Downclocking i7 2600K during stress test.

[potis]

Hello, as you can see its the first post and i am making it to ask something.

This is my first overclock i have read as much as i can on how (Here, Tom's Hardware, random people with same Motherboard/CPU and why but:

I certainly have an i7 2600K at 4.3GHz, at 1.255V (Motherboard seems to give 1.272 during idle though and 1.312 during Intel Burn Test) with an MSI P67A-GD65, i am not using the 1.19 BIOS or the 4.3 (If someone knows i would love to know if i should install the 4.3, MSI says i shouldnt since its for Ivy, others report its awesome, others report problems, i simply dont know!) since i had an issue with cold restart during boot because of some USB related issues combined with the newer BIOS (This mobo is a nightmare -.-) so i am using the 1.E versions or 1.14 as its called otherwise.

It seems to be stable, i have run Intel Burn Test at Very High and Maximum multiple times, letting it do 10 tests or 5 tests depending on how much time i have.

I have also tried 4.5GHz at 1.310V on the Motherboard, which is if i recall 1.34 on CPU-Z, and 1.38 during Intel Burn test but here is the problem, i would get BSOD of 0x124 which is not having enough Vcore as far as i have read so i kept increasing it by 0.05 as the motherboard allows till it stopped giving BSOD on the Intel Burn Test.

I cant test properly because the moment the test gets heavy 10 seconds in, the CPU downclocks at 4.3GHz and 10 seconds after at 4.2GHz, it could be temperature related since its hitting 88C on the third core while the others are at 82C since i read that some mobo's do that, but i cant find if mine does that and if its something else.

I know 88C is a bit high, i am using a Xigmatek Aegir but the idle temps are good enough for 1.344V compared to what i had for 2 years on the useless stock cooler during stock values... i am not gonna be running the PC at 100% for ages, i just play games and it wont reach that high, but my problem is that i cant test if its stable since it downclocks.

For other reasons that it would take ages to mention, mostly not wanting to waste money, i cant change my case to make the airflow better, i could reaset the Cooler with a new paste but thats about it..

Is it the heat that causes the downclocking or something on the damn MSI board that causes it?

EDIT: 5mins after i wrote this i just had an idle crush, i increase the Vcore a bit and enabled C State, now need to wait and see but i confirmed that the CPU downclocks when it reaches 85C (I change the volate to 1.270 from 1.255 so the load went to 1.336 from 1.312 and it went from 82C to 85C and it downclocked instantly), should i change the voltages to what i had and let it run the Intel Burn tests multiple times? Since it downclocks how can i know its stable? I wont even stress the CPU that much at daily use.

 

[EarthDog]

88C is fine for a 2600K. The limt per say, but fine. It should not be throttling there. Did you raise any of the power thresholds on that board?

 

[potis]

88C is fine for a 2600K. The limt per say, but fine. It should not be throttling there. Did you raise any of the power thresholds on that board?

And this is me being a noob.

Which Power Thresholds exactly?

The MSI UEFI, the old one is rather horrible which apparently many things got fixed at the CLICK BIOS II after the Z68 etc but i cant put that because its a 50% chance to be good or horrible for my CPU.

And i cant find anything related to motherboard and throttling in the BIOS/Manual/MSI.

 

[EarthDog]

I dont see why there is any chance a Z68/Z77 board would be bad for your CPU. That said, you should still be able to o/c just fine on that board.

You should find something like overspeed protection, long/short duration power limits, etc. Set the protection to disabled, and the power limits to 250 if possible and the 'maintained' to 1000 and see if that helps.

 

[potis]

I dont see why there is any chance a Z68/Z77 board would be bad for your CPU. That said, you should still be able to o/c just fine on that board.

You should find something like overspeed protection, long/short duration power limits, etc. Set the protection to disabled, and the power limits to 250 if possible and the 'maintained' to 1000 and see if that helps.

Ah those.

Yes, those are all set following http://www.tomshardware.co.uk/forum/265056-29-2600k-2500k-overclocking-guide mostly.

I just enabled C State because i had an idle crash and most posts concerning this issue shows me towards that way.

 

[EarthDog]

Disable C states... EIST... Check out the guide in my signature and try that...

 

[potis]

Disable C states... EIST... Check out the guide in my signature and try that...

Alright, i updated the bios to the latest version, if the USB bug reoccurs i will see then.

I followed your guide to the letter, 2 things.

1)There is no LLC anywhere or it has a different name therefor i cant set it.

2)Setting RAM to 1.5V will cause Windows to crash after boot, only mouse would work and thats it, setting it on Auto everything is fine, it takes the RAM to 1.65V though, but it doesnt crash during testing, the timings remain the same on Corsair Vengeance 9 9 9 24.

3)The Vcore is at 1.310 in the BIOS since the last few days thats where i found it stable, MSI motherboard overvolts it anyway.. no crash during the tests, just the downclock as i have mentioned.

4)I think i have PLL Overvoltage Enabled, i remember reading that my Mobo likes that and its more stable.

The end result is that the picture..

The VCore in the BIOS 1.31 because MSI apparently scales it up, its 1.328 during idle and 1.376 during full load on CPU-Z

The CPU downclocks to 4.2 the moment 2 cores or more hit over 85, well only 2 of them do >_>

You can see the maximum CPU it hit, but generally the temps were 80/87/89/83 most of the time it would randomly show +3C for a second and back to hose.

I will run Prim95 for a few hours now that i will go out, but running Intel Burn Test 5 turns as your post says there hasnt been any crash, as there hasnt been any other problem same as the last few days, just the downclocking that occurs after 85C on 2 cores

Idle temps seem to be 46 41 40 42, not great but not horrible considering i had 52 48 48 48 with stock at stock speeds a week ago.., i probably need to buy a new paste and reseat my cooler, i dont know,i will log onto a game and check the load, doubt it goes over 60C in any game really.

The picture was taken 5-8 seconds after the Success message popped.

 

[EarthDog]

2. Set ram to whatever yours needs.
4. You likely do not need PLL Overvoltage enabled at this low of a clockspeed. Disable it. It is only there to provide stability at much higher clocks.

Remount your cooler and try to lower voltage if possible. Temperatures as you noted, seem to be what is causing the throttling.

 

[potis]

2. Set ram to whatever yours needs.
4. You likely do not need PLL Overvoltage enabled at this low of a clockspeed. Disable it. It is only there to provide stability at much higher clocks.

Remount your cooler and try to lower voltage if possible. Temperatures as you noted, seem to be what is causing the throttling.

Alright, i will disable PLL and check again, and well Auto seems to work just fine for the RAM, so as a noob i will leave it there.

Edit:

I left Prime95 running 15 minutes just to check something, will leave it for 4 hours+ a bit later but i noticed something.

Prime95 took my temperatures to 82-86-88-82 after a minute..40 seconds slower but same as Intel Burn Test did, but no downclock happened on CPU-Z over these 15 minutes.. what the ?

 

[potis]

Well, i had an idle BSOD 2 days ago..I looked around and i enabled EIST/C1E, the temperatures went down to 43/33/33/35 while idle from 45/40/40/41 and there hasnt been a BSOD since.

Of course its too early and random but enabling EIST and C1E was the suggestion everywhere about SB so i just did that, the BSOD happened while AFK for ~4mins..since then i left it multiple times alone for 10 mins up to a hour, nothing happened, but its too early, hopefully what worked for others will work for me!

Full load testing remains the same, it goes to 1.376V from 1.320V and it finishes without a problem..