If
somebody asked me at this time last year if there is such a thing
as fast DDR2, I probably would have laughed in their face.
Not
so much anymore. Why the change now? As we've seen throughout
the year, speeds have been increasing and latencies have been
lowering.
Motherboard
chipsets have also become much more sophisticated, and as we've
seen with the nForce 4 SLI Intel Edition, there's a strong case
for DDR2 now. While DDR2 has been something of a hard sell for
many users, it has steadily, albeit slowly, gaining acceptance
among enthusiasts. With AMD slated to possibly offer DDR2 with
their next-gen platform, things are looking up for the oft maligned
ram format. It also helps that dual core processors can do more
to saturate the memory bus than their single cored brothers.
Today
we'll be checking out Corsair's
TWIN2X1024-8000UL kit.
These modules are based on the XMS2 line, and feature black coloured
heat spreaders.
Specifications
|
Part
Number
|
Speed
|
Size
|
Package
|
CAS
Latency
|
|
|
XMS2-8000UL
|
1GB
|
2x240
DIMM
|
5-4-4-9-T1
|
You can view the full specifications in this PDF,
but what you need to know is the modules are validated with multiple
Intel based motherboards at 1000MHz. Yes, those are three zeros
at the end.
The
two ram modules, each weighing in at 512MB, arrived in a hard
plastic case, with each ram module fitted in the clear plastic
shell. It's worth pointing this out is because there should be
no confusion in what you're getting. You have a clear view of
the ram, and the labels (indicating the type of modules), and
the packaging is snap sealed. The seal changes to a slightly translucent
white when opened, so it should be pretty obvious to the buyer
if the package has been tampered with.
As
previously mentioned, the modules have black coloured heat spreaders.
Along with improving the cooling of the ram, the spreaders (from
what I was also told by a marketing rep) will reduce EMI since
it shields the ram, but I don't have any quantitative proof of
that.
The
kit is rated at PC2-8000, which works out to 1000MHz, or a cool
1GHz. The ram timings are quite low, 5-4-4-9 at 2.2v, and as many
enthusiasts know, tighter timings normally result in better performance.
Still a far cry from the 2-2-2-5 we're used to seeing from lower
speed DDR, but as memory speeds increase, it is extremely difficult
to maintain stability with such low timings. Lately we've noticed
that the timings seem to be having a lesser effect on performance
as memory speeds increase.
All
of the TWIN2X kits are tested in pairs and in a Dual Channel environment,
which all of Intel's new chipsets (including offerings from VIA,
NVIDIA and ATI) support. Pictured above is our test environment
which includes an Intel Pentium Extreme Edition 840 and MSI P4N
Diamond nForce 4 SLI.
Overclocking
(or underclocking?)
A RAM
Stress Test Professional (R.S.T.) Pro2 card will be used to
test for reliability as well as stability during our overclocking
tests. The card is quite expensive, but if you're testing a lot
of ram, or you're a technician troubleshooting systems, this card
is well worth the money.
How it works is that the R.S.T. Pro2 goes into
a free PCI slot and after POST, it bypasses the OS completely,
loading a proprietary OS and software package (embedded on the
card itself). The card allows you to perform stress and performance
benchmarks without the Operating System's memory footprint, memory
resident programs, or drivers interfering with the memory installed.
Before continuing, first a disclaimer, your mileage
may vary. The MSI
P4N Diamond is a great overclocker, but according to the
BIOS, 900MHz was the ceiling of this board. Basically, there's
no way to even hit the maximum supported speed of this kit. We'll
see what we can do, but this is the information we had before
we even started.
By default, the ram was configured to 667MHz upon
installation. We had to set things to manual in the BIOS to change
these settings, and after doing so, we managed to hit 720 (despite
setting it to 725, it dropped to 720 once we hit Enter). At this
point, we began to experience problems. We raised the voltage
slightly and the problems went away.
Next stop was 800MHz. This was no problem provided
we boost the voltage to the Corsair supported max of 2.2v. We
continued upwards and got as high as 900MHz, the board's maximum.
For the heck of it, I went for 1000MHz, which the
board allowed us to enter. I kinda figured it just displayed
that, but was really running at 900MHz. We rebooted and were greeted
with no POST. We rebooted the BIOS several times to no avail.
Before firing off angry emails to Corsair and MSI,
I removed one of the sticks and rebooted. The system restarted
and POSTed. The BIOS reported the ram running at 667MHz (this
was expected since we reset it), so I figured we killed a stick.
To be sure, we swapped modules and behold, that one was working
as well. We boosted the ram to 800MHz, and no problems at all.
We then tried both sticks in the other ram slot
and the system would not boot. For kicks, we placed a ram module
into the working slot, reset the BIOS again, then placed the second
stick in the non-working slot and did another reset. Power was
turned on and the system booted up with both channels working.
To make a long story short, MSI's safety measure seems to be it
deactivates a channel during a bad OC and our totally nonscientific
method of avoiding a RMA return worked out.
Oh, lest we forget, at 2.2v, the maximum stable
overclock (passing the R.S.T. Pro2's tests) was 930MHz. We did
get as high as 940MHz, and the system did run, still passing in
the R.S.T. Pro2 test, but would lock up whenever we fired up a
benchmark.
Test
Setup
MSI
P4N Diamond nForce 4 SLI: Pentium 4 840 Extreme Edition, 2 x 512MB
Corsair TWIN2X-8000UL, 2 x MSI NX7800GTX, 160GB Seagate SATA 7200rpm,
Windows XP SP1.
The comparison memory will be Crucial's Ballistix 6400 (800MHz),
and Corsair's XMS2-5400UL (675MHz), both 1GB kits, running at
their default speeds and timings. The Corsair
TWIN2X-8000UL will be run at both 800MHz and 930MHz at
5-4-4-9.
SiSoft Sandra 2005 Memory
Just a reminder that the MSI's FSB increased past
200 once we passed 800MHz memory, so the overclocked (for the
motherboard) scores will not be "true" memory benchmarks
in some cases. That being said, we can see that the overclocked
settings run away from the rest, but at 800MHz, the slightly better
timings (despite having a higher CAS setting) on the Corsair keeps
it slightly ahead of the Crucial. These results are synthetic
though, so let's look at real-world scenarios.
Battlefield 2
Looks like the trends seen in SiSoft Sandra is continuing
here. Naturally, the 930MHz settings help the most, but at 800MHz,
the Corsair 8000UL still stays on top of Crucial's kit which is
running at the same speed.
Unreal Tournament 2004
Another close finish between the Corsair and Crucial
kits, but again the 8000UL finishes on top. We haven't talked
much about the 5400UL, but given the differences in clock speed,
it's last place finish is to be expected. What is surprising though
is it does stay close thanks to it's more aggressive default timings.
