Printer Friendly

CONVEX COMPUTER INTRODUCES META SERIES LINKING SUPERCOMPUTER PLATFORMS, PA-RISC PROCESSORS

CONVEX COMPUTER INTRODUCES META SERIES LINKING SUPERCOMPUTER PLATFORMS,
 PA-RISC PROCESSORS
 RICHARDSON, Texas, Oct. 13 /PRNewswire/ -- Convex Computer Corporation (NYSE: CNX), unveiling the initial phase of its MPP development strategy, today announced the Meta Series, a cluster system that links Convex supercomputers with Hewlett-Packard Company's (NYSE: HWP) PA-RISC processing nodes.
 The Meta Series announcement comes seven months after Convex and HP signed an extensive business and technology agreement. As part of that agreement, HP assumed a 5 percent ownership stake in Convex and Convex agreed to adopt HP's PA-RISC technologies in the MPP supercomputers that Convex currently has under development. An OEM/VAR agreement, signed by HP and Convex one month ago, provides the framework under which each company can market selected products of the other.
 The Meta Series couples the large, shared memory, vector/parallel architecture, and fast I/O that is characteristic of Convex C Series systems with the high scalar performance of PA-RISC. "The Meta Series hardware, combined with the cluster-specific parallel software tools, provides users with an introductory platform for scalable algorithm and application development that will be employed on our future MPP systems," said Steven J. Wallach, Convex's senior vice president of technology. "This product also gives users access to a high throughput engine for solving complex computational problems."
 Toward Metacomputing -- Empowering the User
 According to Wallach, the Meta Series is one of several Convex products that fit within the context of the metacomputer, a concept pioneered by Dr. Larry Smarr at the National Center for Supercomputer Applications at the University of Illinois. "The metacomputer will empower users by providing transparent access to a variety of heterogeneous computing systems -- such as large vector multiprocessors, MPP systems or clusters -- along with data/file service, sharing the same binary data sets, and visualization tools," said Wallach.
 "The metacomputer is similar to an electricity grid. When you turn on your light, you don't care where the power comes from, you just want the light to come on. The same is true for computer users. When they hit a return, they want their job to run on the best possible machine and they really don't care how that gets done. The metacomputer feels, looks, and operates homogeneously, but physically, it's heterogeneous. This is absolutely what users want in a computing environment, and it's a direction Convex totally supports," added Wallach.
 Linking High-Performance Vector and
 Scalar Technologies
 "Cluster users are innovators trying to solve computationally intensive tasks in applications such as physics, computational chemistry, fluid dynamics, financial analysis, and design verification," said Terrence L. Rock, Convex's chief operating officer. "We believe that combining Convex's C Series systems with PA-RISC nodes extends the user's computing capabilities to include larger data storage options and the ability to process large, single jobs because of greater physical memory. The Meta Series will allow users to expand their range of third-party application software and to speed up design productivity though the use of powerful visualization tools," added Rock.
 The Meta Series cluster is connected by a single ring Fiber Distributed Data Interface (FDDI) link operating at 100 megabits per second. For higher performance applications, Convex has designed a high-speed, low latency, custom interconnect. The Meta Series software has been optimized for both FDDI and the low latency interconnect.
 Although the number of PA-RISC nodes that can be physically attached to a single Convex system is virtually unlimited, the company expects that most users will configure systems ranging from four to 32 nodes. Up to eight PA-RISC processors can be housed in a single cabinet measuring 25 inches wide by 70 inches high. Memory is configured at 32, 64, or 128 megabytes per node.
 The Meta Series, currently in beta test, is expected to be available in the first quarter of 1993 starting at $500,000 for an entry-level system.
 Software Suite Provides Homogeneous Interface
 Working under an independent software vendor (ISV) relationship, Convex has developed a suite of software products that will be used on the Meta Series as well as HP's production cluster, also announced today. "Computing trends are toward increasing parallelism, but the real challenge in this area is software-related. It is software that will provide usable user interfaces," said Wallach. "The Convex Meta Series extends that paradigm by providing a heterogeneous platform with a homogeneous software interface." Convex's new software suite includes:
 -- ConvexPVM (Parallel Virtual Machine), a popular message passing library jointly developed by Oak Ridge National Laboratory and the University of Tennessee, that allows application programs to run in parallel over the cluster. ConvexPVM treats heterogeneous computer systems and networks as components of the same machine. It provides a homogeneous environment that allows data sharing using the same IEEE floating point format.
 -- ConvexNQS+, a network queuing program offering automatic cross mounting, a feature that allows data files resident on the Convex system to be mounted on the executing PA-RISC processing node. ConvexNQS+ also supports automatic load balancing, which is used to route a job to under-utilized processors so it can be completed faster.
 -- ConvexMLIB, a collection of highly optimized subroutines of basic mathematical and computational kernels to enhance program performance on the PA-RISC clusters. These subroutines accomplish many of the commonly occurring tasks in scientific programming, which helps to improve user productivity.
 In addition, Convex has under development a comprehensive set of Administration Tools for the Meta Series. This software facilitates the management and administration of the Meta Series -- such as adding or removing users, setting up ConvexNQS+ queues, and installing new software releases -- through a single set of commands. By using these tools, the appropriate actions are executed on all Meta Series processing units thus saving system administration time and effort.
 Convex Establishes MPP Computing Group,
 Cluster Integration Support
 Convex is the founder and a sponsoring member of the "Scalable Computing Working Group," an independent organization of scientists and engineers from industry, academia, and research. The SCWG meets every three months to advance scalable algorithm and application development. The group's first meeting was Sept. 14, 1992.
 "We are dedicated to easing the transition to MPP for those customers who have applications that will benefit from this processing technology. In sponsoring the group, we will make our engineers available to the members for collaboration and consultation and offer members early access to our scalable computing technology," explained Rock.
 Convex also established a cluster integration group that will work with both Convex and HP customers to provide a single source of support for the Meta Series product and the HP clusters that use Convex software. "Until today, the state of cluster customer support meant you had to call three to five different vendors to get your cluster question answered. We're changing that by offering support from one source -- Convex," said Rock.
 Cluster labs, where customers can work with engineers to investigate emerging parallel technologies and further their parallel efforts, have been set up at Convex's headquarters as well as at HP's Fort Collins, Colo., facility.
 Convex Computer Corporation, a leading supplier of air-cooled supercomputers worldwide, markets its products primarily to scientific, engineering, and technical users for a wide variety of applications in areas such as seismic processing, reservoir simulation, computational chemistry, computer-aided engineering, image processing, aerospace simulation, and molecular biology. Convex is a member of the Precision Risc Organization (PRO), an association of industry-leading companies that promote the use of HP's PA-RISC technology. Convex, which is listed on the NYSE with the symbol CNX, has sold more than 1,100 systems to nearly 600 customers in 44 countries. The systems are sold and serviced through direct sales and an extensive distriution network.
 -0- 10/13/92
 /CONTACT: Donna Burke, 214-497-4230, or Mary Dudley, 214-497-4109, both of Convex/
 (CNX HWP) CO: Convex Computer Corporation; Hewlett-Packard Company ST: Texas IN: CPR SU: PDT


