can an oced cpu cause BSOD at idle?

[Magic Carpet]

Different software makes different load on hardware. I haven't seen a single piece of software able to utilize hardware fully.

So I say this:

Test with the actual programs/games, you use... and forget everything else people are saying ;-p

For example, I had a Sempron 140 unlocked into two cores, that were only stable in specific software. So I just used it within them. People tend to generalize too much but issues should be looked at on a case by case basis.

If it crashes when idle... hell, just powercfg.cpl and increase minimum cpu state to 50% or smth. You don't care about power consumption right? I take it, there is no K10Stat equivalent for Intel. I had a similar issue with AMD so I just upped the voltage for that p-state specifically.

 

[bryanW1995]

Idle stability is often just as hard as or harder to achieve than "100%" load stability. Don't fret about it, just bump up your vcore a notch or two. I always do that as standard practice when I OC, anyway.

 

[blackened23]

Toyota, if you're using LLC and 1.22vcore at load your vcore could vary wildly while idle. I'm not sure from your wording whether you're using LLC or not, but if you are:

it sounds like a load line calibration issue. What I would do is disable LLC and use a 24/7 vcore. I have a P8Z68 pro Gen3 and LLC is very finicky, last time I tried it my vcore would range from 1.1vcore up to 1.5. It was extremely hard to dial in, and if my load vcore was at a good level my idle vcore would be too low. I said screw it and am using a 24/7 vcore, LLC disabled.

Secondly, why do have C1 disabled? You run at 4.5ghz 24/7 with no downthrottling? You shouldn't be idling at your turbo speed Typically you idle at 1600 mhz on a 2500k , which is ideal.

 

[toyota]

its actually 1.23 and I tried to correct all my posts to reflect that. and as I said in my other thread, if I change the voltage and multi on this board then power saving features are disabled. even if I could get it back to 1.6, the voltage would still be at 1.23 since I have no offset options.

I leave LLC on auto and according to cpu z, it stays right at 1.23 at idle and load.

 

[chimaxi83]

OP, I used to use a G43 board and had this same exact issue. Unfortunately all I could do was raise vcore slightly, don't remember to what.

As you know, this board lacks a lot of fine tuning options as far as all the voltages we like to play with, so really the only option was a slight bump up. I think it was 0.02V or something close to that, stable under load and my idle BSODs went away.

On a side note, this issue started after I updated my BIOS. Almost everything was the same except for it was now downclocking CPU as it approached TDP, something I could do nothing about (because the board lacked any power draw increase option). I downgraded BIOS and from then on I had the issue you have.

 

[VirtualLarry]

I had problems with my E2140 CPUs, overclocked from 200 to 400FSB, from 1.6Ghz to 3.2Ghz. I could run 24hour stress tests all I wanted, and it would pass just fine.

But leave the machine running, for a week or more, and it would spontaneously reboot.

 

[Avalon]

Load Line Calibration could explain that.

Either way, 1.22v is pretty low for 4.2Ghz.

Was just about to come in and say check your LLC setting. You might be getting more vdroop at idle than under some decent load. Turn it from either auto to disabled or enabled, and see what kind of voltage your system reports for each setting.

 

[toyota]

Was just about to come in and say check your LLC setting. You might be getting more vdroop at idle than under some decent load. Turn it from either auto to disabled or enabled, and see what kind of voltage your system reports for each setting.

idle or load it was always at 1.23 though according to cpu z. for vdroop settings all I have is auto or high.

 

[Kenmitch]

idle or load it was always at 1.23 though according to cpu z. for vdroop settings all I have is auto or high.

Hard locked vcore and no load shouldn't ever bsod at idle I'd think. Did you change anything in your rig before the bsods started? Drivers? Software? Maybe a recent windows update is conflicting? Maybe roll back to a restore point your sure wasn't bsoding.

Tip: Maybe edit your initial post clearly stating that you have your overclock and vcore hard locked in bios. Might save you the trouble of having to repeat it over and over....I know how you get grumpy sometimes

You said you were gonna run at stock speed for awhile....How goes that?

 

[toyota]

Hard locked vcore and no load shouldn't ever bsod at idle I'd think. Did you change anything in your rig before the bsods started? Drivers? Software? Maybe a recent windows update is conflicting? Maybe roll back to a restore point your sure wasn't bsoding.

Tip: Maybe edit your initial post clearly stating that you have your overclock and vcore hard locked in bios. Might save you the trouble of having to repeat it over and over....I know how you get grumpy sometimes

You said you were gonna run at stock speed for awhile....How goes that?

I cant think of anything that I changed around the day that it started.

no BSODs in the 2 days of running at default cpu settings. I will give overclocking a go again next week once I am positive that nothing else is causing it.

 

[Kenmitch]

I cant think of anything that I changed around the day that it started.

no BSODs in the 2 days of running at default cpu settings. I will give overclocking a go again next week once I am positive that nothing else is causing it.

Maybe you could eliminate your memory while your testing the rig under stock setting. This is from the sticky in the forum.

HCI memtest is superior to LinX, Prime (large or small), and memtest86+ for determining memory stability.

• Launch one instance per thread supported by CPU (8 instances of HCI memtest for 2600K)
• Set each HCI instance to use an appropriate fraction of the memory...16GB on a 2600K means each HCI instance (there will be 8 instances) are to use 2048 MB.
• Stability Criterion: Must pass 200% coverage minimum, passing 1000% coverage is preferred (considered gold standard)

 

[toyota]

Maybe you could eliminate your memory while your testing the rig under stock setting. This is from the sticky in the forum.

HCI memtest is superior to LinX, Prime (large or small), and memtest86+ for determining memory stability.

