Overclock fails when gaming

[Terminashunator]

I have a 3570k at 4.2 on my Asrock Z77 Pro4-M. CPUz says 1.192 - 1.2v

My settings in BIOS are Offset -.05v, +.05 Turbo 50% LLC, 1.99 PLL

 

I ran Prime95, Furmark, AND Intel XTU along with Watching a movie in VLC for 4 hours last night to absolutely test the bejeezus out of the overclock. Temps were fine at 65-68 on each core. No sweat.

Today, I was just using my computer. I play World of Warcraft. Did some dungeons, questing, arena, and LFR. Five minutes after I step into a 25m guild raid, WoW crashed. What gives? I hurriedly dropped the OC to 4.0, same voltages and raided the entire night no more crashes. 

 

I tried to test absolutely everything i could last night, why did the OC fail? Do I need to give more Turbo Voltage? My board does not allow for Constant Voltage, does anyone have some advice? 

[Terminashunator]

Up voltages or down clocks. Crashing means it's not a stable OC. Super simple stuff

I know, right? I just don't understand how it can be stable in my tests, and I mean really stable, but when I actually USE the computer I get program crashes. 

I have a feeling the motherboard doesn't like overclocking. I had it boot at 4.4 but the vCore was >1.3 and I wasn't comfortable with that. But it wasn't stable with 1.3v and anything over 4. Sadface. 

[-rascal-]

1) 4 hours of stress / stability testing is not enough. Generally should be 8 hours (IMO I prefer 24 hours).

2) Furmark and VLC movie for CPU stress test?? Furmark is for graphics cards. Nothing to do with CPU. Plus, watching a movie clocks the GPU to 2D speeds, not the full 3D speeds.

3) Overclock on component at a time. Don't do CPU + GPU + RAM all at the same go. If sheet happens, how are you suppose to know which one is causing the instability.

What CPU cooler are you using for your 3570K?

[Terminashunator]

1) 4 hours of stress / stability testing is not enough. Generally should be 8 hours (IMO I prefer 24 hours).

2) Furmark and VLC movie for CPU stress test?? Furmark is for graphics cards. Nothing to do with CPU. Plus, watching a movie clocks the GPU to 2D speeds, not the full 3D speeds.

3) Overclock on component at a time. Don't do CPU + GPU + RAM all at the same go. If sheet happens, how are you suppose to know which one is causing the instability.

What CPU cooler are you using for your 3570K?

I ran a stress test for 4 Hours, then I ran BOINC for 12 more. I didn't want to keep testing it that much more instead of me doing work. And don't worry, I'll run P95 for 12 hours of fully loading the CPU. 

 

P95, IntelXTU, Furmark, AND VLC. I was doing a general system stability test, not just the CPU. 

 

I'm only overclocking my CPU. Ram is Stock XMP 1600.

 

Hyper 212+, as it says in my signature. Sorry I didn't put that in my post. I'll edit that.

 

 

Up voltages or down clocks. Crashing means it's not a stable OC. Super simple stuff 

 

I looked at your beautiful build log "[project] direct" and saw your beautiful work, along with an Asrock motherboard. Do you have any advice concerning UEFI settings? 

[BlackOutCrisis]

I know, right? I just don't understand how it can be stable in my tests, and I mean really stable, but when I actually USE the computer I get program crashes. 

I have a feeling the motherboard doesn't like overclocking. I had it boot at 4.4 but the vCore was >1.3 and I wasn't comfortable with that. But it wasn't stable with 1.3v and anything over 4. Sadface. 

I could have just been a random WoW crash? Software has it's issues also. I suggest keeping your overclock and seeing if you can reproduce the crash and if so then you can say it may be your CPU. You should be fine at those voltages though, I run my 3570k@ 4.3Ghz 1.76v - 1.184v.

 

If you can reproduce the crash then some things to look out for are double checking your RAM timings in BIOS and turning PLL voltage to auto. Also try setting LLC to auto because it can get into a argument with your vdroop.

[Terminashunator]

I could have just been a random WoW crash? Software has it's issues also. I suggest keeping your overclock and seeing if you can reproduce the crash and if so then you can say it may be your CPU. You should be fine at those voltages though, I run my 3570k@ 4.3Ghz 1.76v - 1.184v.

