Gabriel said:
Hi I now found some (better?) ram that is the same as mine on ebay but I
have a few questions about it. Here is the ebay site:
http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=300576735839&ssPageName=STRK:MEWAX:IT
and here is my ram type but I have two 1gig sticks:
http://pcwizkidstechtalk.com/index.php/kingston-hyperx-review.html
Can I use the two 2gig sticks in channel (zero) and the two 1gig sticks in
channel (one) and still have dual channel support? As they are the same make
and type but not the same size.
Thanks
GK.
First consideration.
What OS is being used ? Is it a 32 bit OS or a 64 bit OS ?
The memory license on a Windows x32 OS is for 4GB.
If you purchase a 2x2GB kit, then that will allow
usage to the limits.
Since you have a 512MB video card, you'd expect the
BIOS to allocate 256MB of memory map for PCI bus and
at least 512MB for PCI Express bus. So you're getting
pretty close to "3GB free" in Windows. You won't be
gaining exactly a whole 2GB by making this change.
You're perilously close to a situation (with a 32 bit
OS), where a 2x512MB kit would suffice for your upgrade.
If you had a 64 bit OS, you could install all four
sticks, and have, say, 6GB total.
Next, consider the specifications.
Kingston HyperX memory - 4 GB ( 2 x 2 GB ) - DIMM 240-pin - DDR2
Storage Capacity 4 GB ( 2 x 2 GB )
Technology DDR2 SDRAM
Form Factor DIMM 240-pin
Memory Speed 1066 MHz ( PC2-8500 ) <---
Data Integrity Check Non-ECC
Latency Timings CL5 ( 5-5-5-15 ) <---
Features Heat sink , unbuffered
Supply Voltage 2.2 V <---
Lead Plating Gold
Limited lifetime warranty ( Germany, Austria and France - 10 years )
The important parameters are marked with arrows.
Speed and Latency, taken as a group, can either be "industry standard"
and nothing special, or the latency can be lower than normal. So
first, you go through a web site with a large number of
product offerings, and see whether DDR2-1066 CAS5 is
standard latency for that speed, or a low latency.
Next, you examine the voltage. JEDEC would list the
voltage requirement as 1.8V. Applying 2.2V is overvolting.
Too much overvolting, in the name of making enthusiast grade
RAM, could potentially shorten the life of the product,
or make it run hot. This is one reason, that some lunatics
use fan cooler assemblies that bolt to their RAM sticks -
it's because the voltage is boosted so much, the RAM
runs hot.
So what you want in a memory product, is a good clock
speed, an industry standard (or lower) CAS, and voltages
you can live with.
So if we were comparing two equal "speed and latency"
memories, and one took 2.2V and the other 2.1V, that
means the 2.1V didn't need as much voltage to make
that tested speed point.
Some motherboards (like perhaps server motherboards),
lack good VDimm adjustment capability, so purchasing
a 2.2V product would be dangerous. The reason would
be, the motherboard would provide 1.8V or 1.9V perhaps,
while the RAM fully meets timing at 2.2V. In such a
case, you'd have to dial down the speed, dial up the
CAS (more slack setting) to make the RAM work. You
may end up doing experiments like that, anyway.
So when you test the RAM, if it throws errors at
the stated speed, then the first question the
tech support will ask is "what VDimm setting did
you use?". They ask that question, to make sure you're
blaming their product, for valid reasons. You as
the customer, only use as much voltage as is needed
to make the RAM error free. If it still isn't error
free at 2.2V, then it's time to return it.
Once you've identified a candidate product, then
you also check what percentage of users experienced
DOA products, or how many people had RAM failures
after six months. You can use the reviews on Newegg
for that.
When I read the reviews here for example, there
are complaints this really isn't DDR2-1066 CAS5 RAM.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820104098
Now, if I go back and start my search again, this time
looking for 2x1GB kit, I can find these.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231144
G.SKILL
Model F2-8500CL5D-2GBPK
Capacity 2GB (2 x 1GB)
Speed DDR2 1066 (PC2 8500)
Timing 5-5-5-15
Voltage 2.0V - 2.1V <-- I take this to be 2.1V
Those are more or less the same specs as some
of their other kits, but the overall review
ratings are better.
When I go through the reviews in detail though, it looks
like the most recently shipping stuff, isn't as good as
when it started shipping.
This makes selecting a product really complicated. DDR2 isn't
mainstream, and perhaps finding good raw materials is more
difficult. Or perhaps they're cleaning out the cupboard.
Or maybe they don't put the time into test that they
used to. There can be any number of reasons why a
product would change.
There have been cases in the past, where the brand of memory
chips, changes in mid production. And this radically alters
the "goodness" factor. Perhaps people were finding the
product would operate one notch lower in latency, and it
got rave reviews with the original chips. And then later,
when an inferior chip was used, there was no test margin
to speak of, and then getting them to meet timing is
touch and go.
They don't change the SKU when that happens. For ordinary
RAM, industry standard speed/latency, I don't really care
if they change chips. But if I'm paying a price premium
for selected chips, I expect to see some benefit from it.
Namely, working at the spec'ed timing when I install it.
Not having to fight with it, to make it work.
If you were to buy a 2x1GB kit, and use them with your
existing 2x1GB kit, you wouldn't want the
characteristics of the RAM to vary too much. If they
were all DDR2-1066, but one was CAS5 and the other was
CAS6, you'll end up running at CAS6. If one was
DDR2-800 and the other kit was DDR2-1066, then
the BIOS will select DDR2-800 for the whole
bunch.
Mixing radically different voltages might not be
a good idea. For example, I wouldn't mix a 1.8V kit
with a 2.2V kit. Maybe the 1.8V kit will run too hot
at 2.2V. Maybe it'll simply blow out.
If you buy a 2x2GB kit, then the mixing issue is no
longer a problem (since you'd use just the 4GB total
with your x32 OS), but you still need to exercise
due diligence on the quality side of things.
It's what the reviews say that counts. And don't just
use the summary numbers ("77% gives 5 stars"), because
I've seen a review section before populated by inexperienced
users, carping about the wrong things. You have to
actually go through the reviews, and weed out the
ones that have reached a mistaken conclusion.
It'll take you a good two or three hours of reading,
to find the best sticks.
Paul
