Memory Tests
Our memory tests first, designed to make sure a memory controller, whether it be on the northbridge as is the case with A7N8X-E, or on the CPU itself in the case of Athlon 64, is working correctly.Our custom Pifast test first, to 10 million places. Low latency memory access is a factor in performance here, along with overall memory controller efficiency. nForce2 Ultra 400 systems usually do well here, we're looking for the A7N8X-E to keep pace with the AN7. Performance comparison can then be drawn with the Pentium 4 and Athlon 64 based systems, to see if Socket A can still hold its own.
A7N8X-E is within twenty milliseconds of AN7, a real world performance that's essentially identical and within our margin of reproductive error. So far so good and with an eye to all other results in this article, prophetic to the extreme. A7N8X-E is going to sit very close to AN7 throughout, if you were expecting any sort of performance miracle, skip to the second last page of the article now. Our Sciencemark 2.0 test pairing will tell us why Pifast performed like it did.
82ns of latency from both nForce2 Ultra 400 boards. They are the slowest of the lot when made to fetch data from main memory, but far from slow. However in memory limited applications, where the CPU has less of an influence on things, the Socket A systems will start to trail.
Just shy of 3000MB/sec of CPU-to-memory controller bandwidth, being a STREAM test it's somewhat of a CPU benchmark too, no bad thing when evaluating motherboard performance. ~93% efficient is in the upper spectrum of memory controller performance on boards like A7N8X-E, a good showing.
CPU bound tests now, to see if A7N8X-E can carry on in the vein it has started in. Competitive performance with AN7 is what we're looking for.