 

If you can reproduce the crash then some things to look out for are double checking your RAM timings in BIOS and turning PLL voltage to auto. Also try setting LLC to auto because it can get into a argument with your vdroop.

I doubt it was a "random" crash. It's the same crashes i was getting when I was pushing 4.4 @ 1.3v. Just random crashes that go away when the overclock is dropped to 4ghz.

 

Ram is at XMP 1600 stock, and the voltage is manually set to the stock 1.8 to prevent the board from auto voltage controlling to high limits. 

 

Though, I will go try them at auto to see what happens. It's cold enough in here, shouldn't make it dangerously hot. 

[bacosta1993]

Run stress test for 24 hours. I had a Blue Screen 16 hours into prime95. It is unusual but it happens. 

[Terminashunator]

Run stress test for 24 hours. I had a Blue Screen 16 hours into prime95. It is unusual but it happens. 

This is my main rig. So I'm not sure I have the luxury to do 24 hours of it. Perhaps 12h for the time being I have time for. 

But you're right, I should test more thoroughly. 

[Terminashunator]

I could have just been a random WoW crash? Software has it's issues also. I suggest keeping your overclock and seeing if you can reproduce the crash and if so then you can say it may be your CPU. You should be fine at those voltages though, I run my 3570k@ 4.3Ghz 1.76v - 1.184v.

 

If you can reproduce the crash then some things to look out for are double checking your RAM timings in BIOS and turning PLL voltage to auto. Also try setting LLC to auto because it can get into a argument with your vdroop.

Checked ram times, XMP3 1600. Though I did set the PLL, LLC, to Auto. Thank you! Voltage is not 1.20 - 1.28v at peak.

 

 

Not sure how similar the BIOS will be tbh, But all the setup in mine was pretty much the same as any other board mate   I don't really dabble much further into the BIOS than OC settings and occasionally raid settings. I run fans and all that from 12v hubs and usually have a pretty straightforward setup 

 

 

 

A 1.3v Vcore is fine unless your running into thermal limits mate   1.4v is the limit before you start running into degradation of the CPU, 1.5v absolute max if your not concerned about the chip lasting a few years less 

 

You'll usually run into that issue where games and programs crash but stress tests wont, at that point just up your voltage and use games as your stress test until you find a point at which your stable   Also run AIDA64 as your synthetic bench for 8 hours instead of 4 

 

EDIT- Thanks for the compliments about Direct too mate   I appreciate the feedback I get on rigs   Direct is actually up on Ebay right now and about to start work on Project Midnight so keep an eye out on that 

 

I was being cautious on the VCore. The last chip I overclocked was a Q9550 and that was FSB overclocking. Dead simple. So i'm being dainty about this chip. I want it to last 5 years+ if I can. Thermals haven't been a problem so I suppose I can just tweak it up.

 

I'll also download AIDA64 and run that tomorrow after 10 hours of Prime as I sleep. 8 Hours of AIDA shouldn't be a problem. 

[bacosta1993]

This is my main rig. So I'm not sure I have the luxury to do 24 hours of it. Perhaps 12h for the time being I have time for. 

But you're right, I should test more thoroughly. 

Just reread your post and I am wondering if WOW crashed or your computer?

If WOW crashed it might be your driver, a bug, WOW or something interfering with your game such as anti virus software.  You should also try playing other games before you change your overclock settings(only if it is WOW that crashes).

[Leonard]

I have a 3570k at 4.2 on my Asrock Z77 Pro4-M. CPUz says 1.192 - 1.2v

My settings in BIOS are Offset -.05v, +.05 Turbo 50% LLC, 1.99 PLL

 

I ran Prime95, Furmark, AND Intel XTU along with Watching a movie in VLC for 4 hours last night to absolutely test the bejeezus out of the overclock. Temps were fine at 65-68 on each core. No sweat.

Today, I was just using my computer. I play World of Warcraft. Did some dungeons, questing, arena, and LFR. Five minutes after I step into a 25m guild raid, WoW crashed. What gives? I hurriedly dropped the OC to 4.0, same voltages and raided the entire night no more crashes. 

