|
Performance & Overclocking
To
be clear, I will always present the timings in CAS-TRCD-TRPD-TRAS. Also for
consistency, here are the settings that did not change through the duration
of the testing:
oCPU:DRAM – 1:1
oAGP:PCI – Fixed, 66:33
oGame Accelerator – Auto
Aside from the FSB, the only other setting that was adjusted was the DDR
SDRAM Voltage, which I set for the highest rated voltage available on the
motherboard. OCZ warranties these modules to 3.0V, which is out of the range
of all but one of my motherboards. The one that would do it, as we will see,
proved to have other issues that prevented its use. OCZ seems confident that
these sticks can handle up to 3.5V, although I have seen reports of users
toasting the BH5 chips at that voltage.
My biggest problem writing this review was the fact that the OCZ EL 3200
sticks just do not want to perform on any of the three systems I had at my
disposal. Two were very closely related – an IC7-MAX3 and an IC7-MAXII
Advance. The reason for that was that the MAX3 up and died on me in the
middle of the review process, so I had to get hold of a similar board to
head off extra problems that I didn’t need. So, first off, here are the
numbers that the MAX3 produced before its untimely demise.
Not impressive, right? After discussing the
issue with OCZ’s tech support, Andy Talamantez at OCZ had this to say:
| “I want to thank Etrememhz
for taking the time to review the OCZ PC3200 Gold memory.
This product uses Winbond BH chips. The nature of these chips is
such that they do very tight 2-2-2 timings at 200mhz with 2.8v,
yet for a good overclock at 2-2-2 timings, they generally
require 3.4v to 3.5v. It has been well documented that the IC7
Max3 board, which is now about 2 years old, does not track Vtt
voltage with Vdimm. Vtt must remain at 1/2 of Vdimm to allow the
voltage to be of any use. Raising the Vdimm past 2.9v leaves Vtt
at 1.4v and the extra voltage only causes instability and
provides no gain. The fact that 210mhz was the maximum overclock
on a stock IC7 Max3 is not surprising. Testing on a more
modern platform such as AMD A64 with a board the can provide the
required voltage or using a DDR booster will yield much better
overclocking results.” |
Ultimately the limiting factor here is the amount of voltage that you have
available to feed these chips. On some of the newer NForce4 boards one has
up to 4V readily available, without any modifications.
With all that said I went ahead and tried the Gold sticks out on both the
Abit NF7-S and the IC7-MAXII Advance, just to put some numbers on the table.
Not everyone has the cash to happily upgrade to the next bleeding edge
platform, so I felt it my duty to give these sticks a fair representation to
those of you still on the Socket A and 478 platforms.
First off, we have the highest stable FSB obtained during my testing with
each motherboard. This was verified by running the SiSoft SANDRA Burn-In
wizard for 10 iterations with no errors. Although the MAXII Advance only had
a maximum VDIMM of 2.8V, it soundly pulled ahead in this race. This is some
of the first memory I’ve had to run at this high a voltage…one of the
reasons my old MAX3 wasn’t modded.
Next are the PCMark04 Memory scores at stock
and overclocked speeds, at 2-2-2-5.
The MAXII Advance did the best here, owing to
the greater stability of its VTT subsystem compared to the MAX3. The NF7,
although not having a chipset recommended for use with these sticks was
probably the most fun to overclock. Having that little Barton 2500+
crunching along at 2.2Ghz at these timings made for a rather nice budget
gaming rig.
|
