| May 13, 2007 |
Testing: Test Bed, Procedure, and Play-by-play
|
Test Bed: Vista Gaming Rig
- Intel E6400 Core2Duo
- ECS NF650iSLIT-A 775 Motherboard
- CoolIT Freezone CPU Cooler
- 1 x Western Digital 160GB SATA HD
- 1 x Foxconn 8800 GTS 320OC
- 2 x 1GB Crucial Ballistix "Tracer" 8500 DDR2 Ram
- 2 x 1GB Crucial Ballistix 6400 DDR2 Ram
|
|
Testing Procedure
Memory overclocking is a tedious process of trial and error. Little by little, I increased the memory speed in 5mhz increments, using memtest86+ to check for errors. If memtest stayed error free for a minimum of 2 hours, I booted into Windows and opened up Orthos. Orthos is a modification of Prime95 which runs 2 itterations of Prime 95 to utilize both cores of a dual core processor. Prime95 was originally developed to calculate Mersenne Prime numbers in an early attempt at distributed computing. Because of its ability to stress even the most powerful cpus, Prime95 has become a staple stability indicator in the overclocking world. For my purposes, I used the Blend torture test, making sure that my memory maintained stability for at least 4 hours before deeming it stable at a given speed.
Stability Testing Software:
- Memtest86+
- Orthos Prime95
Benchmarking Software:
- Everest
- PCMark2005
- SiSoft Sandra 2007
- SuperPi
- Games
Play by Play
Lately, I've run into a number of problems in testing memory, thanks to the positively dismal performance of the 650i motherboard ECS supplied for our Get Ready for Vista Gaming Rig. Testing Crucial's latest 8500DDR2 modules was no exception. When I tried to run this particular memory at its native speed (DDR1066) I not only could not boot into windows, I could not even boot into the BIOS. Frustrated, I cleared the CMOS and started over. Try as I might however, there was nothing I could do to get this memory to boot. Finally, I went over to ECS's website to see if there was a BIOS ubdate available. Sure enough there was, but the BIOS notes did not indicate any changes with regards to the way memory was handled. Desperate, I installed it anyway. With fingers crossed, I set the memory to the speed on the label and pressed the power button. Sure enough, it worked, but to my chagrin, I went back into the BIOS only to discover that ECS had nerfed the memory voltage options. No longer could I boost voltage to 2.08V, now I could only go as high as 1.95V, .25V less than Crucial's recommended setting. Fingers still crossed, I booted into windows to run Orthos for stress testing. Miraculously, Crucial's latest memory was totally stable at 1066 speeds, despite my motherboard's inability to supply proper voltage! This is a testament to Crucial's excellent memory, that it prevailed against the lack-luster performance of ECS's 650i. Unfortunately, because I could not supply these modules with any more volts, I could not get even a modest overclock from them without a loss of stability, though I have no doubts that with a proper motherboard, these modules would REALLY shine.
| Speed |
Timings |
Memtest86+ Stable? |
Prime95 Stable? |
|
DDR 800 (400mhz) |
4-4-4-12 |
Yes |
Yes |
|
DDR 1066 (533mhz) |
5-5-5-15 |
Yes |
Yes |
|