Overview of the power minimization techniques employed in the IBM PowerPC 4xx embedded controllers
ISLPED '95 Proceedings of the 1995 international symposium on Low power design
Power considerations in the design of the Alpha 21264 microprocessor
DAC '98 Proceedings of the 35th annual Design Automation Conference
Reducing power in high-performance microprocessors
DAC '98 Proceedings of the 35th annual Design Automation Conference
Low-Power CMOS Design
A case for dynamic pipeline scaling
CASES '02 Proceedings of the 2002 international conference on Compilers, architecture, and synthesis for embedded systems
Pipeline stage unification: a low-energy consumption technique for future mobile processors
Proceedings of the 2003 international symposium on Low power electronics and design
Stall cycle redistribution in a transparent fetch pipeline
Proceedings of the 2006 international symposium on Low power electronics and design
Variable latency caches for nanoscale processor
Proceedings of the 2007 ACM/IEEE conference on Supercomputing
A mathematical solution to power optimal pipeline design by utilizing soft edge flip-flops
Proceedings of the 13th international symposium on Low power electronics and design
Instruction-driven clock scheduling with glitch mitigation
Proceedings of the 13th international symposium on Low power electronics and design
Minimizing the energy cost of throughput in a linear pipeline by opportunistic time borrowing
Proceedings of the 2008 IEEE/ACM International Conference on Computer-Aided Design
A novel sequential circuit optimization with clock gating logic
Proceedings of the 2008 IEEE/ACM International Conference on Computer-Aided Design
Reducing branch misprediction penalties via adaptive pipeline scaling
HiPEAC'07 Proceedings of the 2nd international conference on High performance embedded architectures and compilers
ICCD'09 Proceedings of the 2009 IEEE international conference on Computer design
Improved clock-gating control scheme for transparent pipeline
Proceedings of the 2010 Asia and South Pacific Design Automation Conference
CRIB: consolidated rename, issue, and bypass
Proceedings of the 38th annual international symposium on Computer architecture
Hi-index | 0.00 |
This paper re-examines the well established clocking principles of pipelines. It is observed that clock gating techniques that have long been assumed optimal in reality produce a significant amount of redundant clock pulses. The paper presents a new theory for optimal clocking of synchronous pipelines, presents practical implementations and evaluates the clock power benefits on a multiply/add-accumulate unit design. Transistor level simulations show that dynamic clock power dissipation can be reduced by 40-60% at pipeline utilization factors between 20-60%, on top of traditional stage-level clock gating, without affecting pipeline latency or throughput.