Memory test – Joe
SUMMARY: Delivers the goods and then some.
“Through fine-tuning of the resistor network on the PCB we have been able to achieve the highest possible speed beyond 400 MHz while maintaining the capability for low latency operation.”
Sounds like a challenge! OK, let’s see how fast these can run.
I used the UXD RST Pro2 Memory Tester. RST Pro2 is a stand-alone PCI card that only requires RAM, CPU and video on the motherboard – it is OS independent, self booting and runs 30 different memory tests to determine the point of failure. This is an updated version that I used in the first tests of this RAM – this supersedes the earlier report.
In running this test, I used an ABIT NF7-S motherboard, detuning the CPU so that it would not be a factor in testing. The only items populating the motherboard were the RAM, CPU, videocard, floppy and RST Pro PCI card – not much of anything to get in the way. I ran two sticks of 512 simultaneously to test how well they interacted.
Each test loop takes about two hours; if a failure was detected, I stopped the test and ran another series until it passed 5 test loops – about 10 hours for each “Pass”. I ran the bulk of these tests at 2.8v, which is the maximum recommended by Mushkin.
Memory Timing @ Voltage
6 2 2 2.0 @ 2.8v
5 x 131
7 3 3 3.0 @ 2.8v
5 x 140
11 2 2 2.0 @ 2.8v
5 x 131
6 2 2 2.0 @ 2.6v
5 x 122
At Mushkin’s spec memory settings, I found it stable up to 436 MHz at 2.8 volts – 9% over the spec 400 MHz. The RAM would post above that but with increasing errors as speeds went up. It’s possible that some may be able to post and run benchmarks with minor errors. Dropping voltage to 2.6v (the lowest on this mobo) significantly impacted stability – 406 MHz at 2.6v compared to 436 MHz at 2.8v. I suspect going over 2.8v will add a few more MHz, but IMHO not worth it.
Relaxing timings to 7 3 3 3.0 bumped the speed to 467 MHz – a 16.8% increase, although relaxed timings such as this will negatively impact on-screen performance. In addition, I tested using 11 2 2 2.0 (some have found nForce motherboards to perform better at these settings) and found the RAM to run no better than at 6 2 2 2.0.
While I can’t attest that this DDR attains the “…highest possible speed beyond 400 MHz…” compared to some other DDR, Mushkin’s PC3200 DDR 222 LII V2 does the job nicely at spec memory settings (6 2 2 2.0) at 2.8 volts and can be pushed further with detuned memory settings.