Greenland is already known for marking a major upcoming GPU Radeon 400 series “Arctic Islands”. According server MyDrivers.com should really go for GPU, which will be produced by 14 nM LPP process developed by Samsung, but he also has a company GlobalFoundries. However, this information is not yet guaranteed, so it is conceivable that in the end really will not be a GPU manufactured 16Nm + FinFET process from TSMC. Another piece of information talking about the fact that Greenland will be produced from the second quarter of next year and the resulting cards should enter the market by the end of the holiday season, that is just for shopping season known as “back to school”. It is therefore rather low season “back to school and home PC.”
Next-generation chip Greenland Arctic Island will also contain still Baffin, Ellesmere falling but logically focuses far less attention, especially since it is expected to fight hi-end AMD and NVIDIA cards. They have to arrive on the market in the same period, which has recently been very common.
It should already be long decided where and how the process can leave the next generation AMD GPU manufacture so that these unwarranted information may be true. Indeed, the actual CTO Mark Papermaster AMD in November, confirmed that his company has received from glof and 14 nM GPU samples that have been identified for further examination. Besides the server MyDrivers.com this report also mentioned Korea Electronic Times, which speaks about the fact that production of the new GPU is not taking a glof, but also Samsung, more so its System LSI Foundry Business Department. Promised the 28nm process compared to double energy efficiency, that is performance per watt, so while keeping the consumption of the new cards should be doubly powerful.
We have therefore slowly begin to address whether it will be thanks to AMD 14 nM process some advantage, though NVIDIA has confirmed that Pascal will 16Nm chips. Potential customers will be especially interested in the performance, and perhaps also a little consumption, but producers there are more variables to consider, such as cost of production, and with it tied the yield and the ability to market the required number of pieces. In the case of the memory on the AMD and NVIDIA should be similar, since both will have the HBM2 for high-end and GDDR5 or GDDR5X for other cards.
Possible to use already known comparison of the processes from which it is seen that there is 14 nm, such as 14 nm. In terms of size consisting of elements that make up the logic chip and know that + 16FF companies TSMC and Samsung 14LPE companies and glof to each other in this regard close. Due process 14LPE but AMD will have a better opportunity to make a single chip telescoped more transistors, while speaking to about 18 billion, or twice the Fiji today. It is widely expected that a new generation of these cards finally bring gaming in Ultra HD resolution into the mainstream. Memory HBM2 turn allows to deploy one card up to 32 GB of memory, which is particularly important for the next generation of FirePro cards. Besides all this you would have AMD may prepare for next year even more special for the upcoming APU powerful Nintendo NX.