Effects of Using Different Resources on Reliability and Power Consumption in Digital Circuit Design
Özet
Designing an application in hardware under inversely competing constraints such as area, performance and power consumption with different objective functions such as reliability of the circuits is a cumbersome task. Having different versions of the same resource type handy during the design process may ease this burden since there can be several alternative resources to meet the given constraints. Although area, power consumption and speed values of commonly used arithmetic circuits can be found in previous researches, reliability analysis of these circuits that are implemented on FPGAs are not done. Thanks to reconfigurability features of FPGAs and their low cost compare to ASIC design at little production numbers make FPGAs desirable also in aerospace industry. Especially in space applications soft errors become a serious issue and requires reliability centric designs. For this reason reliability concern in FPGAs is a salient subject. \\ \\
In this study, some commonly used arithmetic circuits in FPGAs in terms of speed, area, power consumption, and vulnerability to error propagation as reliability parameter are characterized to create a library. Specifically, four well-known adders and two multipliers in SRAM-based FPGA part of Xilinx's Zynq-7000 SoC platform are implemented. Then errors are injected to the configuration bits of the circuits to evaluate the error propagation. The number and the ratio of the bits which causes miscalculations are determined. The results show that different versions of the same resources can have different reliability values in addition to area, latency, and power values. Finally separately analyzed circuits are mixed in a topology consists of six adders. In this circuit impact of different resource selection on meeting constraints is presented.
