Low OC limit despite keeping cool

[Spuff]

i5 4670K
H60 cooler
Asus Maximus Hero

I'm monitoring temperature with HW Monitor and Real Temp. They agree within 3 degrees.
Using the net now I read a temp of 33 degrees (which is 7 degrees above my room temp (+/- 3 degrees error range on my room thermometer).

Under Prime95 100% Cpu load at 4.2Ghz on the processor I read 66 degrees maximum.

If I set the CPU multiplier to 44 I can't even boot into Windows.
The only BIOS settings I've changed is the RAM set to use the XMP profile, and just changed the CPU multiplier. With the Bios setting remaining in Auto, right now the CPU is being fed .0704V. Under test it goes to 1.365V (but stays as cool as I described).

I can't imagine an increase of .2 would be sending my temperature overboard (if the monitors are to be believed). Is stability of a processor independent of temperature?

It would seem a very 'unlucky' Haswell if it can't go beyond 4.2.

EDIT: 4.2Ghz isn't stable despite the temperature not going above 70 degrees under 100% load. The system crashed without the CPU going above 52% load in use.
4.0 does seem to be stable.

 

[tom_gr7]

Haswell and IVY are known not to be great overclockers. Sure some chips might be a bit higher, but its all luck to which chip you get. Its something to do with the thermal paste inside the cpu, (el cheapo thermal paste). Whereas sandbridge cpus have a much better solder inside instead of Dav's el cheapo paste. Something like that anyway.

I'd just leave it at 4.2 mate, i think the highest I've seen an Haswell chip was about 4.4 (for normal person)

 

[keynes]

I assume there is a good reason why PCS offers only up to 4.2 Ghz overclock on the i5, try contacting them if they have the correct settings. Last time I did it , it involved more than just changing the multipliers between 4.5 Ghz or higher.

 

[Rustilol]

i5 4670K
H60 cooler
Asus Maximus Hero

I'm monitoring temperature with HW Monitor and Real Temp. They agree within 3 degrees.
Using the net now I read a temp of 33 degrees (which is 7 degrees above my room temp (+/- 3 degrees error range on my room thermometer).

Under Prime95 100% Cpu load at 4.2Ghz on the processor I read 66 degrees maximum.

If I set the CPU multiplier to 44 I can't even boot into Windows.
The only BIOS settings I've changed is the RAM set to use the XMP profile, and just changed the CPU multiplier. With the Bios setting remaining in Auto, right now the CPU is being fed .0704V. Under test it goes to 1.365V (but stays as cool as I described).

I can't imagine an increase of .2 would be sending my temperature overboard (if the monitors are to be believed). Is stability of a processor independent of temperature?

It would seem a very 'unlucky' Haswell if it can't go beyond 4.2.

EDIT: 4.2Ghz isn't stable despite the temperature not going above 70 degrees under 100% load. The system crashed without the CPU going above 52% load in use.

4.0 does seem to be stable.

You should be able to get stable at 4.2
Try these steps to see if you can get stable at 4.2ghz

1. Go into bios,
2. change core voltage from adaptive voltage to manual voltage.
3 set v-core to 1.25v (Never go over 1.3v else your temps will be insane and possibly kill chip on a stress test)
3. enter windows and test your stability, monitoring temps as you go.

This will hopefully be stable. If it is, reduce the vcore by 0.01 each time and test again. so 1.25, then 1.24, 1.23 etc until its unstable.

If you are trying to reach 4.4Ghz, I would do the above steps but ofcourse change multi to x44. If it is stable in 1.25, gz, if not then crank up the vcore to 1.3 (warning, watch your temps very carefully)

If you then get x44 stable at 1.3v, try and reduce it again by 0.01 until unstable, then go back to your minimum. If you cant get stable with x44 at 1.3, I would say forget about it and stick with 4.2, which is respectable.

I have 4770K, for 4.4ghz I need minumum of 1.25. The 4670K stock clock is .2ghz lower so expect .2ghz lower

Also, if you scroll down all the settings in your asus bios, in the description some say "recommended disabling when overclocking" so disable them too. Just helps with stability

Please report back Good luck!

 

[Rustilol]

Also, I would recommend downloading intel extreme tuning utility. They have a benchmark test which is basically a very short stress test. Very rarely has my system passed the bench test then crashed on a 6hr+ Aida64 or other stress tests.

 

[jasonorland88]

This might help ..
http://www.overclock.net/content/type/61/id/1648491/flags/LL

 

[Spuff]

You should be able to get stable at 4.2
Try these steps to see if you can get stable at 4.2ghz

1. Go into bios,
2. change core voltage from adaptive voltage to manual voltage.
3 set v-core to 1.25v (Never go over 1.3v else your temps will be insane and possibly kill chip on a stress test)
3. enter windows and test your stability, monitoring temps as you go.

I've done that but Prime95 crashed the system (not so at 4.0) just now. At other attempts at 4.2 Prime95 has been OK, but other things have crashed the system. Temperatures remain in the low 60's.
I do have yet to try turning off the "recommended disabling when overclocking" bits.

 

[Rustilol]

Try cranking up the voltage if you want the over-clock, but keep an eye on temps. Increase in 0.01 til stable but dont go past 3.0v imo.
You can try increase cache v to 1.15 or 1.2 if you have over clocked that at all.
3 Other settings which can help, but obviously not as important as voltage: (I use Asus z87-a) CPU Power Phase Control -> Optimized / CPU Load-line calibration -> Level 4 / CPU Spread Spectrum -> Disabled

Good luck