SGI DEBUTS OPTIMIZED LINUX ENVRONMENT FOR HIGH-PERFORMANCE.
SGI (SGI, Sunnyvale, CA, www.sgi.com) A manufacturer of workstations and servers, founded in 1982 by Jim Clark. The company was founded as Silicon Graphics, Inc., but changed to its acronym in 1999. (NYSE NYSE
See: New York Stock Exchange :SGI), Mountain View, Calif., has announced leadership in Linux(R) for technical computing with an optimized Linux software environment that scales to hundreds of Intel(R) Itanium(R) 2 processors, with up to 64 processors in a single Linux node. The company has achieved this feat with an exclusive suite of tools and features designed to maximize performance on the most demanding technical and scientific applications. Offered exclusively for its SGI(R) Altix(TM) 3000 family of servers and superclusters, SGI's software environment also offers a unique capability for Linux clusters -- global shared memory (1) Using part of main memory to support a low-cost display circuit that does not have its own memory. See shared video memory.
(2) The common memory in a symmetric multiprocessing system that is available to all CPUs. See SMP.
1. across cluster nodes -- raising the bar for open source computing and creating the most powerful Intel Itanium 2 processor-based computing environment in the marketplace.
A key component of SGI's software suite is SGI ProPack(TM), a new set of high-performance Linux optimizations that enhances overall system scaling, data handling and resource management while maintaining binary compatibility See binary compatible. with existing 64-bit Linux applications running on the Intel Itanium 2 processor. Additionally, a complement of SGI(R) filesharing and data migration tools enables technical and creative professionals to optimize workflows and manage their largest and most complex data.
"For the first time, developers can advance beyond any perceived limits of Linux to tackle their most data-intensive problems with high-performance 64-bit solutions on massively scalable, big memory systems," said Jan Silverman, senior vice president for SGI. "With today's news, SGI is enabling an entirely new class of Linux applications."
Fueled by two decades of leadership in high-productivity computing, the SGI implementation uniquely leverages the company's extensive experience with scaling, data handling, and large system management in its IRIX A Unix-based operating system from SGI that is used in its computer systems from desktop to supercomputer. It is an enhanced version of Unix System V Release 4. IRIX integrates the X Window system with OpenGL, creating the first real time 3D X environment. (R) operating system operating system (OS)
Software that controls the operation of a computer, directs the input and output of data, keeps track of files, and controls the processing of computer programs. . As a result, the Linux environment from SGI will support thousands of processors in a shared memory supercluster su·per·clus·ter
A group of neighboring clusters of galaxies.
A large group of neighboring clusters of galaxies, along with isolated galaxies scattered between them, the entire collection .
Optimized for users in physical and life sciences, manufacturing, oil and gas, and government and defense markets, this powerfully enhanced environment includes features never before available on Linux. The SGI enhancements to Linux support the third-generation SGI(R) NUMAflex(TM) architecture and modular system interconnect, allowing users to further scale these powerful systems to hundreds and even thousands of processors by combining the computing power of multiple systems into superclusters capable of tackling the most complex computing problems.
For example, life sciences researchers could accelerate their drug discovery analytical runtimes by a factor of 10 by taking all of their genomics databases off of disk and into shared memory. Currently, many of these analyses take several weeks as researchers assess potential compounds sequentially against dozens of genomics databases in a traditional distributed memory (architecture) distributed memory - The kind of memory in a parallel processor where each processor has fast access to its own local memory and where to access another processor's memory it must send a message via the inter-processor network.
Opposite: shared memory. Linux cluster. By simultaneously putting all databases into memory, researchers can reduce their runtimes to a matter of days, thus requiring far fewer processors and overall compute cost.
SGI ProPack(TM) for Linux(R) Fuels Performance Gains
Resource management tools within the SGI ProPack v2.1 help users achieve excellent, repeatable performance on real-world applications, and increase the system's productivity by tuning processors and memory allocation Reserving memory for specific purposes. Operating systems and applications generally reserve fixed amounts of memory at startup and allocate more when the processing requires it. If there is not enough free memory to load the core kernel of an application, it cannot be launched. . SGI advancements, including its NUMA (Non-Uniform Memory Access) A multiprocessing architecture in which memory is separated into close and distant banks. NUMA is similar to SMP, in which multiple CPUs share a single memory. However, in SMP, all CPUs access a common memory at the same speed. architecture and its specially tuned libraries, streamline the process of achieving massive and reliable application scalability.
SGI ProPack data management tools, including the SGI(R) XFS XFS X Font Server (Sun)
XFS Extended File System
XFS X-Fleet Sentinels (gaming clan)
XFS Extensions for Financial Services (software interface specification) (TM) file system, cost-effectively maximize I/O (Input/Output) The transfer of data between the CPU and a peripheral device. Every transfer is an output from one device and an input to another. See PC input/output.
