The technology scaling impact on FinFET-based Field-Programmable Gate Array (FPGA) components (Flip-Flops and Multiplexers) and cluster metrics is evaluated for technology nodes starting from 20[Formula: see text]nm down to 7[Formula: see text]nm. Power consumption, delay and energy (Power Delay Product, or PDP) trends are reported with FinFET technology scaling. Cluster metrics are then evaluated based on three benchmarking circuits: 2-bit adder, 4-bit NAND and cascaded flip-flops chain. The study shows that power, delay and PDP of the FPGA cluster are improved as we scale down the technology. An example for improvement is that for 7[Formula: see text]nm 2-bit adder, circuit speed is 15% higher than its value at 20[Formula: see text]nm and PDP at 7[Formula: see text]nm is reduced by 43% compared to its value at 20[Formula: see text]nm. The impacts of temperature and threshold voltage variations on FPGA cluster performance are also reported after evaluating a 2-bit adder circuit as a benchmark which is then used to calculate the design constraints to meet 99.9% yield percentage.

No abstract

Analysis of FinFET based transmission gate (TG), sense amplifier (SA), and semi dynamic (SD) Flip-flops metrics are evaluated with technology scaling from 20nm down to 7nm technology node. The impact of supply voltage variation on delay, power, and energy is reported. The power delay product of Flip-flops is enhanced with technology scaling. The optimum supply voltage value at each technology node from minimum energy perspective is evaluated which is used by the industry to optimize logic and memory circuitry designs. For instance, considering the 7nm TG Flip-flop, the optimum supply voltage from energy saving point of view occurs at 0.65V. The work also characterizes each Flip-flop according to the obtained simulation results. SD Flip-flop has the best performance, however it exhibits high power consumption. TG Flip-flop is the best choice from power dissipation perspective, but it has high clock load. SA Flip-flop has a very useful feature of monotonous transitions at the outputs, which drives fast domino logic, however it might have glitches and it is the most vulnerable to soft errors.