Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!pyramid!amdahl!mat
From: mat@amdahl.uts.amdahl.com (Mike Taylor)
Newsgroups: comp.arch
Subject: Amdahl 5990 benchmarks
Message-ID: <30964@amdahl.uts.amdahl.com>
Date: 6 May 88 22:19:44 GMT
Organization: Amdahl Corporation, Sunnyvale CA
Lines: 15
Keywords: System/370 Linpack Livermore kernels


For those who expressed some interest in actual performance numbers
for the Amdahl 5990, I do have some numerical benchmarks.  These all
refer to a single CPU (of 2 or 4).

Harmonic mean of 14 Livermore Kernels (LFK14)   15.26 MFLOPS  *
Harmonic mean of 24 Livermore Kernels (LFK24)   11.9  MFLOPS
Single Precision LINPACK (LIN32)                16.78 MFLOPS
Double Precision LINPACK (LIN64)                14.49 MFLOPS

These numbers are all with the VS-Fortran compiler except for *
which is the Facom compiler (which we have running under UTS).
-- 
Mike Taylor                        ...!{ihnp4,hplabs,amdcad,sun}!amdahl!mat

[ This may not reflect my opinion, let alone anyone else's.  ]