|SSIA Summit 2016 - The dichotomy of software flexibility and hardware efficiency in future semiconductor platforms|
Mr Bolsen started by sharing some observations on the semiconductor growth. Despite a slower overall growth in the industry, the semiconductor R&D spending continued to grow with an 8.3% CAGR from 2010-2015, there have been fewer ASIC design starts and the industry has continued on its consolidation path. An interesting chart was on the transistor cost trend with scaling - bucking of the trend is noted at 20nm and 14nm. The non-scalability of traditional compute architectures has also ushered in heterogenous computing platforms to fill the void.
The above set the tone for the advent of Programmable dichotomy, the main theme of Mr Bolsen's presentation. The dichotomy includes being defined by software and optimized by hardware. Software brings the advantages of flexibiity, intelligence and scalability while hardware includes Big data and video, multi-level security, real time sensor fusion and others.
The talk then zoned into 4 major high growth markets – Cloud Computing, Embedded vision, Industrial Iot and 5G Wireless. Some megatrends including the driving forces in each of these markets was shared. Some of the interesting spaces for Xilinx in these markets included the usage of FPGAs in the streaming architectures required for machine learning, as smart hardware accelerator in Virtual Network Functions, programmable software defined storage, heterogenous paltform in automotive systems and others.
It was an interesting share of the increasing use of software and programmability in the semiconductor space.