SUMMARY: SIMM placement in SPARCstation-2 ??

From: Ole Holm Nielsen (ohnielse@ltf.dth.dk)
Date: Mon Mar 11 1991 - 05:36:46 CST


My original question:

>I just received a SS-1 to SS-2 upgrade kit, and now I want to move my added
>(i.e., those beyond the base 8 MB) SIMM modules to the new SS-2 box.
>However, a neat little trick in the SS-2 Installation Guide is that the SIMM
>installation instructions, which previously (for SS-1s) were in an Appendix,
>have now been moved into a separate manual called "Installing SPARCstation 2
>SIMMs". Obviously, I don't own that manual, so I wonder:
>
>Would anyone offer me SS-2 SIMM installation maps (i.e., which SIMM sockets
>do I fill) ? I imagine that I still need to install groups of 4 identical
>SIMM modules. Do I fill the groups with the same symmetry as the factory-
>installed modules, e.g., should I add four modules in U0312-U0321-U0310-U0308 ?

Summary:
========
Some folks report working with my guessed configuration of placing extra
SIMMs behind the Sun-installed ones. Beware however that the SS-2 requires
80 nanosecond chips, as discussed previously on sun-managers. Thus Sun-
supplied SS-1/1+ SIMMs, which are slower than 80 ns, are not supported.
I am still missing the official Sun guidelines for SIMM placement, and
I ask any owner of the manual "Installing SPARCstation 2 SIMMs" to
send me the Correct placement rules.

A useful report was posted to alt.sys.sun by Jim Lick <jim@cave.sba.ca.us>:
! There were questions earlier about adding memory to a Sparc 2. I recently
! performed this operation and have some information to share...
!
! First, the arrangement of the banks have changed. In the Sparc 1 and 1+
! the banks were arranged as follows:
!
! Back Of Machine
! 4 ---------------- ---------------- 2
! 4 ---------------- ---------------- 2
! 4 ---------------- ---------------- 2
! 4 ---------------- ---------------- 2
! 3 ---------------- ---------------- 1
! 3 ---------------- ---------------- 1
! 3 ---------------- ---------------- 1
! 3 ---------------- ---------------- 1
! Front Of Machine
!
! In the Sparc 2 however, the arrangement is as follows:
!
! Back Of Machine
! 1 ---------------- ---------------- 1
! 2 ---------------- ---------------- 2
! 3 ---------------- ---------------- 3
! 4 ---------------- ---------------- 4
! 1 ---------------- ---------------- 1
! 2 ---------------- ---------------- 2
! 3 ---------------- ---------------- 3
! 4 ---------------- ---------------- 4
! Front Of Machine
!
! In each machine there are 4 banks of 4 simm slots for a total on 16
! simm slots. To add more memory in each machine you need to add a
! homegenouse block of 4 simms to the next available bank. In the
! Sparc 1 and Sparc 1+, the banks were organized such that banks were
! in blocks on the board. In the Sparc 2, the memory is organized
! so that each byte of the 32 bit word is in the same block on the
! board. I assume this was done to simplify the address line traces
! on the board.
!
! The other thing I found out was that 100 ns simms definetly don't
! work in the Sparc 2. I knew that even some 80 ns chips supposedly
! don't work. I didn't mean to try this either. What happened was
! that the memory arrived at a time when the Sparc 2 was working on
! some critical runs and couldn't be disturbed. So I threw the memory
! into our Sparc 1 in the mean time. When I got around to moving it
! over to the Sparc 2 where it was supposed to be, I managed to pull
! out one bank on 70ns simms and another of 100ns simms, which were
! what the Sparc 1 originally shipped with.
!
! I learned from this experience that the error codes given on the
! keyboard lights aren't always accurate. They told me that the
! problem was in the first bank of original Sun memory. I spent
! several minutes making sure that all simms were snug in their
! sockets before bothering to check the speeds. Once I got the
! 100ns chips out, and replaced with 70ns chips, the machine booted
! and ran fine with 24 megs of memory.

Also, someone learned from Parity the following bank installation order
(i.e., fill banks in the order 1-2-3-4):
  1-------------- ---------------1
  3-------------- ---------------3
  2------------- ---------------2
  4------------- ---------------4
  1-------------- ---------------1
  3-------------- ---------------3
  2------------- ---------------2
  4------------- ---------------4

Thanks to:
"Anthony A. Datri" <datri@lovecraft.convex.com>
jipping@frodo.cs.hope.edu (Mike Jipping)
dasun!murray@sunkist.West.Sun.COM (Mary Murray (Murray))
Claus Assmann <ca@idefix.informatik.uni-kiel.dbp.de>
Keith Pilotti <Pilotti@Jupiter.SAIC.COM>
steve%eastend@uu.psi.com (Steven Roth)



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:11 CDT