benchmark 1.3

[Francesco Potorti`]

As Rick Niles <address@hidden> pointed out there is a bug
in benchmark 1.3 which sometimes causes a division by zero on slowest
machines.  The patch sent by Matthias Roessler
<address@hidden> fixes this.  

I am collecting the results that many of you are sending to me and to
the list.  I'll collect and post them when I have some.  

I'll also apply some suggestions to the source of benchmark, and
probably we'll use Sparc 10 as a reference, as Rick Niles points out
that it is the machine used as a reference for the Spec95 numbers, and
he has access to one.

-- 
Francesco Potorti`                  Voice:    +39-50-593203
Satellite Network Group             Operator: +39-50-593111
CNUCE-CNR, Via Santa Maria 36       Fax:      +39-50-904052(G3)/904051(G4)
56126 Pisa - Italy                  Email:    address@hidden