Microarchitectural Wire Management for Performance and Power in Partitioned Architectures
Authors:
- 1 April 2005
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15300897,p. 28-39
Abstract
Future high-performance billion-transistor processors are likely to employ partitioned architectures to achieve high clock speeds, high parallelism, low design complexity, and low power. In such architectures, inter-partition communication over global wires has a significant impact on overall processor performance and power consumption. VLSI techniques allow a variety of wire implementations, but these wire properties have previously never been exposed to the microarchitecture. This paper advocates global wire management at the microarchitecture level and proposes a heterogeneous interconnect that is comprised of wires with varying latency, bandwidth, and energy characteristics. We propose and evaluate microarchitectural techniques that can exploit such a heterogeneous interconnect to improve performance and reduce energy consumption. These techniques include a novel cache pipeline design, the identification of narrow bit-width operands, the classification of non-critical data, and the detection of interconnect load imbalance. For a dynamically scheduled partitioned architecture, our results demonstrate that the proposed innovations result in up to 11% reductions in overall processor ED/sup 2/, compared to a baseline processor that employs a homogeneous interconnect.Keywords
This publication has 40 references indexed in Scilit:
- A design space evaluation of grid processor architectures2005
- Clustered multithreaded architectures -pursuing both IPC and cycle time2004
- A Global Interconnect Optimization Scheme for Nanometer Scale VLSI With Implications for Latency, Bandwidth, and Power DissipationIEEE Transactions on Electron Devices, 2004
- Exploiting Low Entropy to Reduce Wire DelayIEEE Computer Architecture Letters, 2004
- Near speed-of-light signaling over on-chip electrical interconnectsIEEE Journal of Solid-State Circuits, 2003
- Modulo scheduling for a fully-distributed clustered VLIW architecture2000
- Dynamic cluster assignment mechanisms2000
- Dynamically exploiting narrow width operands to improve processor power and performance1999
- Digital Systems Engineering1998
- The case for a single-chip multiprocessor1996