Re: Dual Channel memory question - Attn: Paul
- From: nospam@xxxxxxxxxx (Paul)
- Date: Tue, 12 Jul 2005 22:40:39 -0400
In article <5nc8d1pp39v1sck26n23r4v7rm1mdbkofu@xxxxxxx>,
miller.90@xxxxxxxxxxxxxxx wrote:
>
> When Corsair came out with their TwinX modules, they did, in fact test
> them together to make sure they were matched and would work in
> dual-channel. They even published the make and model of the board on
> which they were tested. They used popular motherboards that were in
> common use. Their current policy as stated on their Web site: "TWINX
> modules are tested in pairs, in dual channel motherboards, at very
> aggressive latency settings for optimum performance and
> compatibility." So, as you say, there's nothing "magic", but, in
> fact, some of the better dual-channel pairs are not simply picked
> randomly from a barrel of DIMMs with similar chips.
>
> Ron
You know, when I read the info they place on some of these
web sites, I wonder about the people running those operations.
The majority of memory testing should be done at the individual
chip level. Imagine with a DIMM, that has 16 chips on it,
what the odds would be of a single bad memory chip making
the whole module bad. The only way to get a decent yield on
the DIMMs, is to test the memory chips first, sort them
for speed, and then assemble the module. Very little test
time should be spent after final assembly, because at that
point, the most likely damage, if any, will be I/O pads on
chips. A cursory test (a few seconds in a hand held memory
tester) should be enough to determine that the whole module
works - since all the memory bits were tested at the chip
level, that kind of testing doesn't have to be repeated.
One thing you will notice about the vast majority of retail
products offered for sale, is attention to cosmetic details.
Workers wear gloves, so that they won't leave finger prints
all over the place etc. Now, to take a brand new DIMM, and
scratch up the gold contacts by shoving the DIMM into a
motherboard, is anathema to that kind of care and attention.
If you are going to test a DIMM, you would want a ZIF socket
of some sort. A DIMM socket that doesn't apply force until
the DIMM is seated.
So, for me, I don't buy that story about using motherboards
for test. Another issue - a DIMM socket on a motherboard
would wear out after about 500 cycles or less - there must
be a steady stream of motherboards going to the repair station,
to have the DIMM sockets replaced.
I'm just a bit sceptical of the whole notion. It sounds like
the kind of operation I would run out of the basement of
my house.
I bet Micron (Crucial) or Samsung don't use motherboards
for this.
Paul
.
- References:
- Dual Channel memory question
- From: ChowBoy
- Re: Dual Channel memory question
- From: Paul
- Re: Dual Channel memory question - Attn: Paul
- From: ChowBoy
- Re: Dual Channel memory question - Attn: Paul
- From: Paul
- Re: Dual Channel memory question - Attn: Paul
- From: Dragoncarer
- Re: Dual Channel memory question - Attn: Paul
- From: Paul
- Re: Dual Channel memory question - Attn: Paul
- From: milleron
- Dual Channel memory question
- Prev by Date: Re: P4C800-E 1394 Capture Problems
- Next by Date: Re: P2B - HD Question
- Previous by thread: Re: Dual Channel memory question - Attn: Paul
- Next by thread: Re: Dual Channel memory question - Attn: Paul
- Index(es):
Relevant Pages
|