• Launch one instance per thread supported by CPU (8 instances of HCI memtest for 2600K)
• Set each HCI instance to use an appropriate fraction of the memory...16GB on a 2600K means each HCI instance (there will be 8 instances) are to use 2048 MB.
• Stability Criterion: Must pass 200% coverage minimum, passing 1000% coverage is preferred (considered gold standard)

I ran that to almost 200% on all of my ram so I am pretty sure ram is not the issue here.

 

[Rvenger]

its actually 1.23 and I tried to correct all my posts to reflect that. and as I said in my other thread, if I change the voltage and multi on this board then power saving features are disabled. even if I could get it back to 1.6, the voltage would still be at 1.23 since I have no offset options.

I leave LLC on auto and according to cpu z, it stays right at 1.23 at idle and load.

I understand that it disables everything but it is still selectable in the bios after you set a certain vcore. Just having the options enabled could be causing issues with stability. I used to be a mod over at MSI forum, let me check with some of my sources and see if I can get you an answer.

 

[Rvenger]

What version Bios are you running?

 

[toyota]

okay I give up on overclocking that way. I have tried adjusting every setting and giving it more voltage but nothing will stop the eventual crashing. again I can be 100% stable for hours of gaming or stress testing but it will crash at some point just doing nothing but idling. also it makes no sense that before I started this thread it was fine for 3 weeks.

I am now just overclocking to 4.2 using turbo. to be clear I have set 4.2 for all the turbo settings and left voltage on auto. I never have any stability issues doing that but as I have said before, the voltage is a bit too high for my liking. I guess that I will have to just deal with it though as there seems no other way to oc on this setup.

 

[fastamdman]

You are telling me there are no manual voltages for that board, or offest voltages, doubt it. How much vcore is it pushing for 4.2? As long as your temps are fine and its sub 1.4 at load you are ok.

 

[toyota]

You are telling me there are no manual voltages for that board, or offest voltages, doubt it. How much vcore is it pushing for 4.2? As long as your temps are fine and its sub 1.4 at load you are ok.

please read through the thread as pretty much any question or suggestion you will have has likely already been discussed. there are no offset options. and as I have said, I used 1.23 for 4.2 and its 100% stable in games and stress testing but after 3 weeks it started crashing randomly at idle. I went back to stock settings and no crashes for days but if i try to oc with the above settings it will crash at some point idling. more voltage up to 1.28 does not address it. and if more voltage than that is required for it to be stable at idle for just 4.2 then I might as well just oc with turbo which uses just over 1.3 but will let me idle at 1.6 and low voltage.

 

[Borealis7]

n/m wrong thread

 

[toyota]

lol, you deleted it.

 

[Rvenger]

please read through the thread as pretty much any question or suggestion you will have has likely already been discussed. there are no offset options. and as I have said, I used 1.23 for 4.2 and its 100% stable in games and stress testing but after 3 weeks it started crashing randomly at idle. I went back to stock settings and no crashes for days but if i try to oc with the above settings it will crash at some point idling. more voltage up to 1.28 does not address it. and if more voltage than that is required for it to be stable at idle for just 4.2 then I might as well just oc with turbo which uses just over 1.3 but will let me idle at 1.6 and low voltage.

This happened to me as well. Do you think you can list off your bios settings? I used to have an overclocked MSI Sandybridge setup.

 

[toyota]

well my overclocking adventures are done as I am just going to use turbo for 4.4. I ran maximum Intel Burn Test with basically stock settings except all 4 cores turboed to 4.4 and it passed. voltage stayed at around 1.312 during the run with the highest temp on any core hitting 71 C. most of the time it was in the mid to upper 60s though. thats not bad since I am only using just the pull fan on my A70. heck I am even using the low fan setting too. during games I am hitting low 50s most of the time for a max temp. also during games that use all four cores the highest voltage was a quick peak of 1.34 but usually stayed at 1.32.

 

[fastamdman]

This is the way you should have overclocked to begin with. I would bet the bsod at idle has something to do with not resuming from sleep due to cpu pll over voltage being enabled.

Chips don't just randomly bsod at idle, I have NEVER heard of this, especially with sandy bridge. If a system is stable at load it should be stable at idle, especially if the vcore isn't fluctuating and is constant. Now if you had a failing power supply that wasn't always providing a smooth power current, sure.

Either way the best way to overclock imho is through the turbo settings. Granted most people do it with offset voltages but if you don't have them then obviously you have to run auto. 1.312 load vcore is not very high at all, especially for 4.4. Hell for some chips it wouldn't even be enough to be stable.

Out of curiosity what kind of gflops are you getting and how long are you "stable" with intel burn test?

 

[Rvenger]

I agree, I didn't have much success through the multiplier on the MSI board. The turbo overclocking was way more effective.

Toyota, your voltage probably won't exceed 1.35v per intel max spec as long as Vcore load line is disabled.

 

[toyota]

I just did the standard 5 passes on maximum but with 16gb that took a while. yes I could run it longer but if that is stable then I will have no issues with games. the highest temp I have hit in a game is 53C and again that is just using only one of my A70 fans and on low rpm too.



hosting images

 

[fastamdman]

Meh you are no where near system stability and could very easily crash and or bsod while gaming or doing anything cpu intensive. Judging by your posts, and that screen shot though I doubt you do anything to intensive like folding at home or anything along those lines.

The difference between 5 tests and 25+ tests are HUGE. I would simply let the test run over night while you are sleeping and if you wake up to a BSOD than you know.

I don't understand why people want half ass stable systems. Why not simply take the time to ensure it's right instead of getting another bsod. You don't want to make another thread next week asking "can 1.312 vcore for 4.4ghz cause a bsod when playing counter strike source and downloading a movie" lol. Just take the time to do it right.

Edit : Also manually set the thread count instead of using auto with IBT.