E.A. Arnould, F.J. Bitz, E.C. Cooper, H.T. Kung, R.D. Samson, and P.A. Steenkiste. The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers. In Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems. ACM, April, 1989.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of packets. The main contribution of the NAB architecture was the idea of smartly partitioning protocol processing between the host CPU and the board, a concept which has seen continued use in most modern NIC designs, including the APIC. The Nectar communications accelarator board (CAB) [1] is a host network interface that connects through a 10 MByte s VME interface to the host system. It too has an on board processor, but the difference is that the CAB processor is responsible for all of the transport protocol processing. By mapping the CAB s on board memory into the address space ....

Arnould, Emmanuel et al; "The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers," ASPLOS-III (ACM SIGOPS Operating Systems Review), Vol. 23, New York, April 1989, pp. 205-216.

....where the rewall is based primarily on IPsec support in the operating systems of the end hosts. The work on TCP wrappers [39] which also runs on a host, uses pattern based access control to network services. Previous work on intelligent I O cards and NICs has focused on performance bene ts [3]. Much of this work has used Myrinet NICs which have a low performance LanAI processor and little memory [8, 16, 40] The SPINE project [15] described o oading multimedia functionality to a processor on a NIC, although the focus is on safe code execution. The Auspex NFS server [18] handles NFS ....

Emmanuel A. Arnould, Francois J. Bitz, Eric C. Cooper, H. T. Kung, Robert D. Sansom, and Peter A. Steenkiste. The design of Nectar: a network backplane for heterogeneous multicomputers. Architectural Support for Programming Languages and Operating Systems (Boston, MA). Published as Operating Systems Review, 23(Special Issue):205-216, April 1989.

....high speed network interfacing for existing workstations is provided by work in the Aurora project [Davie,91] Traw,91] The Aurora interfaces offer considerable performance improvements over conventional interfaces by offioading segmentation and reassembly tasks from the host. The Nectar system [Arnould,89] extends this approach to include sophisticated protocol processors which also aim to offioad the transport protocol handling from the main workstation processor onto the network interface itself. The system bearing closest comparison with the Lancaster MNI unit is Pandora s Box, which was ....

Arnould. E., Bitz, F., Cooper, E., Kung E., Sansom, R., Steenkiste, P., "The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers", Computer Architecture News, Vol 17, No 2, pp205-216, April 1989.

....is to improve the spatial and temporal locality of protocol code. Another approach is to move transport protocol processing off the host and onto the network adapter. This is usually done by adding a dedicated microprocessor to the network adapter. The Nectar CAB (communication accelerator board) [17, 18] is an interface to a 100 Mbit second fiber optic network (called Nectar) that uses an on board 16.5 MHz SPARC processor and connects to a host via the VME bus. Unfortunately the on board processor is not fast enough to deal with small packets. When the packet size is 512 bytes, only 12 ....

E. Arnould, F. Bitz, E. Cooper, H.T. Kung, R. Sansom and P. Steenkiste, The design of Nectar: a network backplane for heterogeneous multicomputers, Third International Conference on Architectural Support for Programming Languages and Operating Systems, April 1989.

....Parallel processing in a heterogeneous network environment has become an attractive option for delivering high performance on a range of applications. Interest in distributed parallel processing has been based on advances in three technology areas, local and wide area high performance networking [4][6] 19] 43] 81] toolkits that enable networkbased parallel processing and job multiprogramming [11] 52] 73] 83] and parallel compilation techniques for distributed memory MIMD computers [12] 33] 41] 57] 76] In this thesis we consider a distributed computing environment known as a metasystem. A ....

E.A. Arnould et al, "The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers," 3rd International Conference on Architectural Support for Programming Languages and Operating Systems, 1989.

.... The incorporation of optical fiber link technology into existing multi hop network designs solves many problem that are associated with electronic interconnections, including driving power, capacitive inductive loading, and relatively low transmitter rates implied by sensitivity to noise [22, 1, 14]. This simple technological replacement provides at least an order of magnitude increase in link throughput. Further, the lack of reactive factors and a high noise immunity is well suited to the demands of faster, larger, and more distributed environments. Such networks, however, still suffer ....

E. Arnould, F. Bitz, E. Cooper, R. Sansom, and P. Steenkiste. The design of nectar: A network backplane for heterogeneous multicomputers. In Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, pages 205--216. IEEE, 1989.

