david@elroy.Jpl.Nasa.Gov (David Robinson) (11/16/87)
Since MIPS is such a hot topic and everyone likes magic meaningless number, does anyone have MIPS ratings for older machines (pre-1980). For example: IBM 370/168 and rest of 370 line IBM 360 line CDC 6000 series old UNIVACS ENIAC? PDP-[1-8] Any other "popular" machines -- David Robinson elroy!david@csvax.caltech.edu ARPA david@elroy.jpl.nasa.gov ames!elroy!david UUCP Disclaimer: No one listens to me anyway!
sewilco@datapg.UUCP (11/17/87)
In article <4839@elroy.Jpl.Nasa.Gov> david@elroy.Jpl.Nasa.Gov writes: >Since MIPS is such a hot topic and everyone likes magic meaningless >number, does anyone have MIPS ratings for older machines (pre-1980). In a Datamation article on used machines, from a computer leasing firm: MODEL Introduced MIPS 7070 9/02/58 .022 1401 10/05/59 .0074 1410 10/11/62 .0154 360/30 04/07/64 .037 360/50 04/07/64 .178 360/65 04/22/65 .680 360/85 01/30/68 2.4 370/155 06/30/70 .670 370/165 06/30/70 1.89 370/145 09/23/70 .32 370/158 08/02/72 .870 370/168 08/02/72 2.3 3033 03/25/77 4.7 3031 06/10/77 1.14 4341-1 01/30/79 .77 I've gotten similar MIPS numbers for these machines from other sources, except IBM. The S/36 (5360, introduced 5/83) is about 0.2 MIPS. I don't remember the introduction dates, but Apple II is 0.15 and TRS80-III is 0.22. Multiple processors and intelligent controllers particularly affect the importance of MIPS. The S/36, for example, has two processors and also has I/O interface processors. One S/36 processor (CSP) seems to be used as an MMU and arithmetic unit. -- Scot E. Wilcoxon sewilco@DataPg.MN.ORG {ems,meccts}!datapg!sewilco Data Progress Minneapolis, MN, USA +1 612-825-2607 "My name is David Small, what makes you think you're David Letterman?"
jejones@mcrware.UUCP (James Jones) (11/19/87)
------------------ Just to add some more numbers...an article in *System 68* looking at various programs written in both assembly and high-level languages indicated that a 2MHz 6809 ran about .4 MIPS. Cheers, James Jones
aeusesef@csun.UUCP (sean fagan) (11/23/87)
[I would e-mail, but I seem to have lost the address]
In a previous article, somebody asked for MIPS ratings for old machines,
CDC machines included.
Based on personal experience, technical reference, and some extrapolation,
a CDC 170/174 gets (yes, we still have one) arount 1.3 MIPS (also the same
MFLOPS); add a second head and it gets around 2.1.
A 730 gets slightly lower than that number, around 1.1 and 1.8, whereas a
750 gets around 4.5 and (guessing here; don't actually have one to base
numbers on) 8.8, and a 760 (same as a 750, only faster RAM) gets around 6.0
and 11.7 (ditto on experience).
A 180/830 gets around 1.1 (like a 174) and 1.8.
Hope this has proven interresting to somebody.
BTW, although the 830 gets around the same MIPS as a VAX or two, personal
experience has proven that the 830 is quicker, due to a) incredibly quick
context switching times (a couple of microseconds), and b) a very efficient
I/O system.
Share and Enjoy.
-----
Sean Eric Fagan Office of Computing/Communications Resources
(213) 852 5742 Suite 2600
1GTLSEF@CALSTATE.BITNET 5670 Wilshire Boulevard
Los Angeles, CA 90036
{litvax, rdlvax, psivax, hplabs, ihnp4}!csun!aeusesefroot@cca.ucsf.edu.UUCP (11/26/87)
In article <919@csun.UUCP>, aeusesef@csun.UUCP (sean fagan) writes: > [I would e-mail, but I seem to have lost the address] > In a previous article, somebody asked for MIPS ratings for old machines, > CDC machines included. [Details on Cyber 170/X, 180/X, 7x0 machines reported.] > Some tests we made on the even older CDC 6400 indicated 1 MIP +or- 10% for a number of programs. The 6600 was about 3.5 times faster. Of course, CDC 6xxx MIPS ratings are unlike those of many other architectures. The machines seemed faster because the OS in use at that time emphasized low overhead compared to competing systems and peripherals did not have to interrupt the main processor for service. Indeed, the early OS's for those machine tried to do too much in the PPU's (Peripheral Processing Units) and this slowed them. Moving selected functions back to the CPU improved performance a lot. Thos Sumner (thos@cca.ucsf.edu) BITNET: thos@ucsfcca (The I.G.) (...ucbvax!ucsfcgl!cca.ucsf!thos) If he says it's "user friendly" watch out; he's a con artist. #include <disclaimer.std>