Browsing by Author "Pang, J"

Now showing 1 - 3 of 3
  • No Thumbnail Available
    ItemOpen Access
    Modeling and simulation of a nanoscale optical computing system
    (Journal of Parallel and Distributed Computing, 2014-01-01) Pang, J; Lebeck, AR; Dwyer, C
    Optical nanoscale computing is one promising alternative to the CMOS process. In this paper we explore the application of Resonance Energy Transfer (RET) logic to common digital circuits. We propose an Optical Logic Element (OLE) as a basic unit from which larger systems can be built. An OLE is a layered structure that works similar to a lookup table but instead uses wavelength division multiplexing for its inputs and output. Waveguides provide a convenient mechanism to connect multiple OLEs into large circuits. We build a SPICE model from first principles for each component to estimate the timing and power behavior of the OLE system. We analyze various logic circuits and the simulation results show that the components are theoretically correct and that the models faithfully reproduce the fundamental phenomena; the power-delay product of OLE systems is at least 2.5× less than the 14 nm CMOS technology with 100× better density. © 2013 Elsevier Inc. All rights reserved.
  • No Thumbnail Available
    ItemOpen Access
    More is Less, Less is More: Molecular-Scale Photonic NoC Power Topologies
    (ACM SIGPLAN NOTICES, 2015-04) Pang, J; Dwyer, C; Lebeck, AR
  • No Thumbnail Available
    ItemOpen Access
    Rhythm: Harnessing data parallel hardware for server workloads
    (International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS, 2014-03-14) Agrawal, SR; Pistol, V; Pang, J; Tran, J; Tarjan, D; Lebeck, AR
    Trends in increasing web traffic demand an increase in server throughput while preserving energy efficiency and total cost of ownership. Present work in optimizing data center efficiency primarily focuses on the data center as a whole, using off-the-shelf hardware for individual servers. Server capacity is typically increased by adding more machines, which is cheap, though inefficient in the long run in terms of energy and area. Our work builds on the observation that server workload execution patterns are not completely unique across multiple requests. We present a framework - called Rhythm - for high throughput servers that can exploit similarity across requests to improve server performance and power/energy efficiency by launching data parallel executions for request cohorts. An implementation of the SPECWeb Banking workload using Rhythm on NVIDIA GPUs provides a basis for evaluating both software and hardware for future cohort-based servers. Our evaluation of Rhythm on future server platforms shows that it achieves 4× the throughput (reqs/sec) of a core i7 at efficiencies (reqs/Joule) comparable to a dual core ARM Cortex A9. A Rhythm implementation that generates transposed responses achieves 8× the i7 throughput while processing 2.5× more requests/Joule compared to the A9. Copyright © 2014 ACM.