A Phoenix from the ashes
18th April 2006
The graph below is one of transistor count, not die size. Inevitably, on the same manufacturing process, a significantly higher transistor count translates into a larger die size. But for the purposes of this article, all I need to show you is a representation of transistor count.
See that big circle on the right? That's Fermi. NVIDIA's next-generation architecture. NVIDIA astonished us with GT200 tipping the scales at 1.4 billion transistors. Fermi is more than twice that at 3 billion. And literally, that's what Fermi is - more than twice a GT200.
At the high level the specs are simple. Fermi has a 384-bit GDDR5 memory interface and 512 cores. That's more than twice the processing power of GT200 but, just like RV870 (Cypress), it's not twice the memory bandwidth.
The architecture goes much further than that, but NVIDIA believes that AMD has shown its cards (literally) and is very confident that Fermi will be faster. The questions are at what price and when.
The price is a valid concern. Fermi is a 40nm GPU just like RV870 but it has a 40% higher transistor count. Both are built at TSMC, so you can expect that Fermi will cost NVIDIA more to make than ATI's Radeon HD 5870. Then timing is just as valid, because while Fermi currently exists on paper, it's not a product yet. Fermi is late. Clock speeds, configurations and price points have yet to be finalized. NVIDIA just recently got working chips back and it's going to be at least two months before I see the first samples. Widespread availability won't be until at least Q1 2010.
I asked two people at NVIDIA why Fermi is late; NVIDIA's VP of Product Marketing, Ujesh Desai and NVIDIA's VP of GPU Engineering, Jonah Alben. Ujesh responded: because designing GPUs this big is "fucking hard".
Jonah elaborated, as I will attempt to do here today.
I'd love an uploader that actually freaking works... Fermi Whitepaper
1st January 2005
There are others whom claim Nvidia has dispelled any rumors of the project being late though, which supposedly was a rumor spun primarily regarding them not being able to effect a 40nm yield, and totally false. As tight lipped as Nvidia is being to the general public, we won't really know until they officially announce it's release. In fact I would argue is ANY DX11 GPU really "late", given that we don't even have an official retail OS release or games supporting DX11 yet?
I respect AnandTech's technical knowledge and avid interest in informing the public, but amongst their informative articles often times they get things wrong too or say things in a way that is misleading to consumers. They did at least offer up a quote, but it could be Nvidia has found they are on better schedule than they originally thought or even that they are leading on about.
On the architecture of the chip itself, sure it is going to have a significantly larger surface area than ATI's, but you have to wonder if ATI using ANY size of chip would be able to handle the heat and power consumption of a 3 billion transistor chip as well. At any rate, whether late '09 or early '10, this monster is going to be worth waiting for.
As I have said before, originally I was holding out for the rumored 1 billion transistor GPUs. This will be like having three 1 billion transistor GPUs in tri SLI, but occupying only one Pci-Ex slot. An amazing achievement to say in the least.