A-Data Vitesta DDR2-800+ Extreme Edition
Author: Zahn Funk
Editor: Shawn Knight
Date: 02-19-2008
Provided by: A-Data
Discuss: View Comments

I am testing the A-Data Vitesta EE modules with a variety of benchmark programs including SiSoft Sandra XIIc, Everest Home Ed 2.2, RightMark Memory Analyzer 3.72 STREAM standard, PCMark05 and SuperPi Mod 1.5XS. For comparison, I ran the memory at three different performance levels: the first at JEDEC SPD and stock system settings, then with the lowest stable EPP values and finally with the EPP values overclocked.

Having the CPU frequency lower on the overclocked setting may skew the results somewhat but unfortunately the modules would not run higher than 440MHz with any kind of stability so this was the closest I could get to the stock CPU speed. First up is the SiSoft Sandra Memory Bandwidth test.

Bandwidth typically favors outright speed over timings and here we can see this occur. Lower timings do show an improvement but not as significant as higher overall frequency. Next is the latency test.

Latency usually yields better results from memory timings over speed alone but in these tests the A-Data Vitesta EE scales pretty evenly. Next we will look at the Everest Read performance.

The Read test shows an even improvement between timings and speed, now let's look at Write performance.

Here we see timings play a more important role than frequency. Now on to the Everest Latency test.

Again the Everest Latency test shows timings offer a bigger improvement over the memory speed.

The RightMark Analyzer uses the STREAM test, a recognized standard for comparing memory performance. Here we see that the speed of the modules yields better results than timings alone. Next up is PCMark05.

The PCMark05 results are inconclusive. Strangely it gives the highest score when the memory is run at the slowest speed and timings. The battery of tests that PCMark uses are fairly intensive, we can only assume that the timing values of 4-5-4-12 are not 100% stable.

The memory operations latency data is consistent with the other tests, indicating an improvement with lower timings and faster speed.

The SuperPi calculations indicate an anomaly in the data, showing a slight improvement in lower timings but a decrease in performance with faster speed. Again, we can only determine that something with the memory overclock may not be 100% stable.

On to final thoughts and conclusion.

  Sponsors Links

  Sponsors Links