Intel® Software Network

Intel Optimized LINPACK Benchmark packages can help you in your quest to obtain the highest possible benchmark results for Intel® architecture-based systems. These free packages are implementations of the LINPACK benchmarks which use BLAS and LAPACK software that has been highly tuned for maximum performance on Intel® Xeon® processor-based and Itanium®-based systems. These same high performance BLAS and LAPACK routines are available to software developers in the Intel® Math Kernel Library.

Intel offers two LINPACK benchmark packages:

1. Intel Optimized SMP LINPACK Benchmark package
   For use on SMP machines. This package contains implementations of the LINPACK 1000 benchmark code*. See below for performance and download information.
   
   The package includes the executables listed below, as well as shell scripts and input files.
   
   

   Intel Optimized (SMP) LINPACK Benchmark 10.0
   File Name	Description
   linpack_itanium	64-bit program executable for Itanium®-based systems.
   linpack_xeon32	32-bit program executable for Intel® Xeon® processor-based systems.
   linpack_xeon64	64-bit program executable for 64-bit Intel® Xeon® processor-based systems.

   
   
2. Intel Optimized MP LINPACK Benchmark for Clusters package
   This package is an implementation of the Massively Parallel MP LINPACK* benchmark for use on distributed memory computer systems (clusters). Use this package to benchmark your computer cluster for submission to the Top500 Supercomputer list.
   
   

   Intel Optimized MP LINPACK Benchmark 10.0 for Clusters
   File Name	Description
   HPL 1.0a	The complete HPL 1.0a distribution, as well as additional files to make HPL easier to use.
   nodeperf.c	A program to test for DGEMM performance consistency on all cluster nodes.

The Intel Optimized LINPACK Benchmark packages have been designed to save you time. Fewer file downloads, no need to compile, and less iterating are benefits that save your valuable time. The Intel Optimized SMP LINPACK benchmark package doesn’t perform any of the time consuming searching that HPL does, and while the Intel Optimized MP LINPACK benchmark package does do searching, it provides information early in the process so you don’t have to wait until the entire run is completed before reconfiguring and beginning another run. Try the Intel Optimized LINPACK benchmark packages and see how much time you save.

See the LINPACK Benchmark chapter of the Intel MKL Users Guide for more information.

Elapsed   Time              Intel Optimized SMP  h:mm              LINPACK binary	Elapsed   Time                          HPL h:mm
0:00                  Start	0:00                  Start
0:00                  Download Intel                          Optimized                          LINPACK                          package	0:00                  Download HPL
0:10                  Configure scripts	0:10                  Download BLAS
0:20                  Start Run	0:20                  Download MPI
1:00                  Collect Results	0:30                  Build MPI
	1:00                  Build HPL
	1:30                  Start Run 1
	2:30                  Review Results
	2:50                  Change HPL.dat for improvements
	3:00                  Start Run 2
	4:00                  Review Results
	Iterate changing HPL.dat and running until satisfied with results. Can take many iterations.
	8:00                  Call it a day and vow to finish                          tomorrow…

The following charts show the impressive performance that the Intel Optimized SMP LINPACK Benchmark package has achieved:

• 92.9 percent of the theoretical maximum performance on Dual Core Intel® Itanium-based systems.
• 76.6 percent of theoretical maximum performance on Quad-Core Intel® Xeon® processor based Systems.

The Optimized LINPACK and MP LINPACK Benchmark packages listed below are available for free download.

Download now

Do you have feedback on the LINPACK benchmark? You can post a comment at the Intel MKL user forum at http://softwarecommunity.intel.com/isn/Community/en-US/forums/1273/ShowForum.aspx

• Intel® Cluster Tools 

Note: This product includes software developed at the University of Tennessee, Knoxville, Innovative Computing Laboratories.

¹Performance tests and ratings are measured using specific computer systems and/or components and reflect the appropriate performance of Intel products as measured by those tests. Any difference in system design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, go to http://www.intel.com/software/products/.