🚀 go-pugleaf

In <438@geowhiz.UUCP> larry@geowhiz.UUCP (Larry McVoy) writes:
> I have a few comments & questions.
>
> 1) The story I have heard (albeit from sources that I don't view as gospel)
>    is that the 32016 (in the symmetrics box) is s*l*o*w, about 1/4 a 68010.
>    Is that true?

You would be well advised to put even less faith in your "sources", for
even though I cannot answer for the Symmetrics box, the Dhrystone results
speak (with the usual truck load of salt) for themselves:

|*----------------DHRYSTONE VERSION 1.0 RESULTS BEGIN--------------------------
|*
|* MACHINE	MICROPROCESSOR	OPERATING	COMPILER	DHRYSTONES/SEC.
|* TYPE				SYSTEM				NO REG	REGS
|* --------------------------	------------	-----------	---------------
|* Colex DM-6	68010-8Mhz	Unisoft SYSV	cc		 378	 410
|* Apollo DN550	68010-?Mhz	AegisSR9/IX	cc 3.12		 666	 666
|* Cadmus 9000	68010-10Mhz	UNIX		cc		 714	 735
|* Cadmus 9790	68010-10Mhz 1MB	SVR0,Cadmus3.7	cc		 720	 747
|* Burroughs XE550 68010-10Mhz	Centix 2.10	cc		 769	 769 CT1
|* ATT 3B2/300	WE32000-?Mhz	UNIX 5.0.2	cc		 735	 806
|* Apollo DN320	68010-?Mhz	AegisSR9/IX	cc 3.12		 806	 806
|* IRIS-2400	68010-10Mhz	UNIX System V	cc		 772	 829
|* IRIS-1400	68010-10Mhz	UNIX System V	cc		 909	1000
|* NSC ICM-3216 NSC 32016-10Mhz	UNIX SVR2	cc		1041	1084
|		-------------
|* ATT PC7300	68010-10Mhz	UNIX 5.0.2	cc		1041	1111
|* Sun2/120	68010-10Mhz	Sun 4.2BSD	cc		1136	1219
|* MASSCOMP 500	68010-10MHz	RTU V3.0	cc (V3.2)	1156	1238
|* Cyb DataMate	68010-12.5Mhz	Uniplus 5.0	Unisoft cc	1162	1250
|* Sun2/120	68010-10Mhz	Standalone	cc		1219	1315
|* Sequent Balance 8000	NS32032-10MHz	Dynix 2.0	cc	1250	1315 N12
|		-------------
|* IBM PC/DSI-32 32032-10Mhz	MSDOS 3.1	GreenHills 2.14	1282	1315 C3
|		-------------
|* Cyb DataMate	68010-12.5Mhz	Uniplus 5.0	Unisoft cc	1470	1562 S

> 2) Another story I hear a lot is that people *claim* that the 32xxx chips are
>    fast, but independant tests show that the claims are consistantly false.
>    Can anyone comment on this?  I'd prefer to hear from people who:
>    A) are not associated with National Semiconductor 
>    B) are using a unix workstation based on a 32xxx chip set.

See Dhrystones above.  People with access to 32000 based
workstations from Tektronix, Intergraph & Whitechapel, etc.,
would be ideal candidates for expanding on this questions.
For floating point fans, it's probably a save bet that the
32k based systems outperform most existing (ie. excluding
68020/68881) 68k systems.  Some concrete figures on this
would be of interest.

> 3) I've heard that the 32xxx series of CPU's are binary compatible.  If this
>    is true, are they pin compatible?  The real question is: can I buy a 
>    symmetrics box, pull the 32016 and replace it with a 32032 or 32332?

I'm not sure what you mean by "pin compatible".  If you mean
signal and electrical compatibility, the answer is basically yes.

There would however, be absolutely no point in replacing the
32016 with a 32032, as they are identical save for the CPU bus
interface.  You would need a modest amount of logic to make a
16 bit data bus look like a 32 bit bus (with possible degradation
in performance).  The 32332 is a different story.  Since it has
dynamic bus sizing, it could indeed easily replace a 32016,
with an increase in performance due to the deeper prefetch queue
and separate ALU/Shifter in the address path.

A small adapter board containing the 32332, could be plugged into
the 32016 socket, although you might need to bring an auxiliary
power source onto the board, to satisfy the 32332s greater power
requirements.

P.S. You can find some benchmark results for 32032 IBM-PC add-on board
in a recent issue of UNIX/WORLD.

			Bjorn R. Bjornsson
			Department of Computing Science
			University of Alberta