DDR-II is a relatively new mainstay in the enthusiast market, therefore I would like to cover some differences between DDR and DDR-II. DDR-II (Double Data Rate Two Synchronous Dynamic Random Access Memory) has become a mainstream solution here in late 2004, one of the main advantages of DDR-II is the ability to work at a higher effective clock speed while reducing the actual internal speed of the memory.

A better explanation might be as follows, using a simplistic 100MHz clock:
SDR - Transfers data on every rising edge of the clock = 100MHz throughput.
DDR - transfers data on every rising and falling edge of the clock = 200MHz throughput.
DDR-II - Transfers data twice on each rising and falling edge of the clock = 400MHz throughput.

Other performance enhancements are achieved through increased number of buffers, improved prefetch algorithms, packaging and on-die termination. Through all of this, latency is, however, increased. Typically DDR-II modules Frequency's are decreased, allowing for lower electrical requirements while increasing overall throughput due to the nature of the "quad-pumped" modules. This allows the modules to operate at a lower temperature, and lower voltage requirement.

Well, now that we got that out of the way, lets look at the modules that Kingston has sent over for testing. The modules come in the standard Kingston HyperX Blue Heat Spreaders, however, I notice immediately that these modules are much more substantial than the last pair I tested as far as weight is concerned, although, still not as heavy as OCZ sticks. The increase in weight could be attributable as a result in the increased number of components on the PCB, none the less, there is an increase.

DDR-II memory is identified using PC2 instead of just PC in its speed rating. The HyperX modules I am testing today are PC2-5400, you can take this as being equal to 675MHz, which is basically a quad pumped PC-168 SDR module (remember PC-150 is where SDR basically stopped).

One of the major issues with testing a new format, is typically you are unable to test that format in a "comparative" manner. Luckily this is not the case today, as I have a motherboard that allows either memory type. Without further ado, here is the test bed:

ASUS P5GDC Deluxe
ASUS N5900 Extreme
WD 80GB 7200RPM SATA
Intel 520 (P4 2.8E LGA775)
OCZ PowerStream 420

Memory:
Kingston HyperX PC2-5400 (1.85V) (4-4-4-10)
OCZ Gold PC3700 (2.6V) (2.5-3-3-7)

Sure, I could have tested the HyperX DDR-II memory against another set of HyperX DDR memory, but what fun would that have been, and I am sure Kingston has already performed that test. Since there was little performance difference between the HyperX PC4300 and the OCZ PC3700, I chose the OCZ so I could get a good feel for how DDR-II is going to better my life as it were.

A few notes on another difference between DDR and DDR2 memory. I am sure like me, you see the CAS ratings stick out like a sore thumb on the HyperX memory over the OCZ. Not to worry, a CL4 on DDR2 is not much different than a CL2.5 on DDR, this is due in part to the increased clock of DDR2. In the most basic terms, if I was comparing DDR-2700 to the HyperX PC2-5400, the CAS rating of the PC2700 would have to be 2-2-2-5 in order to maintain pace...

The test suite will include the following:

SiSoft Sandra 2004 SP1b
PiFast
TMPGEnc Plus 2.5
PCMark 2002
3DMark05 Build 110
UT2004 (minimal settings 640x480)

Overclocking

The ASUS P5GDC Deluxe is probably one of the simplest overclocking motherboards you will run into. There are several ways to implement an overclock, the easiest of which is selecting a percentage of overclock from the BIOS. I simply select the amount of overclock I desire and it configures it automatically, FSB, Vcore and VDDR2 as required to achieve the desired amount of OC. There is, of course, manual overclocking, which is what I will be using for today's testing.

With the motherboard set to manual I was able to achieve 250FSB (3.5GHz) without any change to the memory timings or voltages (Vcore / VDDR2) using the HyperX DDR2 memory. A slight increase in stock Vcore to 1.40 allowed me to get to 255FSB (3.57GHz), but there was nothing I could do to get any higher. I continually received a "Drive Unknown or Bad Boot Sector" error for anything above 255FSB. There was nothing I could do to achieve a higher overclock. The OCZ did not fair as well, my maximum OC using the OCZ PC3700 was 215FSB (3.010). Loosening timings or increasing VDDR did nothing to allow me to boot above that threshold. Just to ensure everything was kosher, I popped in my Kingston HyperX PC-4300 sticks, and ran into the same ceiling as the OCZ.

What does all of this mean? Well, my take on it is that the 915P chip set is still in its infancy to a degree. Hopefully later iterations of the BIOS can get around this boot drive issue (after running into it I researched it and found a few other sites that have run into the same issue). The Kingston HyperX DDR2 memory shows nothing but headroom above and beyond 255FSB, this is after all, 675MHz memory. I would not be surprised to see someone eventually getting a 300FSB out of this memory, but that is just theory unfortunately.

NEXT