14 DAY TRIAL // Taiwan Semiconductor Manufacturing Co. confirmed that is gearing up to mass produce 10nm process technology within a year from now. If volume production is to start somewhere the next year, it means that its clients will probably be receiving them at the start of 2017.
Apparently, TSMC believes that it advances steadily to the 10nm goal without any major obstacles on its way. Costumer products qualifications and technology risk qualifications have already started this year, and while this process is happening right now, enough 10nm wafers will be manufactured to satisfy consumer needs in two years from now.
As we mentioned in a previous article, it will take about 100 days from when a FinFET wafer to reach delivery time, so the dates expressed represent a rather realistic time span. However, this announcement is no small thing as other companies like Intel still have major difficulties in developing the 10nm FinFETs in time forcing them to develop longer product cycles for a certain family of CPUs that breaks even more Intel’s cadence of CPU performance.
Pushing 10nm process ahead of others at all costs
This is not stopping, however, other companies like TSMC and IBM that make important breakthroughs in process technology development as they push forwards beyond the 14nm limits promising even more powerful chips in the near future. These breakthroughs are extremely important as the first to reach these process technologies will offer more stable and mature products to their clients, unlike those who mass produce 10nm processes too late.
TSMC managed to disclose the different characteristics of its 10nm FinFETs on different occasions. The semiconductor manufacturer believes that its 10nm FinFET (CLN10FF) fabrication process will have 110 – 120 per cent higher transistor density compared to its 16nm FinFET+ (CLN16FF+), meaning 15 per cent higher frequency at the same power and 35 per cent lower power consumption at the same frequency and complexity.
However, KitGuru believes that the performance improvements of its 10nm process fabrication compared to its 16nm FinFET Plus aren’t that different as both come with similar improvements of clock-rates and power consumption over the classical 16nm CLN16FF. This way, it seems that TSMC cuts corners far more than Intel to reach the 10nm process benchmark with little to no performance differences than its existing 16nm process.