The transition to 5G comes with huge cost and power implications and requires a shift back from FPGA platforms to ASICs. What are the reasons for the move, and how can it be done cost-effectively without sacrificing all of the flexibility of FPGA’s?
As Zhengmao Li, executive vice president of the world’s biggest operator put it at MWC this year, 5G will require three times as many base stations to deliver the same coverage as LTE, will require three times as much power as LTE, and will cost four times as much as LTE.
5G creates several challenges in terms of power, cost, and range, thus precipitating a shift for the cellular infrastructure sector away from FPGAs/DSPs used in 3G/4G systems and back to ASICs, which are better suited. This doesn’t need to be the preserve of only the richest companies. Yes, the likes of Tesla, Facebook, and Google have all made headlines with multi-billion-dollar ASIC developments. But these are for AI and autonomous-driving equipment, where the most advanced technologies are essential. 5G equipment doesn’t need the same bleeding-edge technologies. As a result, costs can be lowered significantly using an ASIC approach.
In this article Ian Lankshear, CEO of EnSilica, takes a detailed look at the tradeoffs between FPGA, DSP and ASIC designs, how to migrate from FPGAs, the challenges of Incorporating mmWave and concluding with a detailed look at the costs involved in both approaches, highlighting the significant advantages of an ASIC approach.
Read the full article at Electronic Design (English) or the original version in Elektroniktidningen (Swedish)