Does good DDR3 really perform better than good DDR2??

[OCChronic]

I'm using an X38 DQ6 board & Corsair Dominator DDR2 1066, CL 5,5,5,15, 2T CMD. I'm getting really good Everest results in the 10000 range for all three memory bench marks and around 52ns for the latency bench. I'm running an E8400 @ 3800Mhz(475 x8, 1.344v@LOAD)

In 1 review, I've seen 11000+ scores for the same 3 memory benches & around 47ns for the latency bench while using top DDR3 2133Mhz kits @ rated speed and timings, an E8500 @ 3780Mhz(7 x 540Mhz), the X48 @ 1.75v on a $350.00 mainboard with water cooling on the MCH. This was after extreme tweaking by talented overclockers from a major review website.

Other reviews using x38 + DDR2 vs. x48 + DDR3 had similiar results and only top DDR3 kits coupled with high FSB & low Trd settings showed only modest real world gains over present X38/DDR2 systems.

 

[DSF]

Well you're really asking two different questions.

Is it really better than good DDR2? Yes, very slightly.

Does the minuscule performance increase justify the extra cost? That's a subjective question, but I think the clear answer is no.

 

[Sylvanas]

For benchmarking is DDR3 worth it? Yes
For general usage is DDR3 worth it? Probably not yet.

 

[GundamF91]

DDR3 is already slightly better than DDR2 (depends on application), but not quite yet right about now in terms of cost/performance. Just think of the early days when DDR transitioned to DDR2. It's not really until the final iteration of DDR2, officially DDR2-800, did DDR2 really shine over the highest spec DDR-400.