Talk:LINPACK

	This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing articles
Mid	This article has been rated as Mid-importance on the project's importance scale.
	This article is supported by WikiProject Software (assessed as High-importance).
	This article is supported by Computer hardware task force (assessed as High-importance).

There are now actually 3 benchmarks called LINPACK (which is confusing).

The first is the original benchmark by Jack Dongarra. The rules state that you are not allowed to change the code in any way, not even the comments. It is therefore a test of your Fortran environment rather than the top speed of your computer. The size of the matrix is small by today's standards.
The second is the LINPACK 1000, or TPP benchmark ("Toward Peak Performance) on nxn matrices with n=1000, but you are allowed to replace calls to particular functions with your own linear equation solver, and even change n.
The third and newest benchmark is the HPC benchmark ("Highly Parallel Computing"), appropriate to clusters where the work can be distributed via MPI or PVM.

I will try to flesh in the article to make these distinctions clearer, and verify that I've actually got the particulars right.

--Wormholio 16:05, 27 February 2007 (UTC)[reply]

A reference for all this is the paper Performance of Various Computers Using Standard Linear Equations Software by Jack J. Dongarra, of the Computer Science Departemnt at the University of Tennessee, Knoxville, TN 37996-1301 and the Computer Science and Mathematics Division of Oak Ridge National Laboratory, Oak Ridge, TN 37831. Document code CS-89-85 dated 30 November 20006. I will track down the link directly to this paper.

--Wormholio 16:11, 27 February 2007 (UTC)[reply]

The paper in question is available at http://www.netlib.org/benchmark/performance.ps Jbradfor 20:54, 8 March 2007 (UTC)[reply]

I think the LINPACK benckmarks, particularly HPL, are important enough, since HPL is used for the Top500 ranking, to have an article of their own, and not just a section in the LINPACK package article. I will create a new article named "LINPACK benchmarks" and extend it, and leave this article just for the package.

--Diego Marcos (talk) 11:23, 8 January 2012 (UTC)[reply]

The HPL - A Portable Implementation of the High-Performance Linpack Benchmark for Distributed-Memory Computers portion of this should be merged into TOP500. The original LINPACK library should be merged into LAPACK (along with EISPACK) and some context given on their historical evoluation. Jeff.science (talk) 09:05, 4 December 2012 (UTC)[reply]