Differing clock speeds on Sparcs?

[wilson.h.bent..jr]

In article <2623 at brazos.Rice.edu> brsmith at cs.umn.edu (Brian R. Smith) writes:
>We have 12 SparcStation 1's with nearly identical configurations.  BUT, as
>was pointed out by one of our users, 3 of them are slower than the rest.
>Here's an example [ highly edited ]
>atto:	        2.6 real         2.4 user         0.1 sys  
>deca:	        4.0 real         3.8 user (!)     0.1 sys  

We had a similar problem; it turned out that the slower machines had
earlier revisions of CPU boards.  No, I don't remember the rev. numbers,
but if it helps, I've got the machine's serial numbers:

	Slow:	925F3928 926F0832
	Fast:	926F0854 926F0893

This is based on experience; we complained about our slow machines, Sun
sent us newer-rev CPU boards (which we installed - easy!), and they turned
into fast machines.

Side note: from what I understand, serial numbers can be decoded thusly:

	925F3928
	^^ ^^----- "real" serial number (sequence number)
	|| +------ Manufacturing location?  Not sure.
	|+-------- Week-of-year (here, 25)
	+--------- Year minus 1980 (here, 1989)

So from our machines, it looks as though somewhere between #832 and #854
they started using a new, faster CPU board.  If you're still under
warranty or have hardware support, you're in luck.  Otherwise...

Wilson H. Bent, Jr.		... att!hoh-2!whb (whb at hoh-2.ATT.COM)
AT&T - Bell Laboratories	HOH L-274A
Holmdel, NJ 07733		(201) 888-7129