elcome to the first part of our large DDR2 roundup. In this first part we're going to take a renewed look at all DDR2 kits we tested so far, and give you the most important facts. This first part is all about Intel's 965 chipset, which was until recently the best clocking chipset around for DDR2 overclocking fun.
On the next pages we're going to have a look at the contestants first, and further on we're going to determine our performance champions on the Asus P5B-deluxe board, one of the best selling 965-based motherboards of the past year. Later this month, we're going to present you with part 2 of this roundup, featuring a different motherboard and chipset, with more DDR2 models added into the mix... Why different parts, you might ask ? I'll try and explain why it is so important to test on different boards at the beginning of our second roundup, for now, let's just say that no two boards, and certainly no two chipsets, handle the ram in the same way... T
o review a pair of DDR2 memory sticks seems like madness or editorial suicide nowadays. Unless you have been living in your closet for the last couple of months, all of you know that DDR3 is upon us. Certainly since Intel launched its new P35 chipset, and with the X38 around the corner, DDR3 is gaining popularity quickly, especially among my fellow reviewers. If you look at many of the competitive websites, one would sometimes think that DDR2 is dead already, and DDR3 is the Holy Grail we've all been waiting for. In the next paragraphs I'll be expressing my own point of view on this newly found Holy Grail, but what is sure is that we've entered what seems to be another transitional period in the memory landscape.
OCZ DDR3 Modules
The big question at this moment is still : are we talking a real, short term transition to DDR3 here, or are we getting the DDR2 transition story all over again? Let's not forget that, at this moment, AMD’s AM2 and Intel Core 2 are barely one year old, with most people only upgrading from DDR towards DDR2 since June or July 2006. As you might know, Intel has been pushing DDR2 since their transition to the 775 pin socket in June 2004, but with very little success in the first two years (mostly due to AMD having the performance crown, at the time). It is the same Intel that tries to convince us to move towards DDR3 in the immediate future, so will they succeed in a faster transition this time around?
The first deciding factor in the DDR3 transition will no doubt be performance. When real world tests prove that DDR3 is indeed faster, I guess DDR2 will be short-lived indeed. As with DDR2 however, DDR3 does not seem to offer a big, immediate benefit over DDR2. The problem - as with DDR2 - seems to lie with latencies. Right now, the performance crown depends on application to application, where sometimes DDR3 is able to benefit from its higher speed rating (and higher bandwidth). Other applications that benefit more from latency have a preference for good "old" DDR2. Until the industry is able to make truly lower latency DDR3, performance does not seem to be a persuasive argument.
A second factor in the discussion will no doubt be the position of AMD. Right now, for the remainder of 2007, and even eatly 2008, AMD does not have plans to introduce DDR3 support in their on-die memory controllers. Support for DDR3 will be there "some time into 2008", together with a brand new socket (AM3) and processor die at 45nm. In other words, if AMD takes back (part of) the performance crown later this year - a big questionmark for anyone still - all the enthusiasts and early adopters of DDR3 will be out of luck. DDR3 will not drive their brand new Barcelona/Phenom CPU's to new performance heights...
Furthermore, in favor of DDR2, here's one very persuasive argument for you: price. Right now, DDR3 is as expensive as DDR2 was one year ago. DDR2 prices have plummeted however, and a 2Gb value kit of DDR2 will set you back no more than $100, whereas a truly high performance kit will not retail for much more than $200. For consumers, getting 2Gb of Ram now is a no-brainer, whereas a year ago this was not as clear. On the other hand, th low prices represent a problem for the memory manufacturers as well : the margins on DDR2 memory products have shrunk to far lower levels than before, and it might be in the manufacturers own interest right now to introduce something (slightly) faster, which they can sell with the right margin.
There is one last thing on the performance point of view though, which could speed up the transition a little: in my opinion, DDR2 development has been at a standstill for the last couple of months. Sure, memory manufacturers have come up with lots of new models, but they seem to rely more on form than function by now. The speed grades of DDR2 have not changed much over the last 9 months, and the kits that were the fastest 9 months ago... still are the fastest today. Micron has played a very big role in the DDR2 story, but even the best Micron chips on the very best PCB's only have that much performance headroom left. Voltages have been increased far above JEDEC's specifications already, but speeds greater than PC2-10000 (or 625Mhz - 1250Mhz DDR) with an acceptable production yield do not seem feasible. In this respect, memory manufacturers have turned to DDR3 for innovation, and speed records. In the end though, DDR2 has a lot of life in it left in my opinion, albeit in the current speed grades, or not.DDR2 Roundup
Until DDR3 becomes mainstream... we present you with Part 1 of our DDR2 Summer Roundup:
Let’s take a closer look at each of these kits on the next pages ->