....Diverse network characteristics There is a considerable amount of excitement today about the rapid strides being made in the field of networking. Nationwide networks at 45 Mb s are a reality, and work is under way to provide gigabit connectivity. Projects such as Autonet [8] at DEC SRC and Nectar [1] at CMU are pushing the current limits of performance in local area networking. Continuing progress in the areas of optic fiber materials and interface technology indicates that this trend will continue for a long time to come. There is considerable debate in the distributed systems community ....

Arnould, E.A., Bitz, F.J., Cooper, E.C., Kung, H.T., Sansom, R.D., Steenkiste, P.A. The Design of Nectar: a Network Backplane for Heterogeneous Multicomputers. In Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, Boston, MA. 1989.

....PMMU 68851 Processor MC 68020 FPP 68881 4 MB Local Memory VMX Bus Interface VME Bus Interface Processor Card Figure 1: Block diagram of a HARTS node. 5 device interface to the application processors which are responsible for executing higher level protocols. The Nectar system [5] uses a communications adapter board (CAB) which features a general purpose SPARC processor. The CAB software organization [11] is similar to our design and provides a high level interface to the host processors. However, these systems do not deal with the problem of real time communication. This ....

E. A. Arnould, F. J. Bitz, E. C. Cooper, H. T. Kung, R. D. Sansom, and P. A. Steenkiste, "The design of Nectar: A network backplane for heterogeneous multicomputers," in Proceedings of the Third Intl. Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS-III), pp. 205--216. ACM, April 1989.

....(MEND) reported in this paper aims to give multimedia networked functionality at a reduced cost. Obviously to maintain the low cost a reduction in quality and functionality is necessary. The MEND is a self contained device that can simply be plugged into existing workstations. Other researchers [Arnould, 89] and [Chesson, 88] also see the advantage of giving the network processing overheads to an autonomous unit. The MEND devices will sit on or under a desk by a workstation, PC, or Mac. They are connected to each other by a dedicated network. MEND Fig 1. Typical MEND Set Up 2. Functionality Each ....

Arnould, E., Bitz, F., Cooper, E., Kung, H., Sansom, R. and P. Steenkiste, "The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers", Computer Architecture News, Vol. 17, No. 2, pp 205-216, April 1989.

....[90] More importantly, it permits management of the communication subsystem by a separate executive designed for communication related operations, such as the x kernel [76] The VMP NAB [90] and Nectar CAB [120] are two examples of this functional partitioning. For example, the Nectar CAB [10, 120] off loads all protocol processing functions from the host processor, freeing it from adapter handshake overheads and permitting greater overlap between useful computation and communication processing. 31 Multiprocessor front ends: Multiprocessor front ends may be designed using specialpurpose ....

E. A. Arnould, F. J. Bitz, E. C. Cooper, H. T. Kung, R. D. Sansom, and P. A. Steenkiste, "The design of Nectar: A network backplane for heterogeneous multicomputers, " in Proc. Int'l Conf. on Architectural Support for Programming Languages and Operating Systems, pp. 205--216, April 1989.

.... databases, parallel languages like Ada or Multilisp, and systems for parallel and distributed programming [11, 2, 10] More motivation for a distributed shared memory facility comes from the increasing interest that hardware designers show in non shared memory multiprocessors: the Nectar project [1] at CMU for instance uses fast fiber optic links. This will reduce the end to end time to send a 1 kilobyte data packet from the tens of milliseconds range of the current ethernet to the tens of microseconds range of the fiber. Fast communication makes distributed shared memory an appealing ....

Arnould, E. A., Bitz, F. J., Cooper, E. C., Kung, H. T., Sansom, R. D., Steenkiste, P. A. The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers. April, 1989. To appear in the Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS-III).

....developed in completely new environments, both in terms of hardware platform and the operating system. In contrast, DeViouS is designed to run on commercially available hardware using widely 9 used software. Paradigm : Paradigm was built on top of the Nectar system at Carnegie Mellon University[22, 8]. Nectar is a fiber optic based backplane of heterogeneous computers. At this time only a prototype exists. Paradigm was designed to take advantage of the low latency and high bandwidth communication channels that a fiber optic based system provides to study methods for task partitioning and task ....

