Print this article

OCZ DDR2 PC2-5400 Gold GX XTC 1 GB


Date
: 04/12/06 – 03:22:08 AM

Author
:

Category
: Memory

Page 1 : Introduction

Manufacturer:
OCZ Technology

Price:
$132 (Newegg.com)

OCZ's DDR2 PC2-5400 Gold GX XTC we're looking at today is an affordable memory that perhaps some of you have been eye balling or already have purchased. Some of the reasons that could interest someone in this memory are brand name, reputation for good overclocking or simply price. You can be sure of the brand name and price here, and in fact this is some of the cheapest PC2-5400 memory on sale at Newegg, but what we're looking at today is how well this memory is able to overclock. I also revisit the CORSAIR TWIN2X1024A-4300C3PR reviewed some time ago, but limited at that time by motherboard/chipset limitations.

Page 2 : Specifications

The following specifications come straight from OCZ's site:

Much of the basic information you want to know is there: CAS timings, voltage, etc. One interesting thing about OCZ's memory is that they warranty the stuff at up to 2.2V. It's nice because you're given some extra peace of mind while you're trying to tweak the frequency or CAS latencies to your liking and adjusting the voltage in the process.

While I was unable to determine the PCB maker, I was able to find some information on the chips. The chips used on these modules are made by AENEON, a division of Infineon (I just read in the news yesterday Infineon is getting out of the DRAM business *shrug*). The product number on these chips is AET93F370. Unfortunately AENEON's module nomenclature seems out of date, because the 3F sequence of the model number is a mystery to me, and the other pieces don't give away anything more than we already know (who makes them, product family, depth and speed). Let's then see how much headroom these modules have in practice.

Page 3 : Performance

What you're here for of course though is to see how these guys perform. If you're curious to see how DDR2 performs, glance over to here where I tested various makers some time back. Nothing's changed in terms of performance and we'll only see a slight negligible variation between manufacturers when tested at the same timings and frequencies. What's important and probably what you want to know when you're buying this stuff is what users are seeing in terms of headroom. We give you our experience below. As an aside, and probably mentioned in the article I linked above, you should run your memory at 1:1 ratio and with the tightest stable CAS settings possible for best performance; Asynchronous frequencies tend to hurt your performance more than help it.

The tests were conducted using the following systems:

The Foxconn motherboard here, and more importantly its chipset have made this review so much easier. Previous boards (about three or four of them) all based on Intel chipsets have driven me up the wall with limited ways of getting memory to its limits. Possibly only motherboard specific (broken BIOS settings, fixed asynchronous memory settings), all my problems were gone on the nForce4 Intel platform. Finally I can show you guys the upper limit of the modules being tested and without overclocking any other component in the system.

The procedure for determining stability was first running a 32M SuperPi calculation without rounding error. Next up the setup had to successfully run through three successive loops of Doom3, FarCry, 3DMark2k1 SE, 3DMark2k3 and 3DMark2k5. Once that was complete the results were considered stable. Some of the setups (namely the Corsair) also suffered through my own Oblivian and WoW playtime.

So the results? A bit disappointing when it came to discovering how much headroom these guys had left in them. The modules did provide some headroom, but they could only run stable at a max frequency of 700 MHz (2.1V). 711 MHz was close to doable, but neither loosening the timings or raising the voltage to 2.2V could get this memory to get past the SuperPi test. What did surprise me was the amount of headroom offered by the Corsair modules which were able to run stable at 720 MHz (2.1V) while still running at 4-4-4-12 timings no less.

On the flip side, both modules were able to run as low as 2-3-2-8 at 400 MHz and 3-3-3-8 at 533 MHz. Both offer plenty of headroom for the overclocker, and the extra headroom of the Corsair is certainly justified in its price: it costs $40 more than the OCZ modules tested here.

Page 4 : Conclusion

I focused on what many of you hopefully find important when shopping for memory. Apart from the brand name and warranty, the first thing I'm wanting out of RAM is headroom. Walking into your favorite retailer and being able to find this out is not likely, so hopefully this straight and to the point article gives you that information.

As for my experience and recommendation, I have to say this is fine RAM for someone looking to overclock. Why? Out of the package you're guaranteed a nice frequency to work your frontside bus up to as you keep the memory frequency at a 1:1 with your processor and the latency as low as you can without the system becoming unreliable. The thing is though, would I buy this particular RAM kit? Let me put it this way: I can get cheaper OCZ or competing kits, with equal latency timing, at a lower cost and still have that same guaranteed frequency.

Advantages:

High frequency
Great warranty

Disadvantages:

Not much headroom beyond stock
Cheaper PC2-5400 Kits available

I'd like to give our thanks to OCZ for supplying the memory to make this review possible.