| OCZ Technology PC2-8500 Reaper HPC CrossFire Edition 2GB Kit
|
Configuration (continued):
As a point of reference, the OCZ Technology PC2-8500 Reaper HPC CrossFire Edition 2GB Kit was tested head-to-head against two other high performance DDR2 memory kits:
» Crucial Ballistix DDR2 PC2-5300 CL 3 BL12864AA663. 16FD2
» Buffalo Firestix DDR2 PC2-6400 CL 5 FSX800D2C-1G
The image below shows the Buffalo Firestix and Crucial Ballistix to the left of the OCZ Technology PC2-8500 Reaper HPC CrossFire Edition 2GB Kit...
These modules are all rated at different speeds, but they have a history of good overclocking ability, and all use Micron D9 ICs. You will notice that I am not using an EPP capable motherboard, nor am I using a crossfire motherboard. The Abit IP35-e is a stripped down no frills motherboard based on the latest Intel P35 chipset. This chipset has been shown to be quite overclockable, and used in conjunction with the e6550 processor. it should allow me to easily reach memory speeds in excess of 1066MHz.
Testing:
For this review I will be using several free benchmark programs to analyze the performance of the three memory kits on hand. These include:
» Everest Ultimate Edition v. 4.00.976 (Cache and Memory Benchmark)
» 3DMark06 (1024x768, No AA)
» SiSoft Sandra Lite XII (Memory Bandwidth Test)
» Super Pi Mod 1.5 XS (32MB)
All settings were kept the same, except where instability required me to boost voltage or adjust timings. Memory timings and voltages are noted on the charts. The following tests were conducted with my e6550 PCU underclocked to 1869MHz (7x267) and the memory multiplier set to 1:2 for a memory speed of 1066MHz. I have also included benchmark results for the OCZ Reaper HPC memory at 533MHz (1:1) for comparison.
Everest Ultimate Edition v. 4.00.976:
Everest Ultimate Edition provides four sets of results: memory read speed (in MB/s), memory write speed (in MB/s), memory copy speed (in MB/s), and memory latency (in ns). Higher values are desirable for memory read, write, and copy speeds, while lower values are desirable for latency.
Note that despite the minor differences in timings, all 3 sets of modules performed nearly identically. I would expect this from modules that all use similar memory ICs (Micron D9). Also of note is that memory write performance doesn't seem to be affected by increases in memory speed. These numbers caused me to go back to the test system and re-run the 533Mhz tests with the same timings I used for the 1066MHz tests (5-4-4-15). The results were similar, virtually no increase in memory write performance despite doubling memory speed. It would appear that memory write speeds are more dependant on CPU and FSB speeds.
|
|
|
|
