Value Compression for Efficient Computation

A processor's energy consumption can be reduced by compressing values (data and addresses) that flow through a processor pipeline and gating off portions of data path elements that would otherwise be used for computing non-significant bits. An approach for compressing all values running through a processor is proposed and evaluated. For the SpecInt2000 benchmarks the best compression method achieves energy savings of more than 20 percent and a peak power reduction of 18 percent.

[1]  M. Horowitz,et al.  Energy dissipation in general purpose processors , 1995, 1995 IEEE Symposium on Low Power Electronics. Digest of Technical Papers.

[2]  Yun Cao,et al.  Reducing access energy of on-chip data memory considering active data bitwidth , 2002, ISLPED '02.

[3]  Krste Asanovic,et al.  Dynamic zero compression for cache energy reduction , 2000, MICRO 33.

[4]  Toshinori Sato,et al.  Table size reduction for data value predictors by exploiting narrow width values , 2000, ICS '00.

[5]  Gabriel H. Loh.  Exploiting data-width locality to increase superscalar execution bandwidth , 2002, MICRO 35.

[6]  Mark Horowitz,et al.  Energy dissipation in general purpose microprocessors , 1996, IEEE J. Solid State Circuits.

[7]  James E. Smith,et al.  Software-controlled operand-gating , 2004, International Symposium on Code Generation and Optimization, 2004. CGO 2004..

[8]  Jun Yang,et al.  Energy efficient Frequent Value data Cache design , 2002, 35th Annual IEEE/ACM International Symposium on Microarchitecture, 2002. (MICRO-35). Proceedings..

[9]  Margaret Martonosi,et al.  Dynamically exploiting narrow width operands to improve processor power and performance , 1999, Proceedings Fifth International Symposium on High-Performance Computer Architecture.

[10]  Kiyoung Choi,et al.  Power minimization of functional units partially guarded computation , 2000, ISLPED '00.

[11]  Michael E. Wazlowski,et al.  IBM Memory Expansion Technology (MXT) , 2001, IBM J. Res. Dev..

[12]  Mary Lou Soffa,et al.  Width-Sensitive Scheduling for Resource-Constrained VLIW Processors , 2000 .

[13]  Margaret Martonosi,et al.  Wattch: a framework for architectural-level power analysis and optimizations , 2000, Proceedings of 27th International Symposium on Computer Architecture (IEEE Cat. No.RS00201).

[14]  James E. Smith,et al.  Very low power pipelines using significance compression , 2000, MICRO 33.