The new guy

If I do it from the bios,all I have to do is put a 0 in the cores I want to disable?

blueknight1960 said:

The 1st 3 cores are very close in temps but the 4th core has a 15c difference (cooler) and that is preventing me from over clocking to the speed I'd like to get at. But on the other hand maybe the cooler core is the one with the problem. Is there anyway to check each individual core?

Click to expand...

How does the temp variation prevent you from overclocking to your desired speed? What speed are you trying to reach?

The ~ max temp(s) range of the other three cores are what you need to monitor... Not the cool running core.

On the same air cooled setup that I posted above... My max temps shot up into the mid 90's! for a very brief run with AIDA64 FPU (AVX) which is similar to Prime95 small fft.

My air cooler just can't handle it. Now that spring has arrived here in Atlanta I should probably consider dropping back down to 4.4GHz or consider changing out my air cooler to something better.

Haswell cpu can run very hot unless you have really good cooling. Sometimes people de-lid their cpu to help improve temps... But that is something that I would NOT recommend to a novice overclocker.

I got a reply from intel-- Here is what they say about the temps.-----------------------------Agent Comments:
Hello,
I understand that you are concerned about the fact that there is a
difference of 15 degrees Celsius between cores.
This difference it totally normal and expected, actually you can see
differences of 20 degrees Celsius between cores.
This is because some cores are more used than others, even if the system
is in idle mode.
Usually the operating system uses more one core than the others, please
do not worry about the cores temperature difference, just pay attention
to the CPU temperature itself.
As long as the CPU temperature does not reach above 73 degrees Celsius,
the processor will be running under thermal specifications. Usually the
cores temperature is higher than the CPU temperature, so check the CPU
temperature not the cores temperature.

- - - Updated - - -

My point about the higher temps in the 1st 3 cores compared to the 4th core is that if the temps were as cool the 4th core,I might be able to get a higher clock without worrying about the temps reaching 100c.

You always go by the hottest core anyway...

I believe the 73C is the package temp, not core temp (which is good to much higher than that).

Intel reps are customer service employees reading directly from a script, and that script is OLD. They still quote cpu/IHS spec temps as Earthdog stated, if you actually got to one of their engineers, which some have posted in Intel help forum years ago, you would get past that cpu temp quote.

And like earthdog said, only hottest core temp matters, real issue is calibration is only accurate to +/-5C at tjmax (100c) so 10C difference is normal even at 100C, below 100C, you can get higher than 10C temp differences.

I appreciate all the replies. Here is where I'm at on my overclocking, which this is where I'm going to leave it unless the temps gets into the upper 80's c. 4.5 Ghz with 1.21 vcore. I ran occt for over an hr and the temps stayed in the middle to upper 70's c with the occasional drift into the lower 80's c on the hottest cores. The only games I play won't use any where near full cpu power so I think I am safe for the time being. Once again Thanks for all the advice.

Well I'm pretty happy with myself right now. I got to messing with the voltages and cpu frequency and I think I hit the sweet spot. Setting are:multiplier-46------vcore--1.25------cpu input volt--1.80 . Here's a look at aida64 with almost an hr of run time.--

Any comments or suggestions are welcomed.

Well I did a numbskull thing and didn't realize it till I read a different post. When I installed the h100i,I put the bracket upside down which was causing the high temps. The temps are closer together now and are also about 10c -15c cooler. So if anyone else buys the Corsair H100i aio,make sure you put the back bracket on correctly.

----------------------------------------This is after I fixed the bracket and aida64 has run over 5 hrs.

Much better... Enjoy!

Thank you.

Ok I'm back at trying to oc again and this time I'm going for 4.8ghz with my i5 4690k. What I have so far is 4.7ghz @ 1.31v , vccin 1.85 with a few other tweaks and it seems to be stable using intel burntest, occt and intel xtu. Now my question is I'm going to try 1.35v on the cpu and 1.90v vccin, will the voltages be safe as long as the temps don't get over say mid 80's c while testing?

Yep, you're all good just watch the temps mid 80s to high 90s isn't a big jump when the V_Core goes up.

Thanks for the reply. My temps with the 4.7ghz 1.31v while running intel burntest is right at 80c average.

Unless you reach a temp wall in your overclocking there is no real point in delidding. If you're pretty sure high temps as opposed to high core voltage is holding you back then delidding would likely be a sensible option.

Well the 4.8ghz oc wasn't stable so I took it back to 4.7. The weird thing is, while it passed all the stress testing and I actually did it for 4 hrs, one of the games I play, which doesn't use much cpu power at all started freezing and I noticed a couple other programs freezing. But no bsod and once I got the game and programs closed (task manager) everything else works fine. So for the time being I'm back at my 4.6 ghz oc. Can anyone shed any light on what's happening?

Could be a buggy video card driver.

Try giving your CPU another 0.01 - 0.02 of vcore and see if that doesn't cure the instability in gaming. Just because the game doesn't utilize a lot of CPU doesn't mean the instability is not coming from the CPU overclock.

IntelBurnTest is a very dated piece of software. It is not a good choice for stability testing any more. It is a good choice for testing your cooling system. Try running four hours of Realbench and 3 hours of OCCT. Run Intel Extreme Tuning Utility over night. I'm betting you need more vcore. I've come to the conclusion that the best way to test for overclock stability is to use a variety of stress tests and ones that emphasize more real world kind of test routines. Ones that are like the apps we actually run. I'm becoming more and more of a fan of Realbench.

Well I ran 4 different stress test for 3 hrs each and the 4th (xtu) for 8 hrs and my 4.7ghz @ 1.31v is stable. No errors or bsod. As far as my game freezing, I think I found the problem, I noticed last night the vram was 100% used and that is when it froze. Time to upgrade my gpu I guess and I haven't noticed any other programs freezing anymore.