It would not shock me if AMD also ups their prices this time around. After the increase with Fiji cards, who knows what will happen.
It would not shock me if AMD also ups their prices this time around. After the increase with Fiji cards, who knows what will happen.
Gtx980 is almost 2 years old crap.Why they need charge 350USD for 16nm GPU with same performance?
That gpu can be around 200mm2 big at 16nm to match GTX980.
Btw x60 GPU always cost 250USD max.
AMD better to have tons of 14 NM cards on their hands... Because tons of people will switch to them... NVIDIA seems that is the one who will lose some market now.
Hopefully, both will have a MASSIVE advantage against Intel HD solution.
If AMD has a winner on their hands w/ Polaris, they'd better darn well charge what they're worth. I would gladly pay top dollar for an AMD GPU if it gives me top dollar performance, and I am sure a lot of other enthusiasts would too.
its a double edged sword pricing. I would say price as high as possible (since I am not buying on first year) to make good profit.
But they could also cut into nvidia's sales by pricing lower. Some people might see the AMD prices, and wait for nvidia's entry while not realizing AMD will drop prices when nvidia releases.
If they are months ahead they have the flexibility to choose right. If they remain fastest option after nvidia launches, well...
They can also be refining their offerings to counter nvidia's later cards. Though not sure how much can be done in a few months.
Nooooooooooooooooooooo!...Oh wait I could care less.
How much less could you care?
Because not only has anything like a die shrink ever happened to cars, such events are commonplace.
So, is this why Newegg was blowing out those MSI GTX950 2GB GDDR5 OC cards for $119.99 a few days ago?
Anyways, I bought several. Did I get hosed, seeing as how Pascal is supposed to land in a few months? Would there have been a $120 Pascal model that would have been faster? At least the cards could be good for HTPC usage too, besides mid-range gaming, since they have HDMI 2.0 and HEVC Main10 support. I suppose I could always re-sell them if I'm not happy with them.
That's the wrong comparison. You're comparing GM206 with GK110, a chip two sizes bigger. Also, Kepler->Maxwell didn't involve a node shrink - frankly, it's a minor miracle that Nvidia managed to squeeze out the performance improvements they did.
Usually, the pattern is that a new-generation chip matches the performance of an old-generation chip one size bigger. For instance, GTX 680 (full GK104) beat GTX 580 (full GF110) by about 20%. GTX 660 (full GK106) beat GTX 560 Ti (full GF114) by roughly 22%.
The original claim was that "1060 should be around GTX980 performance.". GTX 980 (non-Ti) is the full GM204 chip. Assuming that by "1060" he means the full GP106 card, the rule of thumb listed above indicates that it should beat GTX 980 by about 20%. And a "6"-series Nvidia card hasn't traditionally started any higher than $249.
? The 960 didn't have a die shrink vs a 780 either though, and wasn't Maxwell's biggest jump in performance either.
This next gen stuff is getting ~2x perf/watt from the die shrink alone.
As for specific models, especially the big ones? No, no one knows Big models more likely to be later than earlier though.
I don't see Nvidia losing any market share unless AMD has a better performing GPU than Nvidia once all the chips fall. If my guess is wrong in my last post and Nvidia does hold off on the GP100 for consumer cards (even though they are the first chips they are making) until Q1'17, there will be quite a few hold outs that will wait and see which is the better card (look at the recent 980ti/Fury releases for a recent example of plenty of hold outs until both cards were out where in Q2'15 the market sales of cards dropped to the lowest level in 10 years, and in Q3 the sales rebounded 9%... basically most people waited to purchase a card even though they could have gotten a 980ti in Q2'15, they waited to see what the Fury was going to do performance wise, and then bought their cards in Q3 instead).
But I don't see Nvidia sitting on their high end card (for which they are already making the GPU for use in HPC/enterprise cards) if AMD has a card that can significantly outperform whatever Nvidia's top consumer card on the market at that time. Nvidia's game is to maximize profits on their entire product line, which means only releasing something when there is a compelling MARKET reason to release it (in other words a competing product is on the market or extremely close to being on the market, or declining sales on existing products due to product stagnation).
You may also be right that Nvidia may lose some market share, but their numbers can really only go down anyway:
- I think AMD's gameplan is fairly straight forward: release first, charge what the market will support, wait for NV to release Pascal and (this is the most important part) undercut Pascal on launch day so reviewers will bring it up in all of their Pascal launch reviews.
Basically do what NV would and has done to AMD.
I'm going ultra wide with my monitor, and I might be buying it sooner than I thought (as older models start to clearance out, I might just pick one up).
My 980 Ti struggles to do 3840x2160 in some games, so I wonder how it would handle 3440x1440.
5-6 months seems so far away :'(
3840x2160 is 1.67x the pixels of 3440x1440. You should get a big speedup in moving from the former to the latter.
Sorry about the wrong comparison my bad. Here is the correct one gtx780 vs a gtx960 http://www.anandtech.com/bench/product/1596?vs=1493
I would be extremely surprised if you get gtx 980 performance in a gtx1060 (or what ever the name will be) card.
Is there some way to put a reminder on this post for 6 months time?
Still the wrong comparison. GTX780 is a cut down 561mm² GK110 die. GTX680/GTX770 is a full 294mm² GK104 die, so that's the comparison you'd want to use for the 227mm² GTX960.
http://www.anandtech.com/bench/product/1596?vs=1494