benchmrk.txt    Mark Riordan    mrr@scss3.cl.msu.edu

This is a list of timings for the generation of a particular
516-bit key by RIPEM.  This benchmark is run automatically by 
"make" when you generate the UNIX version of RIPEM.  It can also
be run by "make benchmark" from UNIX, or by running
test\testgen.bat from Microsoft operating systems.
In any case, the command line used is:
   ripem -g -H . -k test1pw -u test -R c -C b
which generates a particular 516-bit key.

This file contains only RIPEM version 2.0 and higher benchmarks.
The x86 version of RIPEM has assembly code in a crucial routine
in RSAREF, which benefits the Intel platforms.

Times are in seconds; user time + system time for UNIX,
or elapsed time for MS operating systems.


Machine                     Time    Notes (see #'s at bottom)
                            secs 
DEC Alpha 400MHZ             0.4    #02: Digital UNIX V4.0C gcc, RIPEM 3.0b4
IBM/RS6000 43P133            0.53   #04: AIX 4.1.5, IBM's xlc -O3, RIPEM 3.0b4
Pentium 133, 512K cache      0.931  #03: WinNT 5.0b1, MSVC 11.00.7022, R 3.0b4
SGI Challenge L              1.005  #06: IRIX 6.2, RIPEM 3.0b3?
Sun UltraSPARC-II-248        1.5    #02: Solaris 2.6 gcc, RIPEM 3.0b4
Sun Ultra-1                  2.1    #07: SunOS 5.5.1, Sun sc 4.0, RIPEM 2.1
Sun UltraSPARC 1 170         2.4    #01: Solaris 2.5.1, GCC 2.7.0, RIPEM 3.0b4
Sun ELC (33MHz SPARC)       13.8    #05: Solaris 2.5.1, gcc 2.7.2.1. R. 3.0b4

Notes on specific results, including the person who ran them:

#01: Matt Curtin, 1997-12-02
#02: Randall S. Winchester, 1997-12-04
#03: Mark Riordan, 1997-12-07, elapsed time from NTIMER.EXE.
#04: Michael Gruber, 1997-10-08
#05: Mark Riordan, 1997-12-07
#06: Jeff Peterson, 1997-03-26.  Not sure if processor 200 or 250 MHz.
#07: Doug Royer, 1996-05-03.  This is Solaris 2.5.x.  Not sure of CPU MHz.
