16 citations found. Retrieving documents...
Alan M. Mainwaring and David E. Culler. Active Messages: Organization and applications programming interface. Technical report, Computer Science Division, University of California at Berkeley, 1995. Available from http:// now.cs.berkeley.edu/Papers/Papers/am-spec.ps.

 Home/Search   Document Not in Database   Summary   Related Articles   Check  

This paper is cited in the following contexts:
A Software Architecture for Global Address Space.. - Giannini, al. (1998)   (Correct)

....for all message sizes using a single, simple interface. FM provides direct user level access to the network and supports multiple processes per node. There are a number of other cluster based messaging systems which have design goals and performance similar to FM. These include Active Messages [11], PM [18] U Net [19] and VMMC 2 [7] Operation Comment shmemJnitialize0 initialize shmem finalize0 terminate shmem get0 remote read (blocking) shmem cget0 remote read (non blocking) shmem put0 remote write shmem wait0 spin on memory location shmem cwait 0 spin on memory location ....

....They both support only the primary data movement routines in hardware. Furthermore, because they are hardware based they operate only on physically addressed and pinned buffers. The Active Messages project has chosen a different approach, integrating the put get primitives deeply into the AMII [ 11] system. The earlier versions of Active Messages did not provide put get capability. Deeper integration enables the put get transfers to move data outside of the messaging buffer pool, finessing deadlock and flow control issues. However, the operations are limited to a single memory segment per ....

A. Mainwadng and D. Culler. Active messages: Organization and applications programming interface, 1995.


High Performance Video Server using Myrinet - Alice Bonhomme Patrick   (Correct)

....it allows the server to be scalable and thus to increase the number of supported streams as we adds nodes in the cluster. Then, we had to choose between a set of available communication systems for the Myrinet network: Fast Message FM, LC97] University of Illinois, USA) Active Message AM, MC95] 2 (University of Berkeley, USA) BIP [PT98] LIP University of Lyon, Lyon, France) PM [THIS97] RWCP, Japan) and GM [Myr96] Myricom Inc. USA) All those softwares have an implementation of MPI as, most of the time, their own API. In contrast, a large subset of these systems does not ....

Alan M. Mainwaring and David E. Culler. Active messages: Organization and applications programming interface. http://now.CS.Berkeley.EDU/Papers/Papers/am-spec.ps, 1995.


Using Embedded Network Processors to Implement Global.. - Coady, Ong, Feeley (1999)   (6 citations)  (Correct)

....into smaller packets and pipelining those packets through the data transfer between host memory, network interface memory, and the network. We modified the Trapeze Myrinet control program to add a general facility for network processor based message handling, in the style of active messages [16]. To do this, we added a handler ID field to the Trapeze message header. As depicted in Figure 2, the handler ID indexes a handlerdispatch table stored in network interface memory. Each entry of this table records the address of a handler procedure that receives control for incoming messages, ....

A. M. Mainwaring and D. E. Culler. Active messages: Organization and applications programming interface. http://now.CS.Berkeley.EDU/Papers/am-spec.ps, 1995.


A PCI-SCI Bridge for Building a PC Cluster with.. - Acher, Hellwagner, .. (1996)   (9 citations)  (Correct)

....basis for parallel processing on the SMiLE hardware. Work that has already been carried out includes: ffl Port of PVM direct routed data transfers over native SCI to support more efficient message passing programming on clustered workstations [10] ffl Implementation of an Active Message [18] [14] layer on top of SCI shared memory transactions [5] ffl Port of a Unix Shared Virtual Memory (SVM) implementation to make use of the SCI interconnect for page transfers among nodes. Since the SMiLE hardware has not yet been available, research work was carried out on a testbed consisting of two ....

