mark@mips.UUCP (Mark G. Johnson) (09/15/87)
We finally located a VMS-VAX 11/780 with SPICE available, so we ran
the three public domain SPICE benchmark circuits. Hopefully this
will let the VAX show off its FP performance, as the VMS-fortran
compiler generates better code than the Unix f77 compiler does.
A companion article includes the input circuit files ("decks").
========================== TABLE 1. DIGSR circuit ==========================
Berkeley-2G6
MACHINE seconds Vaxes OS, compiler, notes
------- ------ ----- -------------------------
VAX 11/780 1354.0 0.60 4.3BSD, BerkF77 V2.0
Microvax-II 993.5 0.81 Ultrix 1.1, fortrel
SUN 3/160 901.9 0.90 SunOS 3.2 f77 -O -f68881
VAX 11/780 848.0 0.95 4.3BSD, fortrel
VAX 11/780 808.1 1.0 VMS 4.4, fortran/opt
SUN 3/260 744.8 1.1 SunOS 3.2 f77 -O -f68881
SUN 3/160 506.5 1.6 SunOS 3.2 f77 -O -ffpa
SUN 3/260 361.2 2.2 SunOS 3.2 f77 -O -ffpa
SUN 4/260 225.9 3.6 SunOS 4-beta2 f77 -O3 -Qoption as -Ff0
MIPS M/800 136.5 5.9 UMIPS-BSD V2.01, f77 V1.21
AMDAHL 470 V/7 125.5 6.4 VMSP-CMS 4.0, FORTVS 4.1
MIPS M/1000 114.3 7.1 UMIPS-BSD V2.01, f77 V1.21
FPS 20/64 48.0 16.8 VSPICE (2g6 derivative)
========================= TABLE 2. BIPOLE circuit ==========================
Berkeley-2G6
MACHINE seconds Vaxes OS, compiler, notes
------- ------ ----- -------------------------
VAX 11/780 439.6 0.68 4.3BSD, BerkF77 V2.0
Microvax-II 394.3 0.76 Ultrix 1.1, fortrel
VAX 11/780 312.6 0.96 4.3BSD, fortrel
VAX 11/780 299.1 1.0 VMS 4.4, fortran/opt
SUN 3/160 285.1 1.0 SunOS 3.2 f77 -O -f68881
SUN 3/260 221.7 1.3 SunOS 3.2 f77 -O -f68881
SUN 3/160 170.0 1.8 SunOS 3.2 f77 -O -ffpa
SUN 3/260 112.0 2.7 SunOS 3.2 f77 -O -ffpa
SUN 4/260 63.7 4.7 SunOS 4-beta2 f77 -O3 -Qoption as -Ff0
MIPS M/800 42.6 7.0 UMIPS-BSD V2.01, f77 V1.21
AMDAHL 470 V/7 39.5 7.6 VMSP-CMS 4.0, FORTVS 4.1
MIPS M/1000 35.4 8.4 UMIPS-BSD V2.01, f77 V1.21
FPS 20/64 12.5 23.9 VSPICE (2g6 derivative)
========================= TABLE 3. TORONTO circuit =========================
Berkeley-2G6
MACHINE seconds Vaxes OS, compiler, notes
------- ------ ----- -------------------------
VAX 11/780 460.3 0.63 4.3BSD, BerkF77 V2.0
Microvax-II 366.9 0.80 Ultrix 1.1, fortrel
SUN 3/160 328.6 0.89 SunOS 3.2 f77 -O -f68881
VAX 11/780 302.9 0.96 4.3BSD, fortrel
VAX 11/780 291.7 1.0 VMS 4.4, fortran/opt
SUN 3/260 266.0 1.1 SunOS 3.2 f77 -O -f68881
SUN 3/160 189.1 1.5 SunOS 3.2 f77 -O -ffpa
SUN 3/260 129.4 2.3 SunOS 3.2 f77 -O -ffpa
SUN 4/260 73.4 4.0 SunOS 4-beta2 f77 -O3 -Qoption as -Ff0
MIPS M/800 41.4 7.0 UMIPS-BSD V2.01, f77 V1.21
AMDAHL 470 V/7 39.3 7.4 VMSP-CMS 4.0, FORTVS 4.1
MIPS M/1000 34.5 8.5 UMIPS-BSD V2.01, f77 V1.21
FPS 20/64 17.5 16.7 VSPICE (2g6 derivative)
############################################################################
Since the first posting, we discovered that we were not running the
standard Berkeley Spice2g6 UNIX distribution on our 780. The standard
distribution includes code for block copy and zero that uses VAX
string instructions. Our copy had that replaced with simple byte copy
loops so as to be portable to our own machines. We went back to the
VAX string instruction version for VAX times so that the results will
correlate with results obtained elsewhere with the standard
distribution. And to be fair, we used the 4.2 bcopy/bzero routines on
other machines that we ran it on (SUNs, MIPSs). The end result: the
780 numbers have improved by 15-20% since the last posting, the MIPS
numbers by 7%. (The 1.21 compiler and 2.1 OS releases also improved
MIPS results.)
We have also added 780 results for spice2g6 compiled with two more
compilers: the fortrel compiler from Lawrence Livermore Labs and the
DEC VMS-fortran compiler. The relative performance column ("Vaxes")
is now normalized to the DEC VMS-fortran compiler.
We would like to thank Aedan Coffey, Joerg Bach, and David Hough for
their contributions to this list.
--
-Mark Johnson *** DISCLAIMER: The opinions above are personal. ***
UUCP: {decvax,ucbvax,ihnp4}!decwrl!mips!mark TEL: 408-720-1700 x208
US mail: MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086