The latest news giving us a clear indication that NVIDIA will reveal a new GPU which will be based on full Maxwell GM200 Core. Most probably in the light of previous rumors new GPU core will power the NVIDIA GeForce GTX Titan II or NVIDIA GeForce Titan X. We will be able to see the debut of this Graphics card at GPU Tech Conference. The event will start from 17th march to 20th of March. There are also lots of chances that the new GPU will be Quadro M6000 but the recent leaks are more inclined to the GeForce Titan side.
NVIDIA GM200 GPU Core is arriving on March 17-20 at GPU Tech Conference
The GM200 core pictures have also leaked in the wild and from above picture you can see that the GPU included 12 Hynix GDDR5 memory chips with a capacity of 8 GB (1 GB) and 32-bit bus, and therefore Graphics Card has at its 12GB of VRAM GDDR5 384-bit interface. There is no doubt it is in like its nick Big-Daddy it is also a big Maxwell monster die the recent analysis give us the clue that it would be over 600mm² large. This means that NVIDIA may have approached the reticle limit of TSMC’s 28nm process.
In Kepler era, NVIDIA GK110 was the first large core chip in the market for professional card and then also for the consumer market, GM200 in general as well. Currently, specifications of GM200 core are not clear, but the unofficial information is indicating that the Maxwell GM200 core will have 3072 CUDA core, 384bit memory interface, with 12GB GDDR5 memory.
With this card you may find high-end performance with low power consumption, low heat and even “Fanless” environment, Yes! The card fan will stop at IDLE mode. The Huang family in Silicon Valley technology puts it into a reality. Some brands as above mentioned implement this technology in recent NVIDIA GTX 980 and NVIDIA GTX 970, but Maxwell architecture also allows this technology to play a greater role, from TechPowerUp latest news pointed out that:
Future drivers accompanying the GTX TITAN X will offer a new feature, which when enabled, lets you choose between a non-linear fan curve that keeps the fan off; and one that runs it at low speeds. This should let the driver power the fan of a GTX TITAN X completely off, until it reaches a temperature threshold, and only then begin to ramp up speeds. It could help not just idle (desktop), but also light-gaming scenarios (think the League of Legends).
We don’t really know what naming structure will eventually take place, but we might see something like X80 and X70, with the GTX prefix going back to its original suffix location. So something like an NVIDIA Geforce X80 GTX, or NVIDIA GeForce X70 GTX where the X represents a 10 as with Roman numerals. For subsequent generations NVIDIA could simply add an additional numeral such as X180 or X280. I’m just brainstorming here and would love to hear what you think Nvidia’s future naming structure should look like. Please share your thoughts in a comment below. Not just for GM200 cards, but for future Pascal based cards as well.
There are also lots of chances that this time NVIDIA will go with new naming structure. We don’t generally comprehend what naming structure will inevitably happen, yet we may see something like X80 and X70, with the GTX prefix backtracking to its unique suffix location. So something like an NVIDIA Geforce GTX X80 or NVIDIA Geforce X70 GTX where the X speaks to a 10 as with Roman numerals can also make its debut. For ensuing eras NVIDIA could basically include an extra numeral, for example, X180 or X280. I’m simply conceptualizing here and would love to hear what you think Nvidia’s future naming structure ought to look like. If you don’t mind impart your musings in a comment section. Not only for GM200 cards, however, for future Pascal based cards too.
Source: WccfTech, sweclockers