A. Mainwaring and D. Culler. Active Messages: Organization and Application Programming Interface. Computer Science Division, University of California at Berkeley, Nov. 1995. 7


Fast Communication Mechanisms - Coupling Hardware.. - Hellwagner, Karl.. (1997)   (2 citations)  (Correct)

....for user level access not only to the network interface but through remote memory transactions direct access to remote node s local memories. The Active Message layer within the SMiLE project attempts to close this possible security loophole by implementing the Active Message specification 2. 0 [8] which extends the original definition with the notion of communication endpoints and bundles through which protection and shared access to the network interface are provided. By this definition, an endpoint consists of ffl a send pool for outbound messages, ffl a receive pool for incoming ....

Alain Mainwaring and David Culler. Active Messages: Organization and Application Programming Interface. Computer Science Division, University of California at Berkeley, November 1995.


Implementing Automatic Coordination on Networks of.. - Weiss, Knopp, Hellwagner   (Correct)

....and generates an ordinary C program. The distributed version enhances the shared memory version with remote communication. Hence, it is able to work in hybrid configurations with several SMPs connected together. In [18] we have examined several remote communication platforms: Active Messages [17, 14], Nexus [6] Chant [7] TPVM [4] The distributed coordinator implementation has been decided to use Nexus since this package supports both multithreaded shared memory and distributed memory applications on heterogeneous platforms. The implementation uses the dedicated node concept for access ....

A. Mainwaring and D. Culler. Active Messages: Organization and Applications Programming Interface. Technical report, University of California at Berkeley, Computer Science Department, 1995.


Incorporating Memory Management into User-Level Network.. - Welsh, Basu, von Eicken (1997)   (59 citations)  (Correct)

.... appropriate for next generation systems have integrated the NI into the virtual address side of the system where it shares or replicates the CPU s TLB [1,11,14] Less aggressive proposals allocate a physically contiguous buffer region for each application and pin it down to physical memory [2,3,12,15]. The application specifies message buffers using offsets into the buffer region which the network interface can easily bounds check and translate. This prevents the NI from issuing illegal DMA accesses because the virtual to physical mapping is fixed. The work presented here addresses the memory ....

A. M. Mainwaring and D. E. Culler. Active Messages: Organization and Applications Programming Interface. http://now.CS.Berkeley.EDU/Papers/am-spec.ps, 1995.


High-Speed LANs: New Environments for Parallel and .. - Geoffray, Pham..   (Correct)

....BIP is described in more details in [13] 2.2 Other systems and perspectives First introduced in 1997, BIP was more an incremental step in a large family of softwares than a complete new design. Especially on Myrinet, we can find many other communication systems: Active Messages from Berkeley[10], Fast Messages from Illinois [11] U Net from Cornell University [16] and PM from RWCP (Japan) 14] All these systems bypass the operating system to shorten the communication critical path and to limit, or avoid completely, memory copies for bandwidth improvements. Today these techniques are ....

Mainwaring, A. M., Culler, D. E.: Active messages: Organization and applications programming interface. http://now.cs.berkeley.edu/Papers/Papers/am-spec.ps, 1995.


Address Translation Mechanisms in Network Interfaces - Schoinas, Hill (1998)   (9 citations)  (Correct)

....segment. A protection mechanism can grant access for a particular process to send or receive messages to a specific message segment. Thereafter, the NI must enforce the access rights when it sends or receives messages. Equivalent models are described in Berkeley s Active Message specification [27] and Intel s Virtual Interface Architecture [15] 3 Address Translation Properties In this section, we present the properties that the address translation mechanism should satisfy and we argue why we consider them desirable for minimal messaging. First, NI address translation mechanisms must be ....

Alan Mainwaring and David Culler. Active Messages: Organization and Applications Programming Interface, 1995.


Promela++: A Language for Constructing Correct and.. - Basu, Morrisett, von..   (Correct)

....University basu research.bell labs.com jgm cs.cornell.edu tve cs.cornell.edu 1 Introduction Recent research in high speed network interfaces has focused on removing the operating system from the critical path of communication. An effective solution is to provide user level messaging [vEBBV95, MC95, BDF 95, PLC95, BJM 96] a technique that enables applications to send and receive messages without kernel intervention. As a consequence, heavyweight, kernel resident protocols can be replaced by streamlined user level protocols that are tailored for specific applications. Flexibility ....

