TSMC launches 16nm FinFET line
December 13, 2013 | 10:29
Companies: #globalfoundries #taiwan-semiconductor #tsmc
Taiwan Semiconductor (TSMC) has announced that it will begin manufacturing of FinFET-based 3D chips using a 16nm process before the year is out - albeit only for small quantities of specific parts.
The race to the next process node, where the size of and distance between components on a semiconductor shrink, is a key part of Moore's Law - but something companies have been struggling with of late. Intel has publicly admitted that a delay to its next-generation chips is a direct result of the difficulties in shrinking process nodes much further, and that's a company not-inconsiderably ahead of its competitors: Intel's next-generation Broadwell chips will be based on a 14nm process with three-dimensional Tri-Gate Transistor technology.
TSMC, meanwhile, claims to have completed prototype work on its own next-generation process node with a view to starting small-scale production before the year is out. Like Intel, TSMC is boasting of three-dimensional components - fin-based field-effect transistors, or FinFETS, which drop the voltage required by the transistor while reducing current leakage compared to traditional planar transistors - but at a larger 16nm node.
The new process node marks the first time TSMC has offered FinFET to its customers, which industry group Common Platform has previously targeted solely for the 14nm process node - but it does so significantly after competitor and AMD spin-off GlobalFoundries, which announced a FinFET-based hybrid process back in September 2012 using 14nm transistors on a 20nm interconnect base - a hybrid model it dubbed 14XM, or Extreme Mobility.
TSMC made its announcement at the International Electron Devices Meeting earlier this week. According to coverage of the meeting by Nikkei, the move to a 16nm process means a drop in power consumption of 55 per cent or a boost to performance of 35 per cent compared to the company's existing 28nm product line. The first 16nm FinFET-based products will, unsurprisingly, be system-on-chip (SoC) components destined for mobile devices.
Like the company's existing 20nm line, however, production is going to be sorely limited with only small quantities of parts being produced at 16nm going into 2014. When the node will be ready for wider adoption, TSMC isn't saying - but the company has had previous issues with manufacturing difficulties and a lack of capacity at its 40nm and 28nm process nodes which resulted in customer complaints - meaning that it will have to balance hitting mass production soon enough to be relevant with getting the manufacturing process stable if it wants to avoid a repeat of past mistakes.
So far, TSMC has not released the names of any customers signed up to the first 16nm production runs.
The race to the next process node, where the size of and distance between components on a semiconductor shrink, is a key part of Moore's Law - but something companies have been struggling with of late. Intel has publicly admitted that a delay to its next-generation chips is a direct result of the difficulties in shrinking process nodes much further, and that's a company not-inconsiderably ahead of its competitors: Intel's next-generation Broadwell chips will be based on a 14nm process with three-dimensional Tri-Gate Transistor technology.
TSMC, meanwhile, claims to have completed prototype work on its own next-generation process node with a view to starting small-scale production before the year is out. Like Intel, TSMC is boasting of three-dimensional components - fin-based field-effect transistors, or FinFETS, which drop the voltage required by the transistor while reducing current leakage compared to traditional planar transistors - but at a larger 16nm node.
The new process node marks the first time TSMC has offered FinFET to its customers, which industry group Common Platform has previously targeted solely for the 14nm process node - but it does so significantly after competitor and AMD spin-off GlobalFoundries, which announced a FinFET-based hybrid process back in September 2012 using 14nm transistors on a 20nm interconnect base - a hybrid model it dubbed 14XM, or Extreme Mobility.
TSMC made its announcement at the International Electron Devices Meeting earlier this week. According to coverage of the meeting by Nikkei, the move to a 16nm process means a drop in power consumption of 55 per cent or a boost to performance of 35 per cent compared to the company's existing 28nm product line. The first 16nm FinFET-based products will, unsurprisingly, be system-on-chip (SoC) components destined for mobile devices.
Like the company's existing 20nm line, however, production is going to be sorely limited with only small quantities of parts being produced at 16nm going into 2014. When the node will be ready for wider adoption, TSMC isn't saying - but the company has had previous issues with manufacturing difficulties and a lack of capacity at its 40nm and 28nm process nodes which resulted in customer complaints - meaning that it will have to balance hitting mass production soon enough to be relevant with getting the manufacturing process stable if it wants to avoid a repeat of past mistakes.
So far, TSMC has not released the names of any customers signed up to the first 16nm production runs.
MSI MPG Velox 100R Chassis Review
October 14 2021 | 15:04
Want to comment? Please log in.