|
Today, I will be running a single stick of this RAM
against a single stick of TwinMOS PC2700 followed
with an additional run with a second stick to utilize
the nForce2's DualDDR features. I know that it seems
a bit unfair, to pit 400MHz RAM against PC2700, but
the FSB of the test rig here is @ 333 which is PC2700
anyhow, so therortically it's a fair match until the
point of overclocking.
The test bed here in the 3DV dungeon is based on
Chaintechs nForce2 Motherboard and has the following
specs:
Chaintech
7NJS nForce2 Motherboard
Athlon XP2700+ @333 FSB
Thermaltake
Volcano7+
Thermaltake
350W PSU
XFX
Graphics GeForce 4 Ti4200
XtremeDDR
400true
TwinMOS
PC2700
My first step is to get the fastest timing possible,
after which I'll inch the bus speed up until it won't
go any higher, then I'll do the same for the TwinMOS.
After each successful boot I will run 3DMark 2001SE,
build 330 to ensure stability. At the stock speed,
max RAM timings and max OC, I'll look at the performance
with PCMark and SANDRA. Finally I will use 3DMark,
Unreal Tournament 2003 ([H]ardOCP script) and Return
to Castle Wolfenstein (Checkpoint demo) at the CPU
test level (640 x 480) to determine performance.
Now that we have a plan…lets go
To start with, I'll set the motherboard
BIOS to "Optimal" and run 3DMark, here is
the base score for the TwinMOS. The RAM timings at
this point are 7-3-3-3, and the CPU interface is set
to "Aggressive."
Here's how the XtremeDDR looks from
the start. Hmmmm strange beginning, I rechecked this
several times and got very similar results each time.
I hope we see better than this.
With a point of reference to begin with, I'll see
how aggressive of timings I can get the RAM to stabilize
at.
The TwinMOS doesn't want to get much faster than
optimal with a max timing of 7-3-3 CAS 3, so I will
refrain from including duplicate scores.
However, the XtremeDDR is quite happy
at 4-2-2 CAS 2. Note the performance improvement.
Now for the FSB increase. With the most
aggressive timings I could reach, I set out to see
just how far I could push the FSB at those timings.
Please remember that the TwinMOS is PC2700, so if
it doesn't go much higher than 266 it's no big surprise;
however, if the XtremeDDR shows weakness there then
something is wrong. After an afternoon of 2MHz increases,
here's what I came up with.
The TwinMOS stayed stable up to 173MHz.
That seems fair for a stick of PC2700; however, I
had to drop the latency to 8-4-3 CAS 3 at 172 for
it to obtain this speed. Note the mediocre performance
boost from having to slow the timing down.
Now the XtremeDDR surprised the hell
out of me. I was able to push it to 172 at 4-2-2 CAS
2 with no trouble at all. I dropped the CAS timing
to 2.5 and kept going. With a little help from Chaintech's
wide range of voltage settings, I was able to give
the XtremeDDR 3.2V and stabilize at 180 MHz with timings
of 4-2-2 CAS 2.5. Now that's aggressive, look at the
performance boost!
To put things into perspective, here's
a graph to show these results side by side. Note the
performance boost from the XtremeDDR at max timings.OC
1,000 points already... I can't wait to add another
stick!
With our range in place, I ran the gamut of tests
I mentioned earlier. Here are screens and graphs that
reflect my findings.
|
