How to separate RAM during OC process
How to separate RAM during OC process
Hi, I wanted to ask about isolating RAM for testing the OC to check if the RAM is causing instability.
I know this is done by covering the CPU multiplier so the CPU overclock stays below the stock frequency. This way, the test results won’t be affected by the CPU and FSB limits, right?
For instance, my current E5450 runs at 440FSB (3,96Ghz) with a x9 multiplier. If I reduce the multiplier to x5 or x6, it would drop below the stock 3Ghz, allowing me to test just the RAM.
I have G.Skill DDR2 1066Mhz modules (2x2GB, F2-8500CL5D-4GBPK) and I experience random errors in timing without any BSODs. The DRAM voltage is 2.22V (stock max is 2.1V) and the timings are 6-6-6-18 in case (stock 5-5-5-15), but I’m struggling to stabilize it at 1057Mhz (FSB strap 333, which is lower than the rated 1066Mhz). The MCH is 1.4V, which should be sufficient, though sometimes errors appear after 30 minutes or more, and sometimes after just a few hours.
I managed to get stable results with a strap 400 and 5-5-5-15 at 2.2V @ 880Mhz for over 24 hours, but not with the next one – 1057Mhz and strap 333.
I’m trying to isolate the RAM so I can test it independently for errors.
By the way, I’m new to OCing; I only did it on an E8500 before and it was straightforward. It ran perfectly for the past four years with auto settings, and I only adjusted the DRAM’s maximum stock voltage. But this E5450 is much more complicated.
Board: ASUS P5Q Deluxe
Any suggestions?
Yesterday night I attempted the most stable configurations I could achieve again. The previous memtest passed 28 tests on a 1:1 880Mhz system. After running the test once more, only 7 succeeded, leaving me uncertain about what was happening and why so many random failures occurred. Now I’m running memtest again, this time lowering the multiplier from 9 to 6, reducing the CPU clock from 3.96GHz to 2.64GHz (the stock is 3GHz). I plan to let it run for about 24 to 36 hours and then share the results.