Partner Newsletter Q2 2010 - Intel® Cluster Ready Articles 2-3

INTEL CLUSTER READY NEWS

Intel® Cluster Checker 1.5 Supports New Technologies, Adds New Features

With Intel's launch of the Intel® Xeon processor 5600 and 7500 series, there's also a major revision to Intel® Cluster Checker aimed at helping you incorporate them into your Intel Cluster Ready solutions and get your certified clusters to market more rapidly and with less effort.

Intel Cluster Checker 1.5 supports both processor families. In addition, the new software takes advantage of the latest versions of Intel® tools, which have also been optimized for the new processors. The updated tools include:

• Intel® C++ Compiler 11.1
• Intel® MPI Library 4.0
• Intel® Math Kernel Library 10.2

In addition to supporting the new processors and software tools, Intel Cluster Checker 1.5 has added improvements to speed up the work of designing, certifying, and producing copies of certified designs, plus a broad set of new features and enhancements. Most significant of these is a new automatic configuration feature that can discover elements of each cluster's software and hardware stacks, such as cluster nodes and software library locations. (See Christopher Heller's Tips & Tricks article for information on using the automatic configuration feature.)

Intel Cluster Checker 1.5 also gives you greater flexibility in establishing base performance thresholds. You now have the option of not specifying specific performance thresholds in the input configuration file for these modules:

• hdparm
• hpcc
• imb_pingpong_intel_mpi
• memory_bandwidth_stream
• mflops_intel_mkl

You still need to run these tests to check functionality and consistency across the cluster, and we recommend you continue to establish performance thresholds for them.

The Intel Cluster Ready team has also developed a new guide describing the allowed hardware variances for a certified solution at deployment time. Look for Intel® Cluster Ready Allowed Hardware Variances in the Intel Cluster Ready Document Library. The guide also provides examples of variances that are not allowed without re-certifying the solution. Using the guide to determine what hardware options can be modified in each cluster sale will streamline the deployment process of orders.

To further speed deployments, we've accelerated the execution runtime for the checks needed to verify a design is a copy of a certified design and is working correctly. We've also revised the default behavior of the test modules having a long execution time to achieve a fast but still representative check.

Download Intel Cluster Checker 1.5 and start using it to develop and manufacture clustered solutions based on the Intel Xeon processor 5600 and 7500 series.

Optimizing Your MPI Apps: What's New in Intel® MPI Library 4.0 and Intel® Trace Analyzer and Collector 8?

Updated versions of two major Intel® software tools enable you to quickly deliver maximum end-user performance on Intel® technology-based clusters using the Message Passing Interface (MPI).

Intel® MPI Library 4.0

Implementing the high-performance MPI-2 specification on multiple fabrics, the Intel® MPI Library 4.0 provides new levels of performance and flexibility for applications that execute on clusters of Intel platforms. The library achieves these advantages via improved interconnect support, faster on-node messaging, and an application tuning capability that adjusts to the cluster architecture and application structure. The library's enhanced architecture enables interconnect vendors and Intel processors to better expose their future maximum performance and functionality.

Key features:
• Optimized shared memory path for multi-core platforms allows greater communication throughput and lower latencies
• Native InfiniBand* interface (Open Fabrics Enterprise Distribution (OFED) verbs) supports lower latencies
• Multi-rail capability enables higher bandwidth and increased inter-process communication
• Tag Matching Interface (TMI) support offers higher performance on Qlogic* PSM and Myricom* MX interconnects
• Increased scalability via connectionless Direct Access Programming Library (DAPL*) UD support
• Improved usability, included updated MPI tuner and compatibility with Intel MPI 3.x versions

Intel® Trace Analyzer and Collector 8.0

The Intel® Trace Analyzer and Collector (ITAC) is a powerful tool for visualizing and understanding MPI application behavior, finding bottlenecks in MPI communication, and increasing the performance of parallel cluster applications. ITAC 8.0 is enhanced with new features that accelerate the analysis and tuning cycle of MPI based HPC cluster applications and help even advanced MPI programmers identify further optimization opportunities.

Key features:
• Application Imbalance diagram is an intuitive diagram that simplifies application analysis by combining load imbalance, MPI interconnect time, and other information in one location
• Ideal Interconnect Simulator helps you explore application imbalance by simulating the app's behavior in an "ideal communication environment"
• Custom Plug-in Framework provides a library template and interfaces to help you simulate application behavior over different interconnects

Get the New Software

Start using the new cluster tools. If you've got an existing license, download new copies from the Intel® Registration Center. For evaluation copies, visit www.intel.com/go/mpi and www.intel.com/go/traceanalyzer.


You Talk, We Listen – and Intel® Cluster Ready Gets Better

Over the past year, many partners have given feedback on how to improve Intel Cluster Ready, and we’ve listened and made changes. Here are two that are making Intel Cluster Ready faster, simpler, and more flexible.

• Intel® Cluster Checker version 1.5 includes improvements aimed at simplifying how Intel Cluster Ready processes integrate into building and deploying certified solutions. Overall execution time of Intel Cluster Checker checks to certify or verify a solution is shorter, and you now have the choice of whether to set specific performance thresholds in the input configuration file for a number of modules. See the Intel Cluster Checker 1.5 User's Guide and search the online Intel Cluster Checker Knowledge Base for more information on these and other enhancements, and don't miss Christopher Heller's article in this issue on using the new automatic configuration feature.

• There's a new guide, Intel® Cluster Ready Allowed Hardware Variances, that describes the allowed hardware variances for a certified solution at deployment time. The guide also provides examples of variances that are not allowed without re-certifying the solution. Using the guide to determine what hardware options can be modified in each cluster sale will streamline the deployment process of orders. Download Intel® Cluster Ready Allowed Hardware Variances.

Intel® Enabled Server Acceleration Alliance – Your Recipe for HPC Success!

HPC enables organizations to substantially increase the speed and complexity of computing. Intel® ESAA recipes help you get to market faster by providing validated hardware and software configurations based on Intel® server products that are optimized to meet demanding data-intensive computing challenges. Program membership is free.
See all HPC recipes.
PARTNER HIGHLIGHTS
INDEED, It’s Intel Cluster Ready!
GNS mbH, a new Intel Cluster Ready partner, has registered INDEED, its sheet metal forming simulation package. The INDEED software system offers a unique range of innovative high-precision calculation models for the simulation of forming processes. Learn more about INDEED www.gns-mbh.com/indeed.html

With a long history of simulation and software development for automotive and aerospace companies worldwide, GNS mbH offers engineering services and software products tailored to meet the needs of its most demanding clients and industry's toughest engineering problems. Learn more about GNS mbH www.gns-mbh.com/3.html

 

For more complete information about compiler optimizations, see our Optimization Notice.