Test bed specifications:

AMD Athlon 64 FX-55 (ADAFX55DEI5AS CAA2C 0442GPMW)
DFI LP NF4 SLI-DR motherboard
EVGA 7800 GTX KO video card
ASUS EA800XL video card
OCZ Powerstream 600W PSU
Western Digital Caviar 80GB HD
Sony DRU-720A DVD/RW
Sony FDD
Windows XP Professional (SP2 and all patches applied)
OCZ EB DDR PC-4000 2x1024MB Platinum Edition Dual Channel Kit

Software Methodology:

SiSoft Sandra Professional version 2005.10.10.69
A64 Tweaker v0.60 Beta
CPU-Z Version 1.31
Memtest version 1.65
Lavalys Everest Home Version 2.20.405
BenchemAll Version 2.651 beta
RightMark Memory Analyzer v3.6
DriverHeaven Photoshop v2 Benchmark
SuperPi SSE2
3DMark®05 (Build 1.2.0)
3DMark®03 (Build 3.6.0)
3DMark®2001 Second Edition (Build 3.3.0)
PCMark®05 (Build 1.1.0) Advanced
AquaMark3

The goal of this review was to see how far this 2GB dual channel memory kit would carry me. I sought to find out how far I could go with the tightest timings of 2-3-2-5 (CL-TRAS-TRCD-TRP ) at a Command Rate of 1. I was able to run the memory at 219 HTT using the default 13X multiplier from my A64 FX-55 Clawhammer processor at 2-3-2-5. I then ramped up the front side bus speeds further, and ascertained stability at 250 HTT using a multiplier of 10X - with the memory timings set to 2.5-3-2-7 at 1T. How high did the memory finally scale with slightly more relaxed timings? I was able to run the memory at 275 HTT again using a 10X multiplier - attaining DDR550 speeds at memory timings of 3-3-2-7 at 1T. There were a few inconsistencies that cropped up while trying to run 3D benchmarks and high end gaming applications which will be mentioned later in this review.

It's time to examine the benchmarks so you can see the visual evidence of our testing. Please click to enlarge images.

Above you'll see the default screen shot results from CPU-Z version 1.31 - with the CPU screen, SPD timings, and Memory screen shots visible from left to right. Please note the SPD timings assigned by OCZ to this memory. The frequency is 250 MHz, with a CAS Latency of 3, RAS to CAS of 3, RAS precharge 2, and TRAS is set to 8. Below, the thumb nail image of a verified CPU-Z screen shot shows a front side bus speed of 219 times a 13X multiplier - still at 2-3-2-5 memory timings with a Command Rate set at 1T. This alone gave great bandwidth results with a frequency of 2847.31 MHz at these overclocked settings, which is about the limit of my FX-55 processor on air, using the stock AMD heat sink and fan assembly.

Above you'll note I ran a front side bus speed of 250 and dropped the multiplier to 10X. The memory timings were gradually relaxed to 2.5-3-2-7 at 1T. I dropped the multiplier to see how high the memory would scale at these timings on the Athlon 64 FX-55 processor.

Finally below you will see the final CPU-Z Verified screen shot taken while running at a front side bus speed of 275, again using a 10X multiplier. I used memory timings of 3-3-2-7 at 1T. Synthetic benchmarks ran fine at this setting however some gaming benchmarks did not.

Was the memory stable? I checked by running the SuperPi SSE2 32M test once stability testing had been completed using Memtest version 1.65. Determining Pi to 32 million places is a daunting task and will not complete if the system settings are not stable. The left screen shot shows the default value at stock settings with an overall time of 28 minutes and 53 seconds. At 270 HTT/10X the time was reduced to 26 minutes and 37 seconds (at 3-3-2-7 1T timings).