A. M. Mainwaring and D. E. Culler. Active Messages: Organization and Applications Programming Interface. http://now.CS.Berkeley.EDU/Papers/am-spec.ps, 1995.


Incorporating Memory Management into User-Level Network Interfaces - Matt Welsh (1997)   (59 citations)  (Correct)

.... appropriate for next generation systems have integrated the NI into the virtual address side of the system where it shares or replicates the CPU s TLB [1,12,16] Less aggressive proposals allocate a physically contiguous buffer region for each application and pin it down to physical memory [3,4,13,17]. The application specifies message buffers using offsets into the buffer region which the network interface can easily bounds check and translate. This prevents the NI from issuing illegal DMA accesses because the virtual to physical mapping is fixed and all buffer pages are guaranteed to have ....

A. M. Mainwaring and D. E. Culler. Active Messages: Organization and Applications Programming Interface. http://now.CS.Berkeley.EDU/Papers/am-spec.ps, 1995.


A Language-Based Approach to Protocol Construction - Basu, Hayden, Morrisett, von .. (1997)   (7 citations)  (Correct)

....performance comparable to hand coded versions of a relatively simple protocol. 1 Introduction Recent research in high speed network interfaces has focused on removing the operating system from the critical path of communication. An effective solution is to provide user level messaging [vEBBV95, MC95, BDF 95, PLC95, BJM 96] a technique that enables applications to send and receive messages without kernel intervention. The use of user level networking architectures has pushed in kernel protocol stacks into user space. As a result, heavyweight, kernel resident protocols that often add ....

A. M. Mainwaring and D. E. Culler. Active Messages: Organization and Applications Programming Interface. http://now.CS.Berkeley.EDU/Papers/am-spec.ps, 1995.


High-Performance Local Area Communication With Fast Sockets - Rodrigues, Anderson, Culler (1997)   (38 citations)  Self-citation (Culler)   (Correct)

....targeted at local area networks of workstations, where processing overhead is the primary limitation on communicationsperformance. For low overhead access to the network with a defined, portable interface, Fast Sockets uses Active Messages [von Eicken et al. 1992, Martin 1994, Culler et al. 1994, Mainwaring Culler 1995] An active message is a network packet which contains the name of a handler function and data for that handler. When an active message arrives at its destination, the handler is looked up and invoked with the data carried in the message. While conceptually similar to a remote procedure call ....

....interface, Active Messages, remains a distinct layer from Fast Sockets. This facilitates the portability of Fast Sockets between different operating systems and network hardware. Active Messages implementations are available for the Intel Paragon [Liu Culler 1995] FDDI [Martin 1994] Myrinet [Mainwaring Culler 1995], and ATM [von Eicken et al. 1995] Layering costs are kept low because Active Messages is a thin layer, and all of its implementation dependentconstants (such as maximum packet size) are exposed to higher layers. The Fast Sockets layer stays lightweight by exploiting Active Message handlers. ....

A. Mainwaring and D. Culler. Active Messages: Organization and Applications Programming Interface, September 1995.


Unresponsiveness-Tolerant Collective Communication - Pakin (2001)   (Correct)

No context found.

Alan M. Mainwaring and David E. Culler. Active Messages: Organization and applications programming interface. Technical report, Computer Science Division, University of California at Berkeley, 1995. Available from http:// now.cs.berkeley.edu/Papers/Papers/am-spec.ps.


Distributed Parallel Computing in Networks of.. - Khoussainov, Patel, al.   (Correct)

No context found.

Alan M. Mainwaring, David E. Culler. Active Messages: Organization and Applications Programming Interface. Technical Document, 1995.


Parallel Communication Mechanisms for Sparse, Irregular Applications - Chong   (Correct)

No context found.

A. Mainwaring and D. Culler. Active Messages: Organization and Applications Programming Interface (API V2.0). University of California at Berkeley, Network of Workstations Project White Paper, September 1995.

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