Intel Corporation, Paragon OSF/1 User's Guide, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....programs. The results of the study are presented in Section 3. Section 4 presents conclusions and directions for future work. 2 2 Experimental Setup 2.1 The Intel Paragon This section briefly describes the architecture of the Paragon Supercomputer. A detailed description may be found in [Int93]. An Intel Paragon Supercomputer consists of several nodes 1 connected in a mesh configuration. The computer used for this study has 66 nodes connected in a 11 by 6 matrix. Several other nodes are dedicated to special tasks such as disk and network control. Each node consists of two Intel i860 ....

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

....The operating system conforms to OSF 1 standards. Although the machine is capable of supporting MIMD (or MPMD) computing model, most applications use the SPMD model for computing. The workloads for this study use the SPMD model. A detailed description of the Paragon architecture may be found in [INT93]. 2.2 The Adaptive Scheduling Policy The adaptive scheduling policy chosen for this study is the Robust Adaptive (RA) scheduling policy described in [RSDSC94] This policy has been shown to have better performance than several other non preemptive scheduling policies. The RA scheme actually ....

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

....and techniques presented in this paper are designed for machines with multidimensional mesh interconnects using wormhole routing. Commercial examples of such machines include the Intel Paragon and the Cray T3D, which utilize two dimensional non torus and threedimensional torus meshes, respectively [7,12]. They are also appropriate for research machines such as the Intel Touchstone Delta (two dimensional non torus) and the J and M Machines (three dimensional non torus) 9,13,14,15,17] Many of these algorithms and techniques also apply to hypercubes (which can be viewed as multidimensional meshes ....

Intel Corporation Supercomputer Systems Division. Paragon OSF/1 User's Guide. 1993.

....in section 3.2 the sz and on switch may be used to select some of the nodes within a partition to run application processes. With the mapping table as the partition definition there is no change necessary to those switches. 10 See page 41 in appendix B 11 For a further discussion refer to [7] NXLib Users Guide Version V1 1 4 6 MappingTable VirtualParagonNode ffl ffi fi fl VirtualParagonNode NodeNumber BoT Workstation fi ffi WorkstationPath ffl fl BoT Login fi ffi ffl fl BoT Arch fi ffi ffl fl NodeNumber NATURALINT WorkstationPath Workstation : ffl ffi fi fl ....

....are available within the NXLib environment. Calls which are additionally available will be explained in section 4.1. Finally section 4.2 will show differences between the Paragon environment and NXLib. Appendix A will give a list of NX message passing calls which are available within NXLib. In [5, 6, 4, 7] an intensive discussion is given. A reference to the commands is provided in appendix B. 4.1 Additional user commands 4.1.1 The compiler front ends The C, C , and Fortran compilers on a Paragon use two special linker switches to generate parallel application. The nx and lnx switch both ....

Intel Supercomputer System's Division, 15201 N.W. Greenbrier Parkway, Beaverton, OR 97006. Paragon OSF/1 User's Guide, 1 edition, April 1993. www: http://www.ccsf.caltech.edu/paragon/man.html.

....other lightweight thread packages. Chant can also accommodate both preemptive and non preemptive thread scheduling policies. 2. 2 Interprocess Communication Communication systems for distributed memory architectures have traditionally been provided by the vendors, such as the Intel NX primitives [26] and nCUBE Vertex primitives [34] In response to the increasing demands of portability, several communication libraries have been established that provide a portable message passing interface over a wide variety of systems. Among these libraries, p4 [8] and PVM [44] have received the most ....

Intel Corporation, Beaverton, OR. Paragon OSF/1 User's Guide, April 1993.

....but a single host remains responsible for the control and access of the whole machine. The Intel Paragon was one of the first machines that was host less and allowed the system administrator to tailor the size of the service partition to the number of users and processes present on the system [14]. Intel s OSF 1 AD operating system provides a single system image that gives users the illusion of a single computer, no matter how many service nodes are configured. The drawback of this particular implementation is that the failure of a single node requires the reboot of the whole machine to ....

Intel Corporation. Paragon OSF/1 User's Guide, September 1992.

....calls, NX provides calls for the process management of parallel applications. Cooperating processes address each other via a node number and a process type (ptype) The node number is derived from the node where the process is executing, whereas the ptype can be modified via corresponding calls [3, 4, 2, 5]. The following section will introduce the concepts whichwere necessary to offer a similar system image on a network of workstations to the one available on the Paragon. This will include a more detailed discussion of some Paragon features where it seems appropriate. 3 The Design of the Paragon ....

Intel Supercomputer System's Division, 15201 N.W. Greenbrier Parkway, Beaverton, OR 97006. Paragon OSF/1 User's Guide, 1 edition, April 1993.

....Further, if each VP explicitly accesses its own subfile, many fine grained file operations are required. Performance suffers significantly when this approach is used. Modes were introduced with the earliest parallel file systems [20, 25] and some more recent systems have adopted this approach [2, 11]. The systems provide a limited number of operations whose synchronization requirements and file access semantics vary depending on the current mode. The implementations of modes on these systems have several drawbacks. They are machine dependent. They give a single file operation multiple ....

