Articles: Memory
 

Bookmark and Share

(23) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

Stability Tests

We verified stability of operation in two steps. First, we used the Memtest86+ utility to make sure there were no errors at work. Then we reconfirmed this result by running Windows-based S&M and Prime95 utilities. This two-step approach ensures that we get trustworthy results.

Now let’s get closer to the actual results. The diagrams below show the maximum frequency the memory modules were stable at (as confirmed according to the above-mentioned method) with the given timings.

The first thing you can notice here is that top-end modules of DDR2 SDRAM from different manufacturers do not differ much. It is especially clear with the memory kits from Corsair and OCZ which have almost the same maximum frequencies at different latencies. It’s only at low timings that the OCZ DDR2 PC2-8000 Platinum Enhanced Latency Dual Channel can work at a few megahertz higher frequency than Corsair’s TWIN2X1024A-5400UL. When less aggressive timings are selected, the memory from Corsair and OCZ behaves identically. For example, neither module can conquer 1000MHz frequency even though this frequency at 5-5-5-15 timings is declared to be the default mode of operation for the OCZ kit.

The memory from Mushkin behaves somewhat differently from the other two kits. It is a little worse at aggressive timings. For example, the XP2-6400 DDR2 modules are not stable even as DDR2-667 at 3-3-2-8 or 3-3-3-10 timings. On the other hand, the Mushkin memory accelerates at high timings and notches a better result than the competitors at 5-5-5-15 (by 4MHz only, so this is not a very convincing victory).

To check how the voltage of the memory may affect its overclockability, we tried to give the modules 2.3 volts (this is the maximum the ASUS P5WD2 Premium mainboard can give out) at 5-5-5-15 timings and see what could be achieved. Our modules reacted quite positively to the higher voltage, allowing to raise the “stability bar” more. To be exact, the maximum frequencies at 5-5-5-15 timings and 2.3V voltage were:

Corsair TWIN2X1024A-5400UL

1044 MHz

Mushkin XP2-6400 DDR2

1048 MHz

OCZ DDR2 PC2-8000 Platinum Enhanced Latency Dual Channel

1036 MHz

In other words, our increasing the voltage by 0.2V improved the frequency potential of all the participating modules by about 6% and made it possible to overcome the meaningful barrier of 1 gigahertz.

So, we can say that PC enthusiasts now have much flexibility in configuring the memory subsystem of their LGA775 platforms. Like with DDR1 SDRAM, it is possible either to increase the frequency or reduce the latencies of the memory. Which method is more profitable? We’ll answer this question in the next section of the review.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

Discussion

Comments currently: 23
Discussion started: 08/23/05 05:53:16 PM
Latest comment: 08/25/06 01:25:03 PM

View comments

Add your Comment