 

I tried to test absolutely everything i could last night, why did the OC fail? Do I need to give more Turbo Voltage? My board does not allow for Constant Voltage, does anyone have some advice? 

You can't have an offset voltage setting for any given overclock with both - and + offset, it is one or the other. The more you raise your frequency the harder it is to use a - offset voltage, + offset voltage is required. Each CPU is different and may need less or more voltage to be stable. Put it back at 4.2 and apply +0.020 offset voltage, put LLC at regular and CPU current capability at 100% and see if you still crash. 

[MeltyMoon]

The thing is when gaming, your other components are also drawing more power! When you're just stress testing your CPU only that part is drawing power and is under load. However when other parts are thrown into the mix, it can destabilize the entire computer (if parts are overclocked).  

[Terminashunator]

Just reread your post and I am wondering if WOW crashed or your computer?

If WOW crashed it might be your driver, a bug, WOW or something interfering with your game such as anti virus software.  You should also try playing other games before you change your overclock settings(only if it is WOW that crashes).

Flash player also crashes. But you're right, it could be a driver. Though all my drivers are up to date with the ones on the manufacturer's websites. 

 

Just reread your post and I am wondering if WOW crashed or your computer?

If WOW crashed it might be your driver, a bug, WOW or something interfering with your game such as anti virus software.  You should also try playing other games before you change your overclock settings(only if it is WOW that crashes).

No, WoW is just crashing itself. The Windows session does not crash. Just WoW, Flash Player, sometimes other apps also hang and crash. When I'm not overclocking I don't see a crash, really ever. I'm only running MSE for Antivirus at the moment, and I've tried some other games (Borderlands 2, Dirt 3, Hearthstone) and they have a mixed bag of results. 

 

You can't have an offset voltage setting for any given overclock with both - and + offset, it is one or the other. The more you raise your frequency the harder it is to use a - offset voltage, + offset voltage is required. Each CPU is different and may need less or more voltage to be stable. Put it back at 4.2 and apply +0.020 offset voltage, put LLC at regular and CPU current capability at 100% and see if you still crash. 

Yes, I can have a -Offset and a +Additional Turbo Voltage. The (-) is for a low Idle voltage, so that when it's not bein g taxed it doesn't waste power and make a lot of heat. I've run 8 hours of P95 with -.05/+.04, LLC at Auto (Per Marto's suggestion). Fiddling with the offset does incrase the voltage, however adjusting the (+) Turbo also increases the maximum voltage while in a Turbo state while not raising the Idle voltage.

 

 

The thing is when gaming, your other components are also drawing more power! When you're just stress testing your CPU only that part is drawing power and is under load. However when other parts are thrown into the mix, it can destabilize the entire computer (if parts are overclocked).  

Every other part is stock settings, to reduce conflicts and to reduce varibles to one part. I understand that a CPU benchmark only tests the CPU, that's why I ran a GPU test along with a second CPU test at the same time. So i was loading the CPU, and GPU.

[MegaDave91]

4 hours is nothing. It can still crash at 4 hours and much beyond as well. Start stress testing a few hours before you go to bed and let it run overnight. Keep it running when you wake up too.

 

Stress testing requires you to give up your computer and leave it in an idle (not using it) state for quite a while. 

[flibberdipper]

I feel I'm the only one here that overclocks their netbook.

 

I take the core to 1.9GHz stable and the GMA 945 to 400MHz (up from 166MHz). Runs stable in Vista Ultimate, but not in my poor old XP Home install.

[Leonard]

 -.05/+.04, LLC at Auto (Per Marto's suggestion).

 

This could be your problem. The offset voltage seems overkill for a 4.2Ghz overclock to me. -.05 means in BIOS terms, -0.050 and at 4.2ghz that is too low of a voltage. I think you want to use -0.005 and not -0.050. Can you put a screen shot of your offset voltage figures?

[Terminashunator]

4 hours is nothing. It can still crash at 4 hours and much beyond as well. Start stress testing a few hours before you go to bed and let it run overnight. Keep it running when you wake up too.

