A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....problematic for fine grained parallelism. Furthermore, the memory cannot be controlled explicitly, such that code and data may be distributed among different processors. In the past, several other parallel programming paradigms were used. Today most of them have lost their importance (e.g. PVM [46]) or never really gained wide acceptance among software and hardware developers (e.g. HPF) due to the lack of portability, a high threshold to use (e.g. Pthreads) or exploring only one key requirement. 4.3 Two parallel implementations Two different parallelization strategies were implemented ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 users guide and reference manual. Technical Manual ORNL/TM-12187, Oak Ridge National Laboratory, May 1994.

....proposes HeNCE [7] which is also a visual parallelprogramming language. Newton compared their characteristics in reference [8] We focus on interprocessor communications in the GRIX system because it represents the most characteristic and di#cult point of message passing parallel programs. XPVM [9] is a GUI system for controlling and visualizing the PVM programs. GRIX and XPVM have the same characteristics for controlling interprocessor communications, but XPVM is designed for run time environments, not for editing programs. 7 Conclusion There are many implementation methods for coding ....

G.A.Geist,A.Beguelin,J.J.Dongarra, W. Jiang, R. Manchek, and V. S. Sunderam. Pvm 3 users guide and reference manual. Technical report, Oak Ridge National Laboratory, 1993. ORNL/TM12187.

....beginning users in particular can be confused by their superficial similarities. Several comparisons of PVM and MPI have been carried out since the mid 1990s [18, 17, 12, 23, 16] We consider it worthwhile to do so again for two reasons. We treat the Oak Ridge version of PVM as represented by [5, 11] as the PVM specification. MPI is represented by the MPI 2 specification. The most obvious is that some convergence has recently taken place in the functionality offered by the two systems (e.g. dynamic processes in MPI, static groups and message contexts in PVM) and the very different ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 1994.

....is achieved by placing appropriate calls to the DLB library to exchange information and redistribute work. The compiler, however, generates code to handle this at run time. Code Generation. For the source to source code translation from a sequential program to a parallel program using PVM [6] for message passing, with DLB library calls, we use the SUIF compiler from Stanford University. The input to the compiler consists of the sequential version of the code, with annotations to indicate the data decomposition for the shared arrays, and to indicate the loops which have to be load ....

....to synchronize. 6. EXPERIMENTAL RESULTS All the experiments were performed on a network of homogeneous Sun (Sparc LX) workstations interconnected via an Ethernet LAN (our model, however, can easily handle processor heterogeneity) External load was simulated as described in Section 4. PVM [6] was used to parallelize the following applications: 1) Matrix Multiplication (MXM) and (2) TRFD, from the Perfect Benchmark suite [8] The overhead of the DLB schemes is almost negligible, since they are receiver initiated, and in the absence of external load, all processors finish work at ....

Geist, A. et al. PVM user guide and reference manual. TM12187, Oak Ridge National Laboratory, May 1993.

....messages that will be exchanged and for governing the translation of messages into a specific wire format. They differ from this research in how they define and manage associated metadata, and in the extent to which they can make use of efficient communication mechanisms. Packages like PBIO, PVM[17], and Nexus[16] support language level features for defining messages in which the communicating applications pack and unpack messages, building and decoding them field by field. Their operational norm is for all parties to a communication to agree a priori on the format of messages. The ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....mechanisms. exchanged and governing the translation of messages into a specific wire format. Their differences with respect to this research lie in how they define and manage this metadata, and to what degree they can make use of efficient communication mechanisms. Packages like PBIO, PVM[17], and Nexus[16] support language level features for defining messages in which the communicating applications pack and unpack messages, building and decoding them field by field. Their operational norm is for all parties to a communication to agree a priori on the format of messages. The ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....and other algorithms, such as Radiance [20] which update the object database in the course of execution. For the current implementation we have used an adapted version of Rayshade [21] which is a sequential public domain ray tracer. The language used is C, extended with the PVM parallel library [22]. Although using standard libraries may be less efficient than hard coding for a specific architecture, it has great advantages in terms of portability. Currently, the demand driven tasks consist of primary rays only. Scheduling is performed statically and there are three scheduling criteria. ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, V. Sunderam, PVM 3 Users Guide and Reference Manual, Oak Ridge National Laboratory, Oak Ridge, Tennessee. Included with the PVM 3 distribution, (1993).

