🚀 go-pugleaf

In article <7377@pdn.paradyne.com> alan@oz.nm.paradyne.com (Alan Lovejoy) writes:

>... nasa7....matrix300... are real programs, not synthetic or toy
   codes.

NASA7 and matrix300 are not completely real codes. NASA7 is the
standard NASA/AMES hoop to jump over before you can open your mouth.
It contains fully vectorized implementations of the key loops of
several popular algorithms:

1)	matrix multiply (4 way unrolled for vector computers (sic))
2)	fft (radix 2, 2d)
3)	cholesky decomposition
4)	vectorized block tridiagonal solver
5)	"compute solid related arrays, gauss eliminate the matrix
	of wall influence coefficients" (sic) (particle simulation)
6)	"emit new vortices to satisfy boundary condition..." (sic)
7)	invert 3 pentadiagonals simultaneously

Each code runs for a reasonable period of time. All are very
vectorized, in a fashion which some vectorfolk used to complain about
being Cray biased. DH Bailey was the original author.

Matrix300 is really a bit of the livermore suite

C     THIS IS BENCHMARK LBMK14, WHICH PERFORMS
C     VARIOUS MATRIX MULTIPLICATIONS, INCLUDING TRANSPOSES.

Matrix multiply is of sufficient interest to such large groups of
people, that having it as part of SPEC is probably a good idea. The
implementation is orthodox LINPACK (the library).

SPEC is a good thing, but we shouldn't blindly assert that it's all
real codes .... it's good stuff ... and that is more than can be said
for must of the commonly bandied about benchmarks.

--
Keith H. Bierman    |*My thoughts are my own. !! kbierman@Eng.Sun.COM
It's Not My Fault   | MTS --Only my work belongs to Sun* kbierman%eng@sun.com
I Voted for Bill &  | Advanced Languages/Floating Point Group            
Opus                | "When the going gets Weird .. the Weird turn PRO"

"There is NO defense against the attack of the KILLER MICROS!"
			Eugene Brooks