KenK said:
I haven't located them yet but from what I have for a manual the MB has
two memory slots and will accept up to 2 G. I already have 0.5 Gb - I
rashly assume in a single slot. Can I add 1G to the empty slot to result
in 1.5G RAM? Any idea?
TIA
You have lots of growth and fallback plans. This is a sampling.
1) Current system 512MB, one DIMM.
2) Buy the 1GB DIMM, try 1GB+512MB two slot config.
Swapping the DIMMs in the slots, should make little difference,
or if there is a difference, it would be too hard to predict the
responsible pattern (stub capacitance).
3) If memtest86+ reports problems, drop to 1GB configuration,
and store the 512MB DIMM in an antistatic bag. Place the
1GB DIMM furthest from the processor. (Keep the 512MB as a spare.
If the 1GB fails and needs to be returned, you'll have the 512MB
to use in its place.)
4) If testing reveals 1GB is not enough, buy the second 1GB DIMM
for a total of 2GB.
The upgrade costs 1X, if (2) or (3) satisfy the requirements.
And costs 2X, if we needed to move to step (4).
*******
When adding or removing DIMMs, all power should be removed from the
computer. The simplest way to do this, is unplug the computer before
you start. This is to avoid standby power flowing in the DIMM slots,
coming in contact with the wrong pin, while the DIMM is being
inserted. On Asus motherboards, a green LED tells you when the
power is drained - on other motherboards, wait 30 seconds to a minute
after you unplug, before beginning work.
In terms of ESD practices, bring your body to the same electrical
potential as the chassis. A grounding strap with internal 1 megohm
resistance is preferred, as it dissipates charge a bit more slowly.
The alligator clip fastens to a shiny I/O screw on the back of
the computer.
(This is now twice the price it used to be! Shop around.)
http://www.radioshack.com/product/index.jsp?productId=2260808
*******
And this always happens, so I have to mention it.
I selected Crucial for a reason. The reason is DIMM composition.
The correct composition for your motherboard, is low density
RAM. 16 chips, each chip having x8 width. Memory chips, with
a common package size, have room for 4 bit, 8 bit, or 16 bit
bus connections. The typical "sweet spot", the one Intel likes,
is x8. If you go x16, it still works, but the max DIMM density
is cut in half. Some older SODIMMs were made that way. If you
use x4, that places too many loads on any per-rank control signals.
The x4 formulation is approved by JEDEC, but specifically denied
by Intel datasheets. And the funny thing is, if you shop for
x4-chip-based DIMMs, they *never* have popular brand names on them.
The companies which make UDIMMs with x4 chips, wish to remain
nameless. They don't want to be associated with what they made.
Which I consider to be pretty strange as a business practice.
You won't find Corsair or Crucial on these - they wouldn't
be caught dead selling that stuff.
Note that, x4 chips exist for a reason, and are perfectly valid
on server DIMMs. When used on a server, "register" chips function
as a buffer between the x4 memory chips, and the system memory
controller. Such RDIMMs even support more ranks than a desktop
product does. So there is a place for the x4 chips, but those
chips should just stay on RDIMMs. We don't need them complicating
matters for desktop users. And the (nameless) manufacturers of
the "wrong" DIMM design, see things differently. A buck is a buck,
even if the customer ends up unhappy.
OK, so where might this be a problem ? Ebay. There are Ebay sellers
who would gladly sell you a nameless x4 based 1GB DIMM, for $5 less
than Crucial sells it. The people I pass memory recommendations to,
they promptly ignore my advice, find the first piece of crap (complete
with warnings right on the web page!), buy it, and are then unhappy
it is throwing errors.
If you want to shop Ebay, by all means, be my guest. As long as you
know you're shopping for x8 width chips PC2700 UDIMMs, there shouldn't
be a problem. Just avoid the temptation to buy the x4 nameless ones.
Mushkin (a memory company), used to keep a web page with five or
six pages of test results. They actually tested x4-width-chip UDIMMs
on popular chipsets of the time. And they could find precisely one
chipset, with enough drive strength to drive 3x1GB configuration of
those things. So if you were really really lucky, you might actually
mix the one capable chipset, with the "bargain DIMMs"
The Mushkin
web URLs prevented archive.org from archiving the information,
and I dumbly forgot to archive all of it myself, assuming they'd
just leave it there. I would love to show people those
test results, as a warning.
HTH,
Paul
