4192 - WHITE PAPER Technical Advances in the SGI UV...

Info iconThis preview shows pages 1–4. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: WHITE PAPER Technical Advances in the SGI UV Architecture WHITE PAPER Technical Advances in the SGI UV Architecture 2 sgi.com/uv TABLE OF CONTENTS 1. Introduction 1 2. The SGI UV Architecture 2 2.1. SGI UV Compute Blade 3 2.1.1. UV_Hub ASIC Functionality 4 2.1.1.1. Global Register Unit (GRU) 4 2.1.1.2. Active Memory Unit (AMU) 5 2.2. SGI UV Platform Interconnect Topology 6 3. Integrated System Functionality 7 3.1 The MPI Offload Engine (MOE) 7 3.1.1. Reduced CPU Overhead and Latency for MPI_Send 8 3.1.2. Fast Barriers and Reductions in Hardware 8 3.1.3. Accessing the MOE 8 3.2. Petascale Memory and Data Intensive Computation 8 3.3. Massive I/O Capabilities 9 3.4. Petascale Compute Enhancements 9 3.5. Enhanced Reliability 9 4. Conclusion 10 WHITE PAPER Technical Advances in the SGI UV Architecture 1 sgi.com/uv 1. Introduction Global shared memory (GSM) architecturesin which all processors can access all memory directly offer significant advantages for a broad class of applications including scientific and engineering applications, real-time complex event processing, very large databases (VLDBs) and advanced business applications. SGI is the industry leader in the development of GSM platforms that deliver exceptional processing, memory and I/O capabilities for high performance computing (HPC) and data-intensive tasks. The SGI UV is the fifth generation of the companys scalable global shared memory architecture. SGI started shipping GSM systems with the 64-core SGI Origin systems in 1996. The new SGI GSM platform is the UV family of systems, which scales to 2,560 cores and 16 terabytes (TB) of memory. These systems are built using SGI patented NUMAlink interconnect that provides the high-bandwidth, low-latency, coherence-optimized functionality required by GSM systems. The NUMAlink interconnect fabric can also be used for efficient communication between OS instances, enabling scaling up to many thousands of CPU cores for MPI applications and those developed using partitioned, global address space compilers like OpenMP or Unified Parallel C. SGI UV adds new architectural capabilities which enhance application scaling and performance, independent of the programming model or models employed. These systems also leverage the proven economy of the Intel Xeon processor family and standard Linux while maintaining compatibility with previously written applications. SGI UV adds new architectural capabilities which enhance application scaling and performance, independent of the programming model or models employed. These systems also leverage the proven economy of the Intel Xeon processor family and standard Linux while maintaining compatibility with previously written applications....
View Full Document

This note was uploaded on 01/12/2012 for the course CMSC 714 taught by Professor Staff during the Fall '07 term at Maryland.

Page1 / 12

4192 - WHITE PAPER Technical Advances in the SGI UV...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online