[fa.info-vax] Unix on a 785

info-vax@ucbvax.ARPA (11/30/84)

From: Ronald A. Jarrell <JARRELLRA@VPIVAX3.BITNET>


Our CS department has Ultrix running on a 785.  The kernal configuration
doesn't include a choice for 785 as a CPU, and all the software was compled
on the basis of it being a 780.  Causes no problems, and there's no
difference except that everything runs twice as fast as the manual says it
will.

        -Ron Jarrell
        Virginia Tech Systems Programming

info-vax@ucbvax.ARPA (12/04/84)

From: ihnp4!ho95c!wcs@BERKELEY

The standard line is that the 785 is "bug for bug compatible" with the 780;
there should be no differences in operation unless you're into real-time
 DEC tells me the performance increase is about 35%; I didn't
think it was worth $75K bucks for that small an increase, so I made my
users speed up their software by 300% instead.
			Bill Stewart at AT&T Bell Labs, Holmdel NJ.

info-vax@ucbvax.ARPA (12/04/84)

From: PAD Powell <padpowell%wateng%waterloo.csnet@csnet-relay.ARPA>

	The standard line is that the 785 is "bug for bug compatible" with the 780;
	there should be no differences in operation unless you're into real-time
	 DEC tells me the performance increase is about 35%; I didn't
	think it was worth $75K bucks for that small an increase, so I made my
	users speed up their software by 300% instead.
				Bill Stewart at AT&T Bell Labs, Holmdel NJ.


Gee, did you use whips or just gentle threats of being fired?

Patrick ("Hmm, if the ding-dongs at Bell can do it, why can't we?")
	Powell

info-vax@ucbvax.ARPA (12/04/84)

From: ihnp4!houxm!hound!royw@BERKELEY

How did you do this marvelous feat? %300 increase in speed?

kaiser@jaws.DEC (Pete Kaiser 225-5441 HLO2-1/N10) (12/06/84)

I also speeded up all user software at one installation.  The method is
quite simple, and I recommend it to any installation whose CPU and disk
resources are nearing exhaustion:

	From every user program remove every fourth bit, and
	compact the result.

The speedup (as measured in CPU seconds used by these programs, and the
time from initiation to completion of each run) was in many cases far greater
than 300%.

---Pete