- last edited on by
ROGBot
I have a question. My system had become completely unstable and any attempt at even a minor OC got instant BSOD and after countless attempts I gave up and am ready to start a new build
However today as a last resort I messed with BCLK and upped it to 100.5 and using my previously stable OC setting for 4.2ghz I got a stable 1 hour bench with OCCT and several passes of realbench
Question is why? how has increasing BCLK stabilized my OC and is it an indication that my RAM is failing? (just a guess)
However today as a last resort I messed with BCLK and upped it to 100.5 and using my previously stable OC setting for 4.2ghz I got a stable 1 hour bench with OCCT and several passes of realbench
Question is why? how has increasing BCLK stabilized my OC and is it an indication that my RAM is failing? (just a guess)
srmojuze wrote:
BCLK increase is putting more "juice" into the whole system? 🙂 But yeah maybe push up the BCLK further, drop the multiplier and loosen RAM timings?
Well this is exciting . Have tried your method, dropped back multiplier from 46 to 44 and upped BCLK to 105 and Vcore is 1.45 and core speed is 46mhz
I havnt touched RAM timings mainly cause Im a newb at this and wouldn't know how. In any case 45 mins of Prime and several passes of realbench and its stable. I guess I will keep upping BCKL until something breaks.
Amazing though I couldnt even get step up 4.2 in the bios to work before playing with BCLK
zoran wrote:
I have a question. My system had become completely unstable and any attempt at even a minor OC got instant BSOD and after countless attempts I gave up and am ready to start a new build
However today as a last resort I messed with BCLK and upped it to 100.5 and using my previously stable OC setting for 4.2ghz I got a stable 1 hour bench with OCCT and several passes of realbench
Question is why? how has increasing BCLK stabilized my OC and is it an indication that my RAM is failing? (just a guess)
Digital logic is an an abstraction on top of analog circuitry... We think in 0's and 1's, but the reality is there is an extraordinarily complex and in analog terms noisy system running underneath. Every collection of electrical components has inherent frequency specific attenuation, filtering, etc... Then you have cross-talk and beat frequencies of components operating against each other asynchronously...
A stable system is able to operate over the full range of these inferences, resonances and noise injections, but if you were able to probe every point on the system and Tom Cruise, Minority Report style swipe your way through them, you'd see ripples, wiggles, stutters and near failures all over the place in terms of the electrical signal conforming to the requirements of either consumer(s) or producer(s) of a signal.
If you had time and measurement capability you could run the system over an infinite range and permutation of settings and if you graphed them all out you'd see a pretty tree of standard deviations at each permutation representing how much "margin" any given setting that worked had.
It would NOT be a flat line, you'd likely see resonance in these measurements with margin increasing and decreasing as you went up and down the frequency spectrum in a quasi-sinusoidal fashion.
Short version: very complex system, it has frequencies it likes and those it doesn't and some it doesn't like are lower than those it does.