 

Stress testing requires you to give up your computer and leave it in an idle (not using it) state for quite a while. 

ran P95 for 12 hours, going to run AIDA64 tonight.

 

 

I feel I'm the only one here that overclocks their netbook.

 

I take the core to 1.9GHz stable and the GMA 945 to 400MHz (up from 166MHz). Runs stable in Vista Ultimate, but not in my poor old XP Home install.

Check my Sig, I overclock my Atom from 1.6 to 2.0Ghz, GMA 3150 to 400mhz as well.

 

 

This could be your problem. The offset voltage seems overkill for a 4.2Ghz overclock to me. -.05 means in BIOS terms, -0.050 and at 4.2ghz that is too low of a voltage. I think you want to use -0.005 and not -0.050. Can you put a screen shot of your offset voltage figures?

I'll post a screen shot shortly. but no, it is -0.050 for the offset, and in Turbo it get 0.040v more when it's in the turbo state. I followed the Asus guide  and a Asrock guide so I could understand how Offset control works. With offset, I set the Minimum voltage that the CPU gets at idle. The additional turbo is when it's hitting its peak frequency, it gets an additional boost of power. 

But I will post a screen shot of my BIOS shortly. 

[flibberdipper]

Check my Sig, I overclock my Atom from 1.6 to 2.0Ghz, GMA 3150 to 400mhz as well.

I'm not the only one who overclocks Atoms! 

[Leonard]

ran P95 for 12 hours, going to run AIDA64 tonight.

 

 

Check my Sig, I overclock my Atom from 1.6 to 2.0Ghz, GMA 3150 to 400mhz as well.

 

 

I'll post a screen shot shortly. but no, it is -0.050 for the offset, 

Like i said, this could be your problem. I think you should decrease the -0.050 to -0.030 and see if you still crash. Remember a guide is just that, a guide and you may need a different approach or use different figures to get the same overclock. Waiting on your screen dumps.

[Terminashunator]

Like i said, this could be your problem. I think you should decrease the -0.050 to -0.030 and see if you still crash. Remember a guide is just that, a guide and you may need a different approach or use different figures to get the same overclock. Waiting on your screen dumps.

I'm sorry I took so long. I've had a rough day. 

[Leonard]

I'm sorry I took so long. I've had a rough day. 

No problem. Drop the CPU Offset Voltage to -0.030 and then play WOW and check for crashes. If it still crashes drop it even further to -0.020v.

[Terminashunator]

No problem. Drop the CPU Offset Voltage to -0.030 and then play WOW and check for crashes. If it still crashes drop it even further to -0.020v.

I actually was doing that yesterday. That's what took up ~5 hours. I was playing it with the upped settings there, and I didn't have a crash. 

Overnight I rain AIDA64 with no problems, though temperatures are 15c higher. Maxing at around 78c. Which is fine, but I'm not needing to push it any higher, you know? 

Sorry for only 7 hours. I didn't get as much sleep as usual.

[Leonard]

Sorry for only 7 hours. I didn't get as much sleep as usual.

Glad it worked out and no problem with the time you took to get back to me. If you asked me you were fast. It took me 3days to get my overclocks stable. The increase of temperatures would be normal but what is your cooling solution? Another thing, when doing your stress testing, un-check the GPU and RAM testing and just do the CPU and FPU if you have only overclocked the CPU.

[Terminashunator]

Glad it worked out and no problem with the time you took to get back to me. If you asked me you were fast. It took me 3days to get my overclocks stable. The increase of temperatures would be normal but what is your cooling solution? Another thing, when doing your stress testing, un-check the GPU and RAM testing and just do the CPU and FPU if you have only overclocked the CPU.

Hyper 212+ with Push Pull, though I don't think the second fan helps much with such loose fin density. 

And you're right, I should have de-selected the GPU/ RAM, hoever the computer didn't crash when just the CPU was being tested before, and I was just being overcautious. 

[Leonard]

Hyper 212+ with Push Pull, though I don't think the second fan helps much with such loose fin density. 

And you're right, I should have de-selected the GPU/ RAM, hoever the computer didn't crash when just the CPU was being tested before, and I was just being overcautious. 

Understandable.