GK -- NY061 -- 9397 10/13/92 14:15 EDT
COPYRIGHT 1992 PR Newswire Association LLC
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:PR Newswire
Date:Oct 13, 1992
Words:1297
Previous Article:TOPOX: AGREEMENT SIGNED TO MANUFACTURE INNOVATIVE MEDICAL DEVICES
Next Article:UCL ORDERS FIRST CONVEX META SERIES HIGH-PERFORMANCE COMPUTER
Topics:


Related Articles
CONVEX COMPUTER ADDS NEW ENTRY-LEVEL SUPERCOMPUTER TO ITS C3400 FAMILY (Product Announcement)
WANG ANNOUNCES FOUR NEW MODELS IN RISC SERIES LINE (Product Announcement)
WANG ANNOUNCES EASY-TO-USE PRODUCTS TO LINK UNIX SYSTEMS WITH PC LANS
HP, CONVEX ANNOUNCE CLUSTER COMPUTING SOLUTIONS; SYSTEMS TARGET COMPUTE-INTENSIVE TECHNICAL MARKETS
SILICON GRAPHICS UNVEILS NEW RANGE OF GRAPHICS SUPERCOMPUTERS
IBM ANNOUNCES NINE NEW MODELS, PARALLEL PROCESSING SYSTEM, ENHANCED GRAPHICS, STORAGE FEATURES FOR RISC SYSTEM/6000 FAMILY
CONVEX ADDRESSES THE DATA EXPLOSION WITH A BROAD RANGE OF DATA MANAGEMENT SOLUTIONS
CRAY RESEARCH DOUBLES TOP PERFORMANCE OF FAST-SELLING CRAY J90 PRODUCT LINE
CONVEX AND SILICON VALLEY RESEARCH ANNOUNCE FIRST SCALABLE COMPUTING SOLUTION FOR EDA MARKET
AlternaTek Introduces New Line of Supercomputers Specifically Designed For Financial Simulations

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters