4.2BSD on uVaxII ??

[Dan Ts'o]

In article <> root at bu-cs.UUCP (Barry Shein) writes:
>I guess what I am more or less confused about is, why are people buying
>microvaxes? I can think of only a few reasons:
>
>	1. You run VMS, so you got no choices (you fool.)
>	2. DEC is giving them to you for free or nearly (good reason.)
>	3. You have significant applications that require
>	the architecture specifically (macro code, q-bus periphs)
>	4. You have great expectations for the box.
>
>Seems from my analysis they are slower and more expensive than similar
>68K boxes and promise to remain so for years, especially I/O (rumour has
>it that some of the 68020s will be approaching the compute speed of the
>8600 for about 2% (20K/1M) the price this fall.) Other 32-bit processors
>(3B2 etc) have a similar outlook.

	But MicroVaxen aren't really that uncompetitive. The MicroVAX I
is slow, but so are many 68000 boxes. Using a series of Unix system benchmarks,
the MicroVAX I running Ultrix-32m was neck and neck with the Callan. The
Callan was faster in incrementing an int, but the MicroVAX I was faster at
nroff, and thats what I care about. MicroVAX I and the Callan are similarly
priced but with the MicroVAX you get VAX binary compatibility, 4.2BSD with
Ethernet capability, and a wide range of Qbus peripherals to chose from.
AT&T's 3B2 is in the same ballpark and it just as slow or slower.
	More expensive 68000 boxes like the Masscomp are faster but much more
expensive. Meanwhile the MicroVAX II has almost 11/780 performance at under
$20000. Its been said before: there is a major difference between chip
performance and system performance. The 68000 chip is quite a performer, but
I don't believe that your average $10000 68000 box delivers half that
performance. You always lose in disk, memory management, floating point and
communications.