Memory Tests
As always, the memory tests get first billing. Pifast is used to get a rough idea and then ScienceMark 2.0 confirms.
It's a couple of seconds off a recent Ultra 400 board (ABIT's AN7) result, the direct effect of running the memory at DDR333 instead of the AN7's DDR400.
Sciencemark should confirm the bandwidth deficit and latency penalty.
90% bandwidth efficiency is in the nForce2 ballpark, some 500MB/sec behind a DDR400 nForce2 Ultra 400.
With an Ultra 400 board having a rough 82ms access latency when the CPU uses the memory controller, it's clear to see the N2VIEW's platform disadvantages. But bearing in mind that during the tests the memory controller was being constantly accessed by the onboard graphics framebuffer, you can see the motivation to equip nForce2 with a dual-channel memory controller, the N2VIEW's performance delta under load being less than the 18% CPU load would imply.