Friday, July 11th 2014

NVIDIA Moving Around the Fabled GeForce GTX TITAN II
NVIDIA is moving around engineering samples of what it describes as "GM200 A1 graphics processor," in its shipping manifest. The sample was making its way from Taiwan, to Bangalore, India, from where it's likely pushed to the company's facilities in Bangalore and Hyderabad. A1 steppings of NVIDIA chips are usually pre-production, and bound for just a few more rounds of testing, before being upgraded to "A2" and mass-produced. German tech site 3DCenter.org also pulled out some likely specifications from its sources.
To begin with, the GM200, like the GM204, will be built on existing 28 nm silicon fabrication process, as both NVIDIA and AMD appear to have suffered design setbacks due to their common foundry partner, TSMC, not being able to set its next-gen 20 nm node up to speed in time. The GM200 is expected to feature over 4,000 CUDA cores, although the exact number is unknown. It is expected to widen the memory bus to 512-bit. Given the existing process, the GPU will be huge. Over 600 mm² huge. NVIDIA will probably bank on the energy efficiency of its "Maxwell" architecture to cope with thermal loads put out by a chip that big. The GM200-based "GeForce GTX TITAN II" could launch in the first half of 2015.
Source:
3DCenter.org
To begin with, the GM200, like the GM204, will be built on existing 28 nm silicon fabrication process, as both NVIDIA and AMD appear to have suffered design setbacks due to their common foundry partner, TSMC, not being able to set its next-gen 20 nm node up to speed in time. The GM200 is expected to feature over 4,000 CUDA cores, although the exact number is unknown. It is expected to widen the memory bus to 512-bit. Given the existing process, the GPU will be huge. Over 600 mm² huge. NVIDIA will probably bank on the energy efficiency of its "Maxwell" architecture to cope with thermal loads put out by a chip that big. The GM200-based "GeForce GTX TITAN II" could launch in the first half of 2015.
61 Comments on NVIDIA Moving Around the Fabled GeForce GTX TITAN II
Wouldn't you think the GTX480 Kepler replacement is the GTX780 non Ti?
Or my i7 3930k is a gimped i7 3960? No, it's designed that way to create a product division.
Whether we like it or not, technology produced on nm processes will always allow vendors to sell products based on 'castrated' products to create a product series. The 7950/7970 or R9 290/R9 290x is the perfect example of this.
This isn't about nvidia dumping bad inventory, it's about a company making money by fusing cores and reducing costs on imperfect wafers. They both do it.
These pr leaks are a tad vague these days imho, also a few sites are throwing around a 2015 release date , now surely that can't be right especially on this node.
Honestly I think what Nvidia does with their GPUs, atleast IMO was smart in a business standpoint. For instance look at the 780 to 780Ti. They release the 780 with a few SMX clusters disabled. Wait for AMD to counter, then then release a full fledge GK110 GPU to compete with whatever AMD release (290/290x) to maintain the performance crown.
Its kind of been this way for a long time, and both companies do it.
GF110 = GTX480 perfected, the GTX580. It was the best performing card.
GK100 = Unproducable at the time. so GK104 becomes GTX680 (beats 7970)
---- 7970 Ghz appears, can be seen to beat GTX680----
GK110 = Titan appears, rules them all at a silly price. Then comes GTX780.
---- R9 290X appears, can be seen to best 780/Titan in many cases.
GK110 Revision B = 780Ti and Titan Black.
I don't see Nvidia dumping any inferior GPU's. Seriously, if you call GF100, GK110 and GK110(B) inferior GPU's then wtf have AMD got? Inferior requires a relative product to be related to. Nv have been on top for years now (unfortunately for pricing).
When wafers are made there are always poorer yield parts at first, until the process is refined or the arch is tweaked to fit the process. It's not about dumping inferior anything. It's a known development process and this is a product of it. I know if i make 'X' wafers, I will not get 'X' 100% parts. Those that don't make the grade become the lower models. AMD and NV both do it. If anything they are intentionally scaled down GPU's to fit a known production yield.
The only true dumping that gets done is when both companies rebrand last gen cards as lower brand current gen. That's intentional dumping.
But hey, after having (over the past two years) 2 x 7970's, a Titan and a 780ti, I feel pissed to know I've had inferior GPU's.
But tbh, you're probably making the same point I am but speaking a different semantic. I'm viewing it as a production process/marketing necessity and seeing it from a neutral stance (both camps do it). You seem to imply NV is only doing it - which would be very wrong. EDIT: No need to apologise :toast:
Although the 7970 and 7970Ghz edition are the same card, AMD just bumped the core clocks on the reference models.
I don't think either company comes out smelling like roses in either instance.
It is the intelligent way of hauling success out of assured failure , and every chip company does it (all have a failure rate too)but they don't always release in the same way.
Find it pointless to spend X amount of money on a GPU and a 4K monitor only to gimp it. People do stranger things though.
(And I'm an AMD guy)
Both AMD and Nvidia's card support 30hz 4k quality. Nvidia have simply found a way to stream it at 60Hz. Of course it's a reduced quality image - it's a HDMI bandwidth limitation. You make it sound as if Nvidia are making up for a gimped card.
Both AMD and Nvidia use displayport for 60Hz
It was a marketing fail not matter how you try to justify it. :)
Also im getting confused because the thread is supposed to be about a rumored Titan 2? Why are we stuck debating the 680 and 7970 at this point...
Lot of people no understanding more for their price. Lot of people have GTX780Ti SLI, Titan SLI,
Titan Black SLI very strong cards and if NVIDIA show up with something 1000$ and AMD offer 5% weaker for 600$ I think that would be one bed year for NVIDIA.
GF110 cost 500-600$, GK110 700-1000$, now people expect GM210 more than 1000$.
I think NVIDIA shouldn't ask more than 800$ for premium card and depend on time when they decide to launch and price more or less people will buy.
That chip would be my next choice, but depend of price. I will not look some locked cards as GTX780 any more.
I think this time Titan II will be not be only reference card. We can expect maybe some EVGA Classified with extreme fabric OC with ACX cooler.
Than I can OC only Haswell E because GPU will be OC by fabric.
As far as textures. That will depend on the game devs if they want to up the quality of models and textures to coincide with 4k resolution. Don't expect a rapid transition since consoles are at 1080p PS4 and below ? X-Box One. Doubtful they are using optimize textures and models for there respected resolutions and adapting it to a higher resolution hasn't been the PC gaming industries strong suit with all the poor ports PC users end up settling for.