E. A. Arnould et al. The design of Nectar: A network backplane of heterogeneous mu lticomputers. Proceedings from the Third International Conference on Archit ectural Support for Programming Languages and Opterating Systems, 17(22):205--216, April 1989.

....system configuration. Moreover, it is not clear how gang scheduling performs with respect to the communication layer if SMP nodes are used. A lot of work has been done in providing reliable communication in traditional, kernel based, communication systems and especially TCP IP. The Nectar system [1] used a general purpose processor in the network interface to run the TCP IP protocol stack. However, this is the only work we are aware of, that argues for and incorporates reliable communication in the network interface, in a user level communication system for SAN networks. AM [63] support ....

....amount of data sent per processor for each application for 1 processor per node, for a total of 16 processors. f N = P 0#i#N (C i S i ) P 0#i#N E i , where E i , C i , S i are the elapsed, compute and cache stall time of the i th processor in the parallel execution. Obviously f N # [0, 1]. Note that the costs in the sequential execution are not directly involved in this definition. Assuming that a protocol is responsible for all overheads except the local cache stall time, this definition penalizes a protocol for every cost incurred. Thus, a protocol that injects overheads, ....

E. Arnould, F. Bitz, E. Cooper, H. Kung, R. Sansom, and P. Steenkiste. The design of nectar: A network backplane for heterogeneous multicomptuers. In The 3rd International Conference on Architectural Support for Programming Languages and Operating Systems, pages 205--216, Apr. 1989.

....a 64K Connection Machine. The initial optical implementation was reported to run at 400 Mbit sec. The next step is for the optical link to run at over 1 Gbit sec, but those results have not been reported yet. The Nectar system is a fiber optic network backplane to connect heterogeneous computers [4, 50, 51, 52, 64]. It is designed to exploit coarse level (task level) parallelism by connecting different types of machines together. A design goal was that a user process on one node be able to send a message to another user process on another node in 100 sec (excluding the transmission delay on the fibers) ....

Emmanuel A. Arnould, Francois J. Bitz, Eric C. Cooper, H. T. Kung, Robert D. Sansom, and Peter A. Steenkiste. The design of Nectar: A network backplane for heterogeneous multicomputers. In Proceedings of Third International Conference on Architectural Support for Programming Languages and Operating Systems, pages 205--16, April 1989.

....twice that presented in Table 1. Interface Performance Bytes 4 54 1500 Msg Sec Bandwidth 657,000 396,000 31,750 21 Mb s 171 Mb s 381 Mb s Table 1: Single Source Table 2 places the performance of the prototype ATOMIC network into perspective with other recently developed LAN technologies: Nectar [9] and Autonet [10] Network Switch (# ports) Max Pkts Sec Latency Channel Bandwidth Aggregate Switch Bandwidth Autonet (12 12) Nectar (16 16) 2 M 14 M 2000 ns 700 ns 100 Mb s 100 Mb s 1.2 Gb s 1.6 Gb s ATOMIC (6 6) 31 M 125 ns 500 Mb s 3 Gb s ATOMIC (16 16) 80 M 375 ns 500 Mb s 8 Gb s Table 2: ....

Arnould, E., Bitz, F., Cooper, E., Kung, H. T., Sansom, R., Steenkiste, P. The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, ACM 1989, pp. 205-216.

....many applications to use these primitives. 1.1. Related work The idea of running parallel programs on workstation clusters has existed for many years. A few of the early examples are Spector s work on the Alto [20] the Multisatellite star work in the V system [11] and the Nectar project at CMU [3]. There have also been several software systems that support a shared memory or a shared object abstraction on workstation clusters for executing parallel programs. Some examples are IVY [16] Amber [10] Munin [9] Orca [4] and Treadmarks [15] Current research in using workstation clusters to ....

E. A. Arnould, F. J. Bitz, E. C. Cooper, H. Kung, R. D. Sansom, and P. A. Steenkiste. The design of Nectar: A network backplane for heterogeneous multicomputers. In Proceedings of the 3rd International Conference on Architectural Support for Programming Languages and Operating Systems, pages 205--216, April 1989.

....SHRIMP, Memory Channel) e.g. TPVM, Chant) e.g. MPI Nexus) e.g. NCS) e.g. U Net, AM, FM) e.g. MPI FM, PVM ATM) e.g. Fast Sockets) Figure 1. Classification of High Performance Message Passing Schemes Hardware based Approach In hardware based approach, the research efforts such as [5], 6] 7] 8] 9] 11] 12] have focused on building special hardware to reduce communication latency and achieve high throughput. The developers of the communication hardware should implement the device drivers and proprietary APIs for their communication hardware. By porting well known ....