....routines make up a majority of the Stream implementation, while a few routines make calls to the machine dependent file system. We assume the underlying file system manages a parallel file in the traditional manner, as a single stream of bytes. Therefore, Stream can be implemented on top of PFS [11], Thinking Machines CMMD I O [2] and the single stream views of more flexible file systems like PIOUS [19] and Vesta [3] Knowledge of the underlying implementation of the file system (e.g. programmable I O nodes, disk arrays, etc. may be used to optimize the machine specific portions of the ....

Intel. Paragon OSF/1 User's Guide, 1993.

....execute in time sharing mode [ZB91, SST93] The focus of this paper is on non preemptive policies as they represent a viable solution for multiprogramming parallel systems. Non preemptive policies are a simple on line alternative to off line allocation policies based on users requests exclusively [Int93]. Off line allocations, either in batch or interactive mode, are often the strategy used on actual parallel machines that operate in production mode. By assigning the number of processors requested by the user, the application s execution time is guaranteed but not the waiting time. ....

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

....for Large Scale MPPs Philip M. Papadopoulos G.A. Geist, II y Abstract This paper presents early experiences with using high speed ATM interfaces to connect multiple Intel Paragons on both local and wide area networks. The testbed includes the 1024 and 512 node Paragons running the OSF [4] operating system at Oak Ridge National Laboratory and the 1840 node Paragon running the Puma [5] operating system at Sandia National Laboratories. The experimental OC 12 (622 Mbits sec) interfaces are built by GigaNet and provide a proprietary API for sending AAL 5 encapsulated packets. PVM is ....

Intel Corporation, Paragon OSF/1 User's Guide, Beverton, Oregon, Document number 312489-001, April 1993.

....virtual machine. Further, if each VP explicitly accesses its own subfile, many finegrained file operations are required. Performance suffers significantly when this approach is used. Modes were introduced with the earliest parallel file systems [22, 27] and some more recent systems use modes [2, 12]. The systems provide a limited number of operations whose synchronization requirements and file access semantics vary depending on the current mode. The implementations of modes on these systems have several drawbacks. They are machine dependent. They give a single file operation multiple ....

....routines make up a majority of the Stream implementation, while a few routines make calls to the machine dependent file system. We assume the underlying file system manages a parallel file in the traditional manner, as a single stream of bytes. Therefore, Stream can be implemented on top of PFS [12], Thinking Machines CMMD I O [2] and the single stream views of more flexible file systems like PIOUS [21] and Vesta [3] Knowledge of the underlying implementation of the file system (e.g. programmable I O nodes, disk arrays, etc. may be used to optimize the machine specific portions of the ....

Intel. Paragon OSF/1 User's Guide, 1993.

....in a rope The key to collective operations is the ability for the programmer (or compiler) to specify the scope of the operation; that is, the entities that will be involved in the operation. Collective operations are typically supported at the process level by the underlying communication system [11], or by standard communication interfaces [6, 22] For example, MPI [6] provides a mechanism for process scoping called groups. However, support for grouping threads within processes is not currently supported by either MPI or the new thread based runtime systems yet such support is clearly ....

Intel Corporation, Beaverton, OR. Paragon OSF/1 User's Guide, April 1993.

....system. Actual applications were scheduled to arrive according to a Poisson arrival stream, which were then actually executed on the assigned partition. The experimental platform is a 512 node Intel Paragon, a wormhole routed message passing multiprocessor system with distributed memory [4]. Results are presented for two system sizes, 32 and 64 processors. Experiments with 128, not reported here, confirm the findings of the smaller systems. The implemented workload is an LU decomposition algorithm. It is executed on various matrix sizes, yielding different speedup curves. Four ....

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

....mixes comprising various classes of components. Experimentation is used to investigate the impact of single and multiclass workloads on the performance under p sav policies. In our experimental setting, a 512 node Intel Paragon, a message passing multiprocessor system with distributed memory [7], is the experimental platform. Experiments with 32, 64, and 128 processors were conducted and results are presented for 64 processors. The omitted cases exhibit similar behavior to the results described here. Actual applications are submitted to the scheduler according to a Poisson process. The ....

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

....to be superscalar. We use a linear model (fi m ) for its performance on vector of size m. In all our experiments we use 64 bit double precision floating point arithmetic. This model corresponds roughly to the currently available commercial machines, such as the Intel iPSC 860 [Dun90] and Paragon [EK93, Int93]. In the remainder of this paper we call fi scal and scal the parameters for the cost of the inversion of the pivot element, the search for the maximum of the vector, the interchange of these two elements, and the scale of the vector by a scalar using the Level 1 BLAS DSCAL function [DCDH90] We ....

Intel Corporation - SSD, Beaverton, OR. PARAGON OSF/1 - User's guide, April 1993.

....explicit thread management routines, encoded for a particular architecture, rather than a portable lightweight threads library. 2. 2 Communication Libraries Communication systems for distributed memory architectures have traditionally been provided by the vendors, such as the Intel NX primitives [17] and nCUBE Vertex primitives [24] In response to the increasing demands of portability, several communication libraries have been established that provide a portable message passing interface over a wide variety of systems. Among these libraries, p4 [4] and PVM [30] have received the most ....

Intel Corporation, Beaverton, OR. Paragon OSF/1 User's Guide, April 1993.

No context found.

Intel Corporation - SSD, Beaverton, OR. PARAGON OSF/1 - User's guide, April 1993.

No context found.

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

No context found.

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

No context found.

Intel Corporation, Paragon OSF/1 User's Guide, 1993.

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