Re: STREAM Bandwidth results

From: Michael Kukat (
Date: Wed Nov 13 2002 - 15:13:29 CST

The last one for today...

Vendor: Dolphin Server Systems
System: Triton/88k, m88100, 32 MB RAM
OS : Dolphin Unix System V Release 3.2
Build : cc -o /tmp/stream -s -O stream_d.c second_wall.c -lm

Results stable, even after code optimizations and checking of the assembler
output. Now, this machine really doesn't throw out more. I expected much more
due to 360 RAM chips needed in parallel. Looked like a very wide bus.

Very rare machine, i heard only 250 were made of them. I have serial numbers
124 and 125 :)

(McCalpin's Note: This machine was first available in 1989 or 1990 -- I
have seen performance results from a December 1990 report. Frequencies
for the Dolphin product were reported to be either 17 MHz or 20 MHz --
I chose 20 MHz for the STREAM tables.)

This system uses 8 bytes per DOUBLE PRECISION word.
Array size = 1000000, Offset = 0
Total memory required = 22.9 MB.
Each test is run 10 times, but only
the *best* time for each is used.
Your clock granularity/precision appears to be 9999 microseconds.
Each test below will take on the order of 1290000 microseconds.
   (= 129 clock ticks)
Increase the size of the arrays if this shows that
you are not getting at least 20 clock ticks per test.
WARNING -- The above is only a rough guideline.
For best results, please be sure you know the
precision of your system timer.
Function Rate (MB/s) RMS time Min time Max time
Copy: 13.4454 1.2080 1.1900 1.2200
Scale: 9.6970 1.6610 1.6500 1.6700
Add: 12.4352 1.9340 1.9300 1.9400
Triad: 10.3896 2.3220 2.3100 2.3400


--     Home Powered by: (Net|Open|Free)BSD IRIX NonStop-UX
Solaris AIX HP-UX Tru64 MUNIX Ultrix VMS SINIX Dolphin_Unix OpenStep MacOS A/UX

This archive was generated by hypermail 2.1.4 : Wed Jan 08 2003 - 09:17:56 CST