RazorProtector: Maintaining Razor DVS Efficiency in Large IR-Drop Zones by an Adaptive Redundant Data-Path
スポンサーリンク
概要
- 論文の詳細を見る
Recently, the DVS (Dynamic Voltage Scaling) method has been aggressively applied to processors with Razor Flip-Flops. With Razor FF detecting setup errors, the supply voltage in these processors is down-scaled to a near critical setup timing level for a maximum power consumption reduction. However, the conventional Razor and DVS combinations cannot tolerate well error rate variations caused by IR-drops and environment changes. At the near critical setup timing point, even a small error rate change will result in sharp performance degradation. In this paper, we propose RazorProtector, a DVS application method based on a redundant data-path which uses a multi-cycle redundant calculation to shorten the recovery penalty after a setup error occurrence. A dynamic redundancy-adapting scheme is also given to use effectively the designed redundant data-path based on a study of the program, device and error rate characteristics. Our results show that RazorProtector with the adaptive redundancy architecture can, compared to the traditional DVS method with Razor FF, under a large setup rate caused by a 10% unwanted voltage drop, reduce EDP up to 78% at 100µs/V, 88% at 200µs/V voltage scaling slope.
著者
-
Yao Jun
Graduate School Of Informatics Kyoto Univ.
-
NAKADA Takashi
Graduate School of Information Science, Nara Institute of Science and Technology
-
Nakashima Yasuhiko
Graduate School Of Information Science Naist
-
SASAGAWA Yukihiro
Graduate School of Information Science, Nara Institute of Science and Technology
関連論文
- An Instruction Mapping Scheme for FU Array Accelerator
- D-10-5 An Instruction Decomposition Scheme to Aid Fine-Grained Online Recovery in Pipeline Processor
- An EDP Study on the Optimal Pipeline Depth for Pipeline Stage Unification Adoption
- A Light Bypass Network Design for Cascading ALU Executions
- A Dynamic Control Mechanism for Pipeline Stage Unification by Identifying Program Phases
- Dynamic Control Mechanisms for Pipeline Stage Unification Based on Program Phase Detection
- Dynamic Control Mechanisms for Pipeline Stage Unification Based on Program Phase Detection(集積回路技術とアーキテクチャ技術の協調・融合へ向けた,プロセッサ,並列処理,システムLSIアーキテクチャ及び一般)
- ReVolver/C40 : A Scalable Parallel Computer for Volume Rendering-Design and Implementation-(Development of Advanced Computer Systems)
- Parallel Cloth Simulation with Adaptive Mesh Refinement and Coarsening Using OpenMP on Fujitsu HPC2500(Parallel Processing Applications)
- An approach towards fast simulation of virtual cloth with adaptive mesh refinement and coarsening on Fujitsu HPC2500
- Priority Enhanced Stride Scheduling
- Quantum Walks on the Line with Phase Parameters
- RazorProtector: Maintaining Razor DVS Efficiency in Large IR-Drop Zones by an Adaptive Redundant Data-Path
- Selective Check of Data-Path for Effective Fault Tolerance
- Tensor Rank and Strong Quantum Nondeterminism in Multiparty Communication
- An Instruction Scheduler for Dynamic ALU Cascading Adoption
- An Instruction Scheduler for Dynamic ALU Cascading Adoption