Principles of CMOS VLSI design: a systems perspective
Principles of CMOS VLSI design: a systems perspective
Low delay-power product CMOS design using one-hot residue coding
ISLPED '95 Proceedings of the 1995 international symposium on Low power design
Guarded evaluation: pushing power management to logic synthesis/design
ISLPED '95 Proceedings of the 1995 international symposium on Low power design
Communications of the ACM
Low power motion estimation design using adaptive pixel truncation
ISLPED '97 Proceedings of the 1997 international symposium on Low power electronics and design
Efficient Wordlength Reduction Techniques for DSP Applications
Journal of VLSI Signal Processing Systems - Special issue on recent advances in the design and implementation of signal processing systems
Low Power Digital CMOS Design
Reduced Power Dissipation Through Truncated Multiplication
VOLTA '99 Proceedings of the IEEE Alessandro Volta Memorial Workshop on Low-Power Design
Energy reduction by systematic run-time reconfigurable hardware deactivation
Transactions on High-Performance Embedded Architectures and Compilers IV
Low power energy efficient pipelined multiply-accumulate architecture
Proceedings of the International Conference on Advances in Computing, Communications and Informatics
Hi-index | 0.00 |
Reducing switching activity of Arithmetic and Logic Unit (ALU) is important for design of low-power processors. Due to two's complement data notation and fixed bit-width, existing ALUs perform many redundant signal transitions during subtraction, dissipating power. This paper proposes a new scheme that adaptively adjusts the ALU bit-width to input data variation. Unlike related techniques, the scheme masks the number of the Most Significant Bits whose values remain unchanged, thus preserving unnecessary signal variations in corresponding hardware. The scheme is simple in implementation yet efficient in performance. According to simulations, it can reduce the total number of signal transitions per subtraction as much as half and save up to 30% of energy/operation without sacrificing the quality of results.