's the case, although I made my calculations based on specs some time ago and it's within the upper range of what's posible.
PS: And yes I notice a massive difference between 30 and 34 fps. At low fps every single one of them improves the experience incredibly. Every extra frame is a blessing. And by your logic AMD shouldn't have bothered with HD7970 or HD7950 because they're "only" ~15% faster that GTX580 and HD6970, respectively. That's why I made that comment about praising them. In the end you'll always find a card that is 10-15% apart from one another, so by your logic, if you don't see a difference between 30 and 34 fps, you'd see any between 27 and 30? And what about 25 and 27? How low can we go (where do you draw the line)? Oh but maybe someone (many actually) can't bare anything below 40, 50, 60... and it's going to be different with every game too, so what's enough? Basically what I'm saying is that a card that is 15% faster than another one, it's 15% faster always (on average) and it has it's place.
Personally I don't spend that much, far from it, but even in my price range I'd pay x% more for x% more performance. IMO your case is a falacy, like I said because there's no card that will give you 30-34 fps in all games. On some games, a particular card might be enough, in others it won't and let's not start talking about settings. So like I said a card that s faster it's faster and always will. It's not my bussiness or your bussiness to decide if those extra $100 are worth it for the people who are willing to pay $400++ for a card.
When you say "anyone who will pay $500 to go from 30 to 34 fps in most of his games an utter fool" you are calling a fool to ALL enthusiasts, because that's what you get. A 15% increment no matter if it's at 30 fps or at 200, its a substantial difference and worth paying for some people. At 30 fps is going to be ever more important than at 200 plain and simple.
According to your logic overclocking is useless, because you'll never achieve much more than 15% more actual performance and the best you would do is obtain those extra 3-4 fps.And this and what I'm saying are not mutually exclusive either. I'd do the same, but that does not mean that a 15% difference is meaningless. If you have to choose between HD7970 and GTX580 for the exact same price (and you can pay that price) you'll go with the Radeon. If the difference is 15% or less, much of the same.