Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!ois.db.toronto.edu!jonah
From: jonah@db.toronto.edu (Jeffrey Lee)
Newsgroups: can.sun-stroke
Subject: benchmarks
Message-ID: <89Aug16.165710edt.9341@ois.db.toronto.edu>
Date: 16 Aug 89 20:57:00 GMT
Sender: list-admin@cs.toronto.edu
Distribution: can
Lines: 15
Approved: can.sun-stroke@mail.ai.toronto.edu
Original-To: sun-stroke@cs.toronto.edu, xspeed@csri.toronto.edu

I was interested in determining whether it would be better to use
integer or floating point for faster arc drawing code under X11.  This
lead to me write a simple benchmark test program.  The following are
approximate timings (in microseconds) for operations on various
processors:

        3/50    3/60    3/80    4/110   4/280   SGI/4D  M120
                              (no FPU)
loop    1.25    0.93    1.17    1.03    0.82    0.20    0.18
iadd    1.15    0.85    0.52    0.82    0.63    0.08    0.12
imul    4.27    3.12    2.95    5.80    5.00    0.49    0.43
idiv    7.42    5.40    5.43    8.17    6.67    1.45    1.31
dadd   15.73   11.10    8.42  493.00    1.62    0.25    0.27
dmul   17.43   12.25    9.15  540.00    1.82    0.34    0.30
ddiv   18.56   13.85   11.12  827.25    5.40    0.89    0.79