5820K overclock help

spooks

spooks

spooks

Associate
Joined
21 Aug 2007
Posts
236
Hello all, my new 5820k