linpack: Linpack Benchmark

In shinra-dev/okcpuid: Interface to the cpuid library

Description

The Linpack benchmark is the standard performance benchmark for numerical computing. Research institutions, governments, and companies with very large computing resources "compete" to build machines that will perform the best on the benchmark. Results of this competition are released twice a year, in June and in November, at the ISC and SC conferences, respectively. Although it is generally understood that Linpack no longer adequately measures the performance capabilities of a system on real problems, it has remained an important measurement all the same. One of the most attractive things about the Linpack benchmark is its host of historical data, which dates back all the way to 1993.

Usage

linpack(nmin = 1000, nmax = "choose", by = "doubling", warn = TRUE,
  verbose = TRUE)

Arguments

nmin	Minimum problem size.
nmax	Maximum problem size. Value should either be an integer greater than nmin or it should be the string "choose". In the former case, the supplied value will be the largest value considered. In the latter case, the largest problem size will be 2/3 of max system ram.
by	How to get from nmin to nmax. Value should either be a numeric value greater than 0, or it should be the string "doubling". In the former case, seq.int() is used, while in the latter, values starting with nmin will be doubled until nmax is reached/surpassed.
warn	Logical; this controls the printing of a warning that I very seriously want you to read. Set warnme=FALSE to disable after you have read it.
verbose	Logical; determines whether or not intermediate results should be printed.

Details

If you are even somewhat serious about using this as a test of the host system's capabilities, I would strongly encourage you to run this exclusively; that is, shut down all other resource-intensive applications (web browsers, system updates, antivirus, whatever). If a lot of your ram is used up by other utilities and the benchmarker has to swap them to disk, the performance will degrade by several orders of magnitude.

Value

The return is a list (formally an S3 'linpack' object), containing:

R.max

The max

References

http://www.top500.org/project/linpack/

shinra-dev/okcpuid documentation built on May 29, 2019, 9:25 p.m.