E. Arnould, F. Bitz, E. Cooper, H. T. Kung, R. Sansom, and P. Steenkiste, "The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers", Proc. of the 3rd International Conference on Architectural Support for Programming Languages and Operating Systems, pp. 205-216, April 1989. 21

....traffic due to several key physical link issues. These include driving power, capacitive inductive loading, and relatively low transmitter rates implied by sensitivity to noise. The second era in computer networking incorporated optical fiber link technology into existing multihop network designs [11, 1, 6]. An order of magnitude increase in link throughput was available. Further, the lack of reactive factors and a high noise immunity were well suited to the demands of the faster, larger, and more widespread environments. Such networks, however, still suffer from throughput bottlenecks and high ....

E. Arnould, F. Bitz, E. Cooper, R. Sansom, and P. Steenkiste. The design of nectar: A network backplane for heterogeneous multicomputers. In Proc. of 3rd Int. Conf. on Arch. Support for Prog. Lang. and Oper. Sys., pages 205--216. IEEE, 1989.

....as a basis for realistic loosely coupled multicomputing. More recently, multicomputer researchers have incorporated some of these ideas into reliable networks for distributed memory multicomputers [25] The Nectar project at CMU connected workstations using a high performance custom network [3], while DEC s VAXclusters system [17] used complex communication controllers to provide distribution services for a cluster. At a coarse granularity, systems like Ivy [19] Munin [8] and Memnet [11] provide coherent remote page access using high level software or hardware. We expect that our ....

Emmanuel A. Arnould, Francois J. Bitz, Eric C. Cooper, H.T. Kung, Robert D. Sansom, and Peter A. Steenkiste. The design of Nectar: A network backplane for heterogeneous multicomputers. In Proceedings of the 3rd International Conference on Architectural Support for ProgrammingLanguages and Operating Systems, April 1989.

No context found.

Emmanuel Arnould, Francois Bitz, Eric Cooper, H. T. Kung, Robert Sansom and Peter Steenkiste. The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers. Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, ACM/IEEE, Boston, April, 1989, pp. 205-216.

No context found.

Emmanuel Amould, Francois Bitz, Eric Cooper, H. T. Kung, Robert Sansom and Peter Steenkiste. The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers. Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, ACM/IEEE, Boston, April, 1989, pp. 205-216.

....remote memory accesses more efficiently, i.e. reduce host overhead, by using a more powerful network adapter. A first approach is to rely on a powerful outboard processor to take over some of the communication tasks normally performed by the host, e.g. similar to the original Nectar systems [2]. An alternative is to provide dedicated hardware support, e.g. similar to the Shrimp interface [6] While both approaches reduce communication overhead significantly, they have the disadvantage that they may lead to expensive adapters, and more importantly, since the adapter becomes more special ....

Emmanuel Arnould, Francois Bitz, Eric Cooper, H. T. Kung, Robert Sansom, and Peter Steenkiste. The design of nectar: A network backplane for heterogeneous multicomputers. In Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, pages 205--216, Boston, April 1989. ACM/IEEE.

No context found.

E.A. Arnould, F.J. Bitz, E.C. Cooper, H.T. Kung, R.D. Samson, and P.A. Steenkiste. The Design of Nectar: A Network Backplane for Heterogeneous Multicomputers. In Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems. ACM, April, 1989.

No context found.

E. A. Arnould, F. J. Bitz, E. C. Cooper, H. T. Kung, R. D. Sansom, and P. A. Steenkiste. The design of Nectar: A network backplane for heterogeneous multicomputers. In Proc. 3rdACM Intl. ConferenceonArchitectural Support for Programming Languages and Operating Systems, Boston, Massachusetts, April 1989.

No context found.

Emmanuel A. Arnould, Francois J. Bitz, Eric C. Cooper, H. T. Kung, Robert D. Sansom and Peter A. Steenkiste, `The design of Nectar: a network backplane of heterogeneous multicomputers', Proc. Third International Conference on Architectural Support for Programming Languages and Operating Systems, 17, (22), 205--216 (1989).

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC