Toward Ideal On-Chip Communication Using Express Virtual Channels

Princeton Univ., Princeton
IEEE Micro (Impact Factor: 1.81). 02/2008; 28(1):80 - 90. DOI: 10.1109/MM.2008.18
Source: DBLP

ABSTRACT Current on-chip networks use a packet-switched design with a complex router at every hop, which imposes significant communication energy, delay, and throughput overhead. we propose reducing energy and delay, and increasing throughput, using express virtual channels. packets traveling along these virtual express lanes, which connect distant nodes in the network, bypass intermediate routers, significantly reducing router overhead.

  • Source
    • "This kind of network topology, commonly referred to as small-world, can be incorporated in NoCs by introducing long-range, high bandwidth and low power links between distant cores [1]. There have been efforts to improve NoC performance by introducing low-latency long-range links and low power express channels between highly separated nodes, where the performance gain is achieved by bypassing intermediate NoC switches/routers [1][2]. These communication channels are more efficient in terms of power and delay compared to their conventional counterparts, but they are still, basically, metal wires. "
    [Show abstract] [Hide abstract]
    ABSTRACT: While traditional cluster computers are more constrained by power and cooling costs for solving extreme-scale (or exascale) problems, the continuing progress and integration levels in silicon technologies make possible complete end-user systems on a single chip. This massive level of integration makes modern multicore chips all pervasive in domains ranging from climate forecasting and astronomical data analysis, to consumer electronics, smart phones, and biological applications. Consequently, designing multicore chips for exascale computing while using the embedded systems design principles looks like a promising alternative to traditional cluster-based solutions. This paper aims to present an overview of new, far-reaching design methodologies and run-time optimization techniques that can help breaking the energy efficiency wall in massively integrated single-chip computing platforms.
    Design, Automation and Test in Europe, DATE 2011, Grenoble, France, March 14-18, 2011; 03/2011
  • Source
    • "The EVC flow control mechanism proposed in [17] enables packets to entirely bypass routers. Figure 3 shows the router architecture supporting EVCs. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Manycore systems require energy-efficient on-chip networks that provide high throughput and low latency. The performance of these on-chip networks affects cache access latency and, consequently, system performance. This paper proposes solutions to address the performance limitations related to the use of snoop-based cache coherence protocol on switched network-on-chip (NoC). We propose a new network flow control technique, Express Virtual Channel with Taps (EVC-T), for transmitting both broadcast packets and data packets efficiently. In addition, we propose a low-latency broadcast packet notification tree network that maintains the order of broadcast packets on an unordered NoC. We evaluate our technique using both synthetic traffic and parallel benchmark suites through detailed system simulation. EVC-T reduces the average network latency by 24% with a negligible change in power for synthetic benchmarks. For NAS parallel applications, EVC-T increases the instructions per cycle (IPC) by 9% on average with minimal increase in power. Our technique reduces the energy-delay product (EDP) by 13% on average across all benchmarks.
    IEEE 19th Annual Symposium on High Performance Interconnects, HOTI 2011, Santa Clara, CA, USA, August 24-26, 2011; 01/2011
  • Source
    • "Besides, complex high-radix routers are necessary again. Therefore, Kumar et al. [1] propose the use of Express Virtual Channels (EVC) to not increase the routers' radix. Nevertheless, EVC necessitates costly virtual channels [8] in any case that also include additional packet buffers and arbitration logic. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Several alternatives of mesh-type topologies have been published for the use in Networks-on-Chip. Due to their regularity, mesh-type topologies often serve as a foundation to investigate new ideas or to customize the topology to application-specific needs. This paper analyzes existing mesh- type topologies and compares their characteristics in terms of communication and implementation costs. Furthermore, this paper proposes BEAM (Border-Enhanced Mesh) − a mesh- type topology for Networks-on-Chip. BEAM uses concentration while necessitating only low-radix routers. Thereto, additional resources are connected to the outer boundaries of a conventional mesh. As a result, overall bandwidth is traded off against hardware overhead. In conclusion, simulation and synthesis results show that the conventional mesh stands out due to its communication performance, whereas clustered and concentrated topologies offer the least hardware overhead. BEAM ranges in between and is an option to balance hardware costs and communication performance.
    13th Euromicro Conference on Digital System Design, Architectures, Methods and Tools, DSD 2010, 1-3 September 2010, Lille, France; 10/2010
Show more


1 Download
Available from