Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
CPU Intel P4 840 D P4 820 D P4 630 P4 640 P4 650 P4 660 P4 670 AMD Athlon64 3500+ 3700+ 3800+ 4000+ X2-3800+ X2-4200+ X2-4400+ X2-4600+ X2-4800+ 1-Way Opteron Opteron 144 Opteron 146 Opteron 148 Opteron 150 Opteron 152 2-Way Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Opteron 248 Opteron 250 Opteron 252 2-Way Dual Core Opteron Opteron 270 Opteron 275 nVidia GF 7800GT GF 6800GT GF 6600GT ATI R X850 XT PE R X850 XT R X800 XT PE R X800 XT R X800 XL Memory Corsair Crucial Kingston Mushkin OCZ What are you shopping for?

3DMark2001SE

It is still a classic and it is reproducible, therefore, regardless of whether it really represents actual games or not, it is representative for system performance, at least as far as we can tell.

Since this is a reference benchmark, any performance hit will hurt twice as badly and 400 points are 400 points no matter which way one looks at it.

FinalFantasy XI Benchmark2

FinalFantasy is a role playing game (RPG) ported to a multitude of individual players within each scene. This makes it very different from the standard first person shooter with limited variables, rather, there is a massive amount of user input from different sources, which requires constant recalculation of the different activities within any given scene and adjustment of the different user actions on the other hand. We don't have any concrete information on the what and how but it appears as if the benchmark follows the execution of a script rather than playing back a pre-recorded demo. This is very similar to the situation in the, by now obsolete, "Expendable" timedemo. Needless to say that in situations with a lot of user interaction, there will be a fair amount of writes and, consequently a lot of bus turnaround, which, in turn, means sensitivity to all kinds of latencies.

We re-ran this benchmark about 20 times in any sequence between modules, with the frequencies set to auto or manually specified and -- we were not able to change the outcome of the benchmark as shown above. At the same latencies, the TSOP modules are 16% faster than the BGA modules, everything else was 100% identical.

Conclusions

There is, at this point, not much to conclude, the results speak for themselves. There are a number of questions, though, particularly with respect to where the performance hit encountered with the Kingston and Samsung modules comes from. We are in touch with the Kingston engineers who are looking into these issues and will follow up as soon as we get any concrete feedback pertaining to the problem.

Ironically, the Kingston DIMMs are the only ones qualified by AMD, everything else is out of spec and I am already anticipating some emails from our friends at Intel, telling me that if it does not conform to JEDEC specs, or if it is not qualified, it does not count even if the performance is better. It would appear, though, that this is another case for the consumer to decide whether qualifications really matter that much when compliance with them has to be bought with a massive performance degradation.

Special Thanks to Heather Jardim, Kingston and the Kingston engineering team for immediate feedback and discussions.
Disclaimer: We have spent two full days running these benchmarks over and again under any conditions we could come up with. We also had similar results replicated by a third party but that still does not mean that there is no margin for error on our side.

next page:    => More =>