Translate to multiple languages

Subscribe to my Email updates

https://feedburner.google.com/fb/a/mailverify?uri=helgeScherlundelearning
Enjoy what you've read, make sure you subscribe to my Email Updates

Friday, September 15, 2017

Beyond FinFET: The Research Alliance’s Plans for Microprocessor Evolution | Evolving Science - Computer Science & Technology

Photo: Deirdre O’Donnell
"FinFET-based computer processors are one of the great breakthroughs of this age. Their form factors and architectures have allowed for the development of nearly every chip in nearly every device that enhance and sometimes define our lives today" argues Deirdre O’Donnell, professional writer for several years. Deirdre is also an experienced journalist and editor.

Photo: Evolving Science

Without these silicon microstructures, computerised gadgets as we know them might not have progressed to the relative power and portability they demonstrate as productivity and media-consumption tools. However, FinFET does have upper limits, particularly in terms of transistor density over a given area. Therefore, if computer processing is to get any stronger or more complex, processor architecture will inevitably have to move beyond this historical threshold and take on a new form. A collaborative effort between IBM, Samsung and other companies may have produced the next generation of transistors that will miniaturise processors beyond the traditional 7nm barrier.

FinFET Structure and Some of its Functions 
FinFET is the transistor standard that has revolutionised the form factors, computing abilities and power consumption of computer processors. It enables chips to have the small footprints we’re all familiar with, as well as their relatively astronomic electronic speeds that modern users take for granted. Without FinFET, consumer-grade electronics, particularly lighter and thinner device types including laptops and smartphones, simply could not exist. The transistors made based on FinFET architecture take the form of non-planar rectangles, or ‘fins’, of silicon atoms at the nanometric scale, stacked alongside one another. The fins can switch between states such as on or off very easily, while using ever-decreasing amounts of current at very high efficiency. This is the essential basis of modern data transfer and storage.

The Limits of FinFET 
Modern increments in processor speed and power depend on the steadily-increasing density and complexity levels at which the fins can be ‘stacked’. However, these properties inevitably meet ceilings, which are mainly determined by the atomic structure of silicon and the forces acting on it. This translates into a real-world limit of 5nm per FinFET-powered chip. However, because of other physical limits, mostly those of current flow between fins, these processors would be beaten by other form factors of a superior architecture on a chip of the same size. For example, silicon ‘fins’ are relatively static, which means they cannot adjust variably to the presence of current. Therefore, the best available FinFET-based processor units, or ‘nodes’, are about 10nm in size, whereas there is potential for more processing power per nanometer in the presence of improved silicon-based transistor types.

5 nanometer transistor -- how they did it


In other words, FinFET is still limited as to how many transistors can be packed into each nanometer of processor size. Currently, the best ratio that industrial research has produced equates to about 20 billion transistors across a 7nm node. This impressive array was developed by the Research Alliance, a consortium of investigators who have come together from major multinationals such as Samsung, GLOBALFOUNDRIES and IBM to assess the next directions for processing technology and how they may be realised. That was nearly two years ago, however. Currently, the Research Alliance has presented the debut of their new, 5nm node. This new device has increased the transistor number to 30 billion, and may represent the new generation of large-scale computing.
Read more... 

Source: Evolving Science and IBM Research Channel (YouTube)