> 
> Bill Studenmund wrote:
> > 
> > > 
> > > On Fri, 14 Nov 1997, Bill Studenmund wrote:
> > > 
> > > > In quadruplets. You need to populate each bank of 4 with the same
> > > > size chip. 
> > > 
> > > Same size _and_ speed - don't mix 70ns chips with slower ones.
> > 
> > Is it really a problem to have different speeds, as long as they are
> > all fast-enough? I didn't think these macs could change the memory
> > timing (so the speed couldn't get set faster than the slower chips)
> > and they are on different memory banks. ??
> 
> I've successfully mixed 80ns with 70ns in the same bank.  I believe it's
> fine as long as you make sure that they are all fast enough.  It's a bit
> of a waste, tho, since you won't see that 10ns speedup if you do mix them.

What speedup? True 70 ns chips are faster than 80 ns, but I was trying
to get at the fact that the Mac doesn't change its memory timing (at least
in these computers), so the CPU will take as long to access 70 ns memory as
80 ns memory. So a full bank of 70 ns will be accessed as fast as a full
bank of 80 ns RAM, which is as fast as a mixed bank.

I agree with Paul about what happens if one of the SIMMs is NOT
fast enough for the machine. :-)

Also, since we started talking about IIsi's, don't forget that that bank
of memory will be accessed as fast as a 100 ns chip! :-)

Take care,

Bill