I've been testing various settings above 450 since I "lost" my overclock at 450 about a few weeks ago. To break my system I added an extra 4gb of the same memory without taking notice that I may have to go through the whole tweak again. Previously, my system was 4GB and doing AUTO settings for voltages save for RAM at 1.8 and CPU for >48hr prime stable.
Previously 8x450 was stable with auto settings. I've discovered that while the system is running fine at 9x400, 8x450 causes random prime failures anywhere past 4hrs all the way to 16 hrs. Makes it a pain to test but making 24hr prime stable a new standard for myself of which my last set of settings seems to have gone 28hrs fine.
Test bench.
PC6400 5-4-4-12 1.8-2.1V Mushkin RAM. Timings relaxed to 5-5-5-18 but running oc'd at 450. RAM overclocking has turned out to be trickier than expected. Regardless of RAM voltage, 1.8 to 2.1V, I ended up failing around the similar times with prime. So I kept RAM voltage at 1.8V and played with only NB voltage, then subsequently FSB voltage. Results showed the NB is really the limiting factor on my system.
For speeds above 400, it seems FSB Termination Voltage and NB voltage go hand in hand. Increasing the NB to 1.40 from auto (min is 1.20) would net me stability up to the 8hr mark. Going to the next voltage up 1.55 logically then should make it stable but instead I failed prime testing in 7 mins!
This is where I started raising my FSB termination voltage. You can look it up online, but the general consensus is that this may need to be modified for FSB overclocking. 1.3V netted me 15hr prime stable. 1.4V FSB Termination netted me the latest result 28hr prime stable, so I've gone and tested my other set of Mushkin RAM which is approximately 9 months younger than the set that passed.
My final round of testing will check 8GB and see if I can get it stable and overclocked and hopefully provided a bit more insight into what goes into trying to run 4 sticks and 8gb systems.
Previously 8x450 was stable with auto settings. I've discovered that while the system is running fine at 9x400, 8x450 causes random prime failures anywhere past 4hrs all the way to 16 hrs. Makes it a pain to test but making 24hr prime stable a new standard for myself of which my last set of settings seems to have gone 28hrs fine.
Test bench.
PC6400 5-4-4-12 1.8-2.1V Mushkin RAM. Timings relaxed to 5-5-5-18 but running oc'd at 450. RAM overclocking has turned out to be trickier than expected. Regardless of RAM voltage, 1.8 to 2.1V, I ended up failing around the similar times with prime. So I kept RAM voltage at 1.8V and played with only NB voltage, then subsequently FSB voltage. Results showed the NB is really the limiting factor on my system.
For speeds above 400, it seems FSB Termination Voltage and NB voltage go hand in hand. Increasing the NB to 1.40 from auto (min is 1.20) would net me stability up to the 8hr mark. Going to the next voltage up 1.55 logically then should make it stable but instead I failed prime testing in 7 mins!
This is where I started raising my FSB termination voltage. You can look it up online, but the general consensus is that this may need to be modified for FSB overclocking. 1.3V netted me 15hr prime stable. 1.4V FSB Termination netted me the latest result 28hr prime stable, so I've gone and tested my other set of Mushkin RAM which is approximately 9 months younger than the set that passed.
My final round of testing will check 8GB and see if I can get it stable and overclocked and hopefully provided a bit more insight into what goes into trying to run 4 sticks and 8gb systems.