Review: Gigabyte GA-K8NXP-SLI

by David Ross on 14 February 2005, 00:00

Tags: Gigabyte (TPE:2376), AMD (NYSE:AMD), NVIDIA (NASDAQ:NVDA)

Quick Link: HEXUS.net/qa6e

Add to My Vault: x

Memory Tests

Let's kick off the testing with a few tests which measure the performance of the motherboards memory subsystem, starting with our own custom Pifast benchmark.

Pifast


Not much to split the three boards here, all performing very similarly in this test.

Sciencemark 2.0 - Bandwidth

Sciencemark 2.0 is up next, used first to measure the peak bandwidth between the CPU and memory controller.


Sciencemark consistently greeted us with a very odd result for the Gigabyte board, giving results that wouldn't look out of place using PC2700 RAM, despite this not being the case. Testing using other tools to measure memory bandwidth on the GA-K8NXP-SLI gave figures on a par with the other two boards on test here, as you would expect considering that the memory controller is integrated into the Athlon64's die, and thus there should be little difference between scores in this kind of benchmark. With that in mind, we can chalk this result down to an anomaly with ScienceMark itself, rather than any kind of performance issue with the Gigabyte board.

Sciencemark 2.0 - Access Latency


ScienceMark's latency test is a similar story, again showing poor results for the GA-K8NXP-SLI. Again, these results are disproved by other benchmarks to test latency, which show the Gigabyte part on par with the other nForce4 board tested here.