....has been programmed using a message passing programming model, however, it is generally accepted that writing message passing programs is much more difficult and error prone than writing the serial equivalent. Message passing programs typically use a library of communications routines such as PVM, [9], and MPI, 11] and often employ complex techniques such as multi threading or tasking. A simpler programming model known as the Data Parallel Model is becoming increasingly popular due to its simplicity. Features of this programming model are the single thread of control, a global name space ....

A. Giest et al. PVM 3 users guide and reference manual. Technical report, Oak Ridge National Laboratory, May 1994.

....conclusion and discuss some directions for future work in Section 5. 2. Background and Related Work In high performance communication packages, the operational norm is for all parties to a communication to have an aprioriagreement on the format of messages exchanged. Many packages, such as PVM[10] and Nexus[9] support message exchanges in which the communicating applications pack and unpack messages, building and decoding them field by field. Other packages, such as MPI[8] allow the creation of user defined datatypes for messages and fields and provide some marshalling and ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....system can be found in the IBM AIX Parallel Environment: Operation and Use manual (1994) The nodes of the machines communicate using message passing. PVM and MPI are the two principal message passing systems used on the SP1 and SP2. PVM is one of the most popular libraries, developed by Geist, Beguelin, Dongarra, Jiang, Mancek, and Sunderam, 1994). PVM stands for Parallel Virtual Machine. One criticism of PVM is its relatively low performance. Message latencies are high and bandwidth is low. Because of its relatively low performance, PVM was not used for this research effort. MPI was used as the message passing system for this research ....

Geist, A., Beguelin, A., Dongarra, J., Jiang, W., Mancek, R., & Sunderam, V. (1994). PVM 3 users guide and reference manual. Oak Ridge, TN: Oak Ridge National Laboratory.

....order of execution. The second stage of the transformation produces parallelism by introducing message passing actions into the intermediate program. An often cited advantage of functional programming languages is their support for formal equational reasoning. Communications libraries such as PVM [3] and MPI [2] consist of large quantities of C code, thus it is impossible to reason formally about them. We use an abstract message passing model, characterised by a set of laws. Program equivalence proofs are then secure in the context of this model. The message passing model is made concrete by ....

Al Geist, Adam Beguelin, Jack Dongarra, Weicheng Jiang, Robert Manchek, and Vaidy Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

.... clusters, 1 hence most works have focused on this model, although there are also proposals for optimizing a shared memory model over ATM [14] Message passing systems consist of three architectural levels: 1) the interface, usually in charge of a user level library such as p4, MPI and PVM [6]; 2) the transport protocols, which map the library s communication primitives to the characteristics of the network service; and (3) the network service itself. In order to achieve the desired performance, this architecture has to be implemented in such a manner that the performance degradation ....

A. Geist et al. PVM 3 Users' Guide and Reference Manual. Oak Ridge National Laboratory, 1993.

.... to be sent can be further classi ed as non personalized (wherein all receivers will receive a same message from a same source) and personalized (wherein each receiver will receive a di erent message from a same source) Some of these communication patterns have also been implemented in PVM [8] and MPI [19] as communication libraries. In this paper, we study the all to all personalized communication, or known as complete exchange or gossiping, wherein each node needs to send a distinct message to each of the rest of the nodes. This represents the densest communication pattern among ....

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 1994.

....model for such a system became a recurrent theme in the year s work. A later approach to the problem (Section 1.3) provided a direct transformation from the speci cation to a monadic implementation; again the model of processes was unsatisfactory. 6 1. 2 PVM PVM, the Parallel Virtual Machine, [7] is a portable message passing programming system. It is designed to link separate host machines to form a manageable single computing resource the virtual machine. This can be composed of hosts of varying types, in physically remote locations. The system is portable to a wide variety of ....