I/O - Input/Output performance, robustness, and flexibility. These tools intelligently manage data and I/O to achieve industry-leading real-world performance. For instance, the SGI ProPack environment has demonstrated in excess of 2GB per second of sustained I/O throughput, enabling Linux applications to overcome the increasing challenge of handling big data in high-performance computing High-speed computing, which typically refers to supercomputers used in scientific research. by increasing throughput to levels beyond those attained by most UNIX UNIX
Operating system for digital computers, developed by Ken Thompson of Bell Laboratories in 1969. It was initially designed for a single user (the name was a pun on the earlier operating system Multics). (R) operating systems.
System management tools, which help users employ resources efficiently and without interruption, include partitioning, which allows system managers to maximize resilience and eliminate single points of failure, and Performance Co-Pilot(TM), which tracks performance at the system-resource level to help identify potential areas for optimization.
Examples of ProPack value-added performance and features include XSCSI XSCSI X Small Computer System Interface high performance I/O infrastructure, MPT MPT Maryland Public Television
MPT Modern Portfolio Theory (investing)
MPT Ministry of Posts and Telecommunications
MPT Message-Passing Toolkit
MPT Master of Physical Therapy
MPT Mitochondrial Permeability Transition (SGI's tuned MPI MPI - Message Passing Interface library), SCSL SCSL Sun Community Source Licensing (Sun)
SCSL Scientific Computing Software Library (Silicon Graphics, Inc.)
SCSL Satyam Computer Services Limited (Secunderabad, India) (math library), FFIO FFIO Flanders Foreign Investment Office (Belgium)
FFIO flexible file input/output
FFIO Freakin' Figure It Out (polite form)
FFIO Flexible File Input Output (memory buffering library), NUMA tools, high performance file systems (XFS), and SGI platform hardware support and drivers.
Global Shared Memory Across Clustered Nodes
With the debut of SGI Altix 3000, SGI introduces the ability to share memory across clustered nodes (each up to 64 processors) to vastly improve application performance and to greatly simplify deploying and managing a multi-node system. By holding more complex job geometries and complete workflows in memory, SGI Altix 3000 also enables new application breakthroughs that traditional Linux clusters can't tackle.
Existing cluster systems are designed to break up memory and distribute it over multiple CPUs for processing, then reintegrate re·in·te·grate
tr.v. re·in·te·grat·ed, re·in·te·grat·ing, re·in·te·grates
To restore to a condition of integration or unity.
re the data to deliver results. Such distributed memory processes add overhead in terms of time, cost and complexity, and have the potential to corrupt the integrity of the results. SGI Altix 3000 superclusters dramatically reduce the time and resources required to run technical Linux applications by managing extremely large data sets in a single, system-wide, shared-memory space. The result is breakthrough system and price performance in an easy-to-manage, standards-based environment. All major parallel programming models, as well as MPI, are supported by SGI Altix 3000 systems.
Ultra-Fast Data Management
The SGI Altix 3000 family also addresses data management, a fundamental weakness of clusters. With CXFS CXFS Clustered Extended File System (SGI) (TM), SGI provides the industry's fastest shared filesystem for storage area networks (SANs). CXFS enables cross-platform data sharing and reduced costs in data-intensive environments by eliminating file duplication and the time it takes to move large files over the network. SGI's storage solutions for SGI Altix 3000 superclusters are designed to solve the data access and management problems unique to engineering, scientific, and creative customers engaged in large-scale computing and visualization. All of this is built on an I/O foundation that more than quadruples the typical I/O performance of other standard Linux systems.
A New Opportunity for the Open Source Community
SGI has long been a supporter of the open source community, and SGI is committed to evolving the scalability, performance and reliability of Linux. SGI has contributed to Linux scalability, scheduling, memory usage, I/O, and other efforts critical to high-demand application performance. These investments have resulted in many enhancements and features in today's standard Linux distributions, along with additional tools to maximize computing productivity.
The SGI optimization of Linux for the SGI Altix 3000 family supplements the standard Linux distribution with a robust set of development, performance and resource management tools for solving compute and data-intensive problems. Only SGI offers users these levels of high-productivity optimization in a fully supported, standard Linux environment.
Availability, Support and Development Tools
Linux for SGI ships with all SGI Altix 3000 models through SGI worldwide sales offices and authorized resellers. SGI provides complete technical support worldwide for its high-performance Linux environment, and works closely with third parties and the open source community to ensure the availability of a rich and robust software offering. To further evolve the market for 64-bit Linux applications, SGI also expects to announce a Linux development tool kit in the near future.
SGI, also known as Silicon Graphics, Inc., is the world's leader in high-performance computing, visualization and the management of complex data. SGI products, services and solutions enable its technical and creative customers to gain strategic and competitive advantages in their core businesses. Whether being used to design and build safer cars and airplanes, discover new medications and oil reserves, predict the weather, entertain us with thrilling movie special effects or provide mission-critical support for government and defense, SGI systems and expertise are empowering a world of innovation and discovery. The company is headquartered in Mountain View, Calif., and has offices worldwide.
For more information, visit http://www.sgi.com or call 650-933-4519.