Hi All
So having built an x99 platform PC back in December, using the 5820k CPU, I finally stepped into the murky waters of overclocking in an attempt to push my already outstanding performance that little further. I originally had opted to leave it be, but that little what if in the back of my mind nagged me for the past 6 months and what the hay, its a K processor. Apologies if the following is long winded;
So basically I have used Intel Extreme Tuning Utility to compare my system against any identical systems and applied their settings to my setup. From that I was able to determine that my system would, manage a rise from the stock 33x multiplier up to the 43x multiplier, giving my system a healthy boost whilst appearing to be stable, stress tested using XTU for 20 mins and AIDA 64 for about 10 mins. Whilst I know these aren't extensive stress test, the impression I got whilst researching this process, was that a test that show consistently 100% CPU activity in AIDA 64 for a duration, should be indicative of being close to a stable OC. So, naturally my next step was to try to push it to 44x multiplier, at which point my PC shut down, I tried a combination of all cores on 44x, 3 cores on 44x and 3 on 43x, to squeeze some performance, but each time to the disapproval of my PC and the inevitable BSOD.
So at this point its safe to say my wall for OC is 43x all be it, currently untested in extensive stress testing. I thought I was happy enough with this (and depending on feedback I will be more than happy to settle) but the very nature of OC from what I can glean is to keep trying to push it. Having made these changes via XTU software, I read that the dynamic mode on the CPU multiplier and CPU voltage, induced by EIST and the power saving technologies could contribute to an unstable system as the dip in voltage can cause premature BSOD on otherwise 'stable' OC. I reset my system to stock, having to clear the CMOS after a failed OC attempt and decided to apply my OC setting via the BIOS, as I had read that it was a preferred way of doing things, not that I'm a purist.
I fixed the multiplier at 43x and set the voltage to 1.3 and noticed that temps rose, not dramatically, but by a few degrees, and that my key benchmark, Cinebench only gained nominal performance. As an extra measure I attempted to bring my cache ratio up from stock 30x to 42x - it crashed. 38x - it crashed. So have opted to leave it at stock 30. Though I didn't make any changes to any other settings to try to stabilise I high cache ratio. Finally I decided to call it a night, returning the power saving measures, enabling EIST and setting both the CPU and voltage to dynamic settings.
So we are now up to date on the state of my system and the questions I have are as follows;
1 - Have I missed a trick in terms of pushing the multiplier as far as I can. Is there a setting that I should use in conjunction with the multiplier to stabilise it when I increase it past 43x.
2 - I have read the temperature charts in the stickies and read around it, my temps reach 74 at full load and idle around 40, though as I say, intensive stress testing is yet to happen. I will report back the temps once I feel I have pushed the OC settings to a comfortable limit.
3 - Is there an argument for trying to up the cache ratio from 3.0Ghz. I have heard its better to have it close to the CPU settings. If so, what should I be tweaking in order to test the stability of an increased cache ratio.
4 - Which kind of trumps the above questions, should I be happy with the seemingly stable overclock i have on my 5820k, 3.3Ghz to 4.3Ghz, rejoice in the apparent increased performance, stress test it extensively for stability and enjoy?
This forum is awesome, without it, I would know even less that I do know about PC's and certainly wouldn't be able to ask any of the above. I appreciate any feedback / advice that you lot can give.
Thanks
So having built an x99 platform PC back in December, using the 5820k CPU, I finally stepped into the murky waters of overclocking in an attempt to push my already outstanding performance that little further. I originally had opted to leave it be, but that little what if in the back of my mind nagged me for the past 6 months and what the hay, its a K processor. Apologies if the following is long winded;
So basically I have used Intel Extreme Tuning Utility to compare my system against any identical systems and applied their settings to my setup. From that I was able to determine that my system would, manage a rise from the stock 33x multiplier up to the 43x multiplier, giving my system a healthy boost whilst appearing to be stable, stress tested using XTU for 20 mins and AIDA 64 for about 10 mins. Whilst I know these aren't extensive stress test, the impression I got whilst researching this process, was that a test that show consistently 100% CPU activity in AIDA 64 for a duration, should be indicative of being close to a stable OC. So, naturally my next step was to try to push it to 44x multiplier, at which point my PC shut down, I tried a combination of all cores on 44x, 3 cores on 44x and 3 on 43x, to squeeze some performance, but each time to the disapproval of my PC and the inevitable BSOD.
So at this point its safe to say my wall for OC is 43x all be it, currently untested in extensive stress testing. I thought I was happy enough with this (and depending on feedback I will be more than happy to settle) but the very nature of OC from what I can glean is to keep trying to push it. Having made these changes via XTU software, I read that the dynamic mode on the CPU multiplier and CPU voltage, induced by EIST and the power saving technologies could contribute to an unstable system as the dip in voltage can cause premature BSOD on otherwise 'stable' OC. I reset my system to stock, having to clear the CMOS after a failed OC attempt and decided to apply my OC setting via the BIOS, as I had read that it was a preferred way of doing things, not that I'm a purist.
I fixed the multiplier at 43x and set the voltage to 1.3 and noticed that temps rose, not dramatically, but by a few degrees, and that my key benchmark, Cinebench only gained nominal performance. As an extra measure I attempted to bring my cache ratio up from stock 30x to 42x - it crashed. 38x - it crashed. So have opted to leave it at stock 30. Though I didn't make any changes to any other settings to try to stabilise I high cache ratio. Finally I decided to call it a night, returning the power saving measures, enabling EIST and setting both the CPU and voltage to dynamic settings.
So we are now up to date on the state of my system and the questions I have are as follows;
1 - Have I missed a trick in terms of pushing the multiplier as far as I can. Is there a setting that I should use in conjunction with the multiplier to stabilise it when I increase it past 43x.
2 - I have read the temperature charts in the stickies and read around it, my temps reach 74 at full load and idle around 40, though as I say, intensive stress testing is yet to happen. I will report back the temps once I feel I have pushed the OC settings to a comfortable limit.
3 - Is there an argument for trying to up the cache ratio from 3.0Ghz. I have heard its better to have it close to the CPU settings. If so, what should I be tweaking in order to test the stability of an increased cache ratio.
4 - Which kind of trumps the above questions, should I be happy with the seemingly stable overclock i have on my 5820k, 3.3Ghz to 4.3Ghz, rejoice in the apparent increased performance, stress test it extensively for stability and enjoy?
This forum is awesome, without it, I would know even less that I do know about PC's and certainly wouldn't be able to ask any of the above. I appreciate any feedback / advice that you lot can give.
Thanks