Read my updated post, direct comparison shows 150% better performance and 150% better power consumption when compared 28nm transistors to 16/14nm Finfet transistors :banana:
For both really. The are both using 16/14bm FinFet. At least AMD has shown some hardware, for Pascal we only have leaks and no release dates.
Last I heard was that Polaris was likely gonna be launching before Pascal. (That was on the rumour thread on notebookreviewforums.) Yes, good for both Polaris & Pascal, I'm looking forward to the reviews & launches!
Excel Sheet summing the speculated/expected information regarding the new GPUs while filling the Blanks: Memory Type, TDP, Processing Power with Personal Touch. Nothing to fuss about until the true specs, all we can do now is to hope for bigger numbers and* better technologies. And I Hope Nvidia will start adopting new technologies to move the market to a better place instead of leading marginal portion of it in stagnation like the DirectX fiasco. If you remember how they refused to support 10.1 until DX11. Let's hope they will make different choices and enable the New DX12 API to flourish and with it it's adoption.
I hope the 1080 based card has more than 6gb of VRam and it's GDDR5x. The rest of the specs seem fine to me. However we will have to wait for an official statement from Nvidia.
There doesn't seem to be enough production capacity for mainstream NVIDIA-card levels of production for GDDR5X. That means that either: a) Pascal won't have GDDR5x (which would fit with these specs). b) Pascal will be available on very limited quantities. c) Pascal will be really late. d) Pascal will be a paper launch. I personally see a combination of a and d.
I think we're all aware of that. However, my point is based on the fact nvidia have never used an X in their product names like that, unless I'm mistaken, only AMD do that and nvidia aren't stupid enough, and know better than, to name their card models too closely to the competition. Mark my words, they'll start a new naming scheme with Pascal, I've absolutely no doubt about that, so this chart must be fake, it's that simple to me! :grin:
You do not need to read the tech specs to realize its a hoax. From a marketing point of view why should NVidia introduce the Ti version at the same time as the high and low end version of the new silicon? When giving innovation to customers and like any recognized brand name NVidia always followed a typical path by introducing 1 or eventually 2 levels at the same time but never 3 ! (Never confuse the customer, reduce communication complexity and costs, keep it simple). Furthermore the Ti name has a strong recognition as signaling a valuable mid-life update of an existing line. Why wasting this opportunity from the beginning when introducing a new line?
The best part is the price, X80 - 600€, X 80 Ti - 900€ and X80 Titan - 1300€. I made it totally up, but sadly it will be not far off.
150% better performance at half the die size is ridiculous, this must be wrong. 150% better performance at half the die size means mean's a 300mm^2 16nm finfet chip = 2.5X 600mm^2 28nm planar chip That's too much, unless you're seriously expecting GP104 to be over twice as fast as GM200
Pascal will be irrelevant :banana: With the STARDOCK technology with dx12 all the new cards will be a moot point. You can mix AMD cards with NVIDIA cards and not bother with sli nonsense. Sooooo you can mix with what you can afford, then pick another card when you feel like it and not be at the mercy of $1300 bllsheet cards. Love it the consumer finally wins!!!!!!:banana: