Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I've tried to overclock my Intel HD Graphics 4600 in my desktop PC.

I've read elsewhere; people managed to get 1600-1700MHz (OC) instead of 1250-1300MHz (default).

The problem here is I am not unable to get reliable consistent results.

At some point I was able to get a decent record, but I'm unsure whether it was "Adaptive" or "Static" voltage as the Intel Extreme Tuning Toolkit sometimes changes this setting.

I used FurMark for these records

SCORE CLOCK    VOLT
393 - DEFAULT
403 - 1350MHz

411 - 1400MHz
419 - 1400MHz (+0.085V)

408 - 1450MHz
407 - 1450MHz (+0.010V)
410 - 1450MHz (+0.005V)
415 - 1450MHz (+0.010V)
412 - 1450MHz (+0.012V)
412 - 1450MHz (+0.015V)


415 - 1500MHz (+0.010V)

411 - 1550MHz (+0.020V)
416 - 1550MHz (+0.040V)
414 - 1550MHz (+0.030V)

418 - 1600MHz (+0.080V)
418 - 1600MHz (+0.080V)
416 - 1600MHz (+0.085V)
419 - 1600MHz (+0.082V)
412 - 1600MHz (+0.084V)
422 - 1600MHz (+0.083V)
426 - 1600MHz (+0.100V)
427 - 1600MHz (+0.085V)
428 - 1600MHz (+0.095V)
435 - 1600MHz (+0.095V)

415 - 1650MHz (+0.165V)
426 - 1650Mhz (+0.175V)
426 - 1650MHz (+0.180V)

What I am thinking I am battling here is either some adaptive voltage or auto scaling.

It's quite random whether I can reproduce the results above or not.

Quite recently I even have problems reaching in the 400+ score (most scores drops below, even while being overclocked), which was not at all a problem earlier.

Without overclock as you can see the scores are from 350-400.

share|improve this question
    
I do have to say... "I've read elsewhere; people managed to get" just because others have success with OC'ing, doesn't guarantee success for all in the same endeavor. –  Bon Gart Jul 12 '13 at 15:46
    
Every physical processor die is different –  Ramhound Jul 12 '13 at 15:57
    
sure, this is all basic OC disclaimers. However I believe results should be noticeable or at least show better results instead of randomly dropping in the 3xx-scores? –  dezza Jul 12 '13 at 23:10

2 Answers 2

I suspect that Intel Extreme Tuning Toolkit merely sets the maximum turbo boost frequency of the iGPU. Then the actual frequency is dynamically scaled between the min and max allowable values based on the CPU thermal and power budgets. See fig. 1, page 7, http://www.intel.com/assets/pdf/whitepaper/323324.pdf

If this is indeed the case, then iGPU performance will depend on the current CPU load. So you might test this hypothesis by loading one or two of the CPU cores and checking how it impacts your score(assuming the benchmark is almost purely testing the iGPU). Also, you might be able to get more consistent graphics performance by increasing the allowable thermal design power(TDP) in your BIOS(not available on all mother boards).

Standard warnings apply.

Edit: http://semiaccurate.com/2012/04/23/overclocking-intels-hd-4000/ suggests that thermal throttling will kick in quickly when over clocking the HD 4000(that may explain your varying performance) and that increasing the TDP by increasing the current limit in Intel's Extreme Tuning Toolkit can resolve the problem.

share|improve this answer

Intel HD graphics are crap, and overclocking it is a terrible idea.

Spend $100 and get a video card that can leave your overclocked Intel HD graphics in the dust.

You probably need a large after market cooler to cool the graphics more.

You could also try a refrigeration system.

Liquid Nitrogen, a constant supply, will lower the temperature to super conductivity and it might work then also.

The point is you can not squeeze water from a rock, and each chip is different and therefore may not achieve your goal with any standard methods. All performance records are achieved with liquid nitrogen or similar. If you really want to push the limits that is how it is done.

share|improve this answer
    
I have bought the Phanteks PH-TC14PE - actually the cooling doesn't affect it that much before 1600/1700MHz. But the idea here is to get better performance without a video card. –  dezza Jul 13 '13 at 14:50
    
@dezza What performance numbers are you trying to achieve? I know you want better performance without a video card, but are your goals realistic. You can not make a intel HD 4600 perform like a geforce 680 no matter what. My point is I have overclocked many things is my time, and the final conclusion, in many cases, is I could spend 20 hours getting it to work or buy a better part and have 18 hours to spend doing something I really enjoy. Also the better the part the more it overclocks. –  cybernard Jul 13 '13 at 16:17
    
Well a few FPS would be great. I have no big expectations, just that it runs stable and marginally better. –  dezza Jul 13 '13 at 16:18
    
@dezza Would you please explain your reason for not getting a real video card? Case to small? Video card to expensive? Power supply to weak? Bragging rights(I overclocked nearly the worst modern video card in the world! hahahaha)? It would be easy to get 10 more fps with a real video card. –  cybernard Jul 13 '13 at 16:41
    
@dezza I used 3dmark/futuremark to determine the performance level of your HD 4600. A couple of examples: $41 for seattlegadgets.com/HP-nVIDIA-GeForce-Express-631078-001/dp/… $50 for $50 outletpc.com/jt7793.html –  cybernard Jul 13 '13 at 17:12

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.