Al Geist, Adam Beguelin, Jack Dongarra, Weicheng Jiang, Robert Manchek, and Vaidy Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....MLn is to choose which parallel programming paradigm to employ. Given that the existing package is written in C , and that we wish the parallel version to be robust and portable, we are effectively restricted to choosing between a distributed memory, message passing paradigm (as supported by PVM [4] or MPI [15] and a shared memory, multiple threads paradigm (as supported by, for example, POSIX threads [13] There are a number of (possibly conflicting) objectives which the choice of paradigm should satisfy. As well as portability, we should consider ease of programming, efficiency of the 9 ....

Geist, A., A. Beguelin, J. Dongarra, W. Jiang, R. Manchek and V. Sunderam, PVM 3 Users Guide and Reference Manual (Oak Ridge National Laboratory, Oak Ridge, Tennessee, 1994).

....code instrumentation . CORE Binary data transfer support . Communication support EXTENSIONS . Figure 1: Services of Event based Monitoring . In this paper, we focus on the core monitoring service of binary data transfer. Popular approaches to binary data transfer, such as PVM [11] or Nexus [10] support message exchanges in which the communicating applications pack and unpack messages on a per field, per datatype basis. Other packages, such as MPI [9] allow the creation of user defined datatypes for messages and message fields and provide some amount of marshalling ....

Al Geist, Adam Beguelin, Jack Dongarra, Weicheng Jiang, Robert Manchek, and Vaidy Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....time, when comparing it to a non flexible approach like MPI s. 2 Background and Related Work In high performance communication packages, the operational norm is for all parties to a communication to have an a priori agreement on the format of messages exchanged. Many packages, such as PVM[9] and Nexus[8] support message exchanges in which the communicating applications pack and unpack messages, building and decoding them field by field. Other packages, such as MPI[7] allow the creation of user defined datatypes for messages and fields and provide some marshalling and ....

Al Geist, Adam Beguelin, Jack Dongarra, Weicheng Jiang, Robert Manchek, and Vaidy Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....hide at least some of the complexities of binary data interchange among heterogeneous machines. The features and semantics of these packages are typically a compromise between what might be useful to the applications and what can be implemented efficiently. For example, many packages, such as PVM[7], MPI[5] and Nexus[6] support message exchanges in which the communicating applications pack and unpack messages field by field. This approach offers applications significant flexibility in message contents while ensuring that the pack and unpack operations are performed by optimized, ....

Al Geist, Adam Beguelin, Jack Dongarra, Weicheng Jiang, Robert Manchek, and Vaidy Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 94.

....developed for it will be portable across platforms. We note that there are two aspects to portability. First, not all parallel platforms support the same programming language or message passing interface. This problem seems to be resolved by widely available message passing libraries like PVM [Geist et al. 1994] and MPI [Message Passing Interface Forum 1993] However, a more fundamental problem is that many parallel programs are architecture dependent, meaning that they are customized for a particular architecture in order to achieve the highest performance. The result of this practice is that parallel ....

....perform local operations and send messages to other processors. Supersteps are separated by a barrier synchronization, after which it is assured that all messages have reached their destinations. This programming model is different from the one provided by most message passing libraries like PVM [Geist et al. 1994] or MPI [Message Passing Interface Forum 1993] These libraries are based on pairwise sends and receives, whereas in the BSP model explicit receives are unnecessary. Instead, a barrier synchronization signifies the end of all communication operations. The cost of a BSP computation depends on the ....

Geist, A., Beguelin, A., Dongarra, J., Jiang, W., Manchek, R., and Sunderam, V. 1994. PVM 3 Users Guide and Reference Manual. Oak Ridge National Laboratory, Oak Ridge, Tenessee 37831.

No context found.

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 1994.

No context found.

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference Manual. Oak Ridge National Laboratory, 1994.

No context found.

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference manual. Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, May 1994.

No context found.

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, V. Sunderam. PVM3 Users Guide and Reference Manual. Available from netlib@ornl.gov. P. Green. Analyzing Multivariate Data. The Dryden Press; Hinsdale, IL, 1978.

No context found.

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, V. Sunderam. PVM3 Users Guide and Reference Manual. Available from netlib@ornl.gov, 1994.

No context found.

A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, and V. Sunderam. PVM 3 Users Guide and Reference Manual. Oak Ridge National Laboratory, Oak Ridge, Tenessee 37831, May 1994.

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