Prime95 test in between overclocks

[Flyingbig]

Ok recently i have been reading up on what to do and not to do when overclocking and the process seems to include overclocking in small increments then testing using Prime95 which is fair enough but i always considered Prime a 12 hour test so it seems useless to me. For instance if i am putting the fsb up in 5 increments and want to see if it is a stable or unstable overclock i hardly want to be sitting around for 12 hours so is there a better piece of software or is 10 mins of Prime 95 enough to indicate a stable overclock.

 

[dguy6789]

Here is what I do. I increase HTT if AMD64 or FSB if I am using a cpu from some other company whose name I cannot recall, by around 10, then I go and use Prime 95 for 5 mins. I keep going up until P95 crashes within 5 minutes. Then I go down by 5, and run P95 for 30 mins. If it crashes at that, then I go down another 5. Once I get it to a speed that goes beyond 30 mins, I just let it run overnight and if it is still on in the morning with no errors, then it is stable. Keep in mind to monitor temps as you do this.

 

[Duvie]

Here is what I do.....I usually look at similar chips and see what they are hitting at default vcore and I go straight for it. I look at the vcore if it the board undervolts a bit I try to get the vcore to be actual as close to stock....I only do 1 hour of testing in the prelim phases...I then move at 5 fsb increments...

Along with 1 hour of prime I would usually do a superpi32mb run and 5 passes of memtest...With A64s I highly recommend these as supplement test to run...With Intel chips it didn't quite make much of a difference for me...

One of the key things is to ISOLATE!!! so when I am oc testing a cpu I only OC the cpu.....I try to keep the ram out of it by running dividers, overly conservative tiings, and safe vdimm.....The A64 memory controller and how it relates to cpu freqeuncy is more sensitive....It could still error out even if the ram isn't the issue...therefore it needs to be tested...