Upon its introduction, DDR2 did show promise, yet failed to stir much enthusiasm on behalf of the PC community. While it did offer high-speed performance, its extraordinarily high latencies hamstrung bandwidth. A dodgy situation was further exacerbated by DDR2 being limited to just two Intel chipsets. Intel championed the BTX standard which introducing PCI-ex was not without merit. Problems arose, however; as PC-users began to realize the cost associated with acquiring Intel's latest would involve the purchase of a CPU, CPU-cooler, PCIe graphics card, PSU and of course DDR2. Then PCI-ex fell short as well which made the move to the new standard more sideways, perhaps even backwards rather then forward. Native PCI-ex GPU's were difficult to find except among ATI based cards. In the rare event an nVidia card could be procured the fact it performed as well as native PCI-ex cards while employing a bridging chip was further proof all one would incur switching to the new standard was credit card debt. Sure you could claim to be ready for the future, but it wasn't necessarily a future anyone wanted (at least initially).
In an attempt to quell potential silicon street riots, Intel had a trick up their sleeve which may involve going back to the drawing board; so far back they landed squarely on an SiS idea from CIRCA 2001. This coincided with the birth of the SiS-735
chipset which offered DDR/SDR capability in a single solution. Jumping ahead to the present the infamous i915x chipset was born. Able to support either DDR/DDR2, AGP/PCI-ex the i915X seemed like the answer to many a prayer. Problem was, the chipset performed about as well as the 735 did back in 2001. In its earliest days only two good things came out of the new standard, motherboard power circuitry better suited to run the Prescott Socket 775 edition and their i925x flagship chipset which did have potential.
Undoubtedly DDR2 has had a difficult birth however the "little BGA-IC that could" is doing very well for itself. As with all things born of technology, the maturation process (while never fast enough for the PC-enthusiast) is quite rapid occurring over a period of a few months. Ideally by the time the bugs are worked out one would hope the standard hasn't become obsolete. All humor aside, considering where DDR2 began in so far as its rediculously high latency was concerned, some remarkable progress has been made. From 4-4-4-12 (some even said 5-4-4-15) were now getting back to reality at 3-3-3-8. In fact one of our candidates ran below their advertised SPD of 3-2-2-8 running at 3-2-2-6. In contacting manufactures for this article I simply chose among those brands "I" would spend my money on. Given the fact a spinal ijury left me dependent upon Social Security Disability, that budget is about as fickle as it comes. Taking into account my knowledge of PC's and strict budget limitations, I think it's fair to say product's recommended here have been thoroughly scrutinized. Without further adieu let’s "scrutinize." The Usual Suspects (listed alphabetically)
Corsair TWIN2X1024-4300C3PRO CL3-3-3-8 @ 538MHz
Corsair TWIN2X1024-5400C4PRO CL4-4-4-12 @ 675MHz
Crucial Ballistix PC2-5300 CL4-4-4-10 @ 667MHz
Kingston KHX5400D2K2/1G CL4-4-4-12 @ 675MHz
Mushkin PC2-5400/1G CL4-4-4-11 @ 667MHz
Mushkin PC2-4200/1GLL CL3-2-2-8 @ 533MHz
Transcend DDR2-533 CL4-4-4-11 @ 533MHzAll modules were either matched 2x512MB, or 2x512MB and ran in Dual Channel | DIMMS 1/3.
|CPU ||Intel P4 550 Retail-3.4GHz LGA 775 |
|Mainboard ||Abit AA8 DuraMax (BIOS 18)|
|Memory || Corsair 4300C3PRO|
Crucial Ballistix PC2-5300
|Graphics ||ATI Radeon X800 XT PCI-ex|
|Power Supply ||OCZ PowerStream 480W|
|Operating System ||Windows XP SP2||Water Cooling ||Alphacool Xtreme Pro Kit 240 12V substituiting Alphacool's NexXxos XP BOLD S.775. |
In recent years the overclocking phenomenon has become as trendy as girls kissing girls on FOX television. Motherboard makers seeking to attract the hard earned cash of this burgeoning population have begun to design products with some remarkable overclocking features. Hardware settings which were once accessible only through BIOS, can now be made in the Operating System. ABIT
the manufactuer of our Duramax AA8 utestbed has made substantial progress in this respect. Their efforts have materialized in a technology known as µGuru
. Unlike ad hoc
software based solutions, µGuru is hardware based taking the form of a Winbond controller chip on Steroids, which has been specifically manufactured just for Abit. µGuru offers a comprehensive GUI utility which monitors temps, voltages, frequencies, fan-speeds all observable in real-time and mates this with a number of hardware settings right at your finger-tips.
Why mention an overclocking utility in a memory roundup? As Overclockers demand more from their PC's and motherboard maker's labor to meet these demands, it's only natural memory maker's would follow suit. In actuality this formula is juxtaposed with memory maker's such as Corsair, Mushkin, OCZ, etc. designing well beyond JEDEC
standards by years. The result has been some versatile SPD (Serial Presence Detect) settings pre-programmed into all overclocking DDR and DDR2 alike. I've found many SPD setting's to be just as effective as manual settings and in most cases synonymous. To minimize the number of variables, SPD settings will be used throughout this review. This not only makes for a more objective test environment, it places the burden of performance on each manufacturer. It must also be said then the Abit AA8 Duramax along with most other boards doesn't begin to reach the speed of current DDR2 parts, which is the reasoning behind dividers (or more accurately memory multipliers). Where we test PC2-4200/4300 parts the Abit BIOS dictates 1:1 FSB performance, where the parts are PC2-5300/5400 the divider runs at 3:4. The following CPUZ thumbnails exemplify each memory under its SPD at 200MHz, and show SPD timings for each manufacturer in general.Corsair 5400C4PROCorsair 4300C3PROCrucial Ballistix PC2-5300 Kingston KHX5400Mushkin PC2-4200LLMushkin PC2-5400Transcend PC2-5300
Of interest would be memory brands/versions with multiple SPD settings as these not only ran up to speed, but may offer future benefits. At the onset of testing Abit BIOS-18
and CPUID CPU-Z
Version 1.24 were installed. For consistency these were left as is, until the first full round of tests were completed. Upon retesting BIOS-22, and CPUZ version 1.28 no changes occurred in either the SPD settings, motherboard/memory performance, nor were additional "features" revealed with CPU-Z v.1.28.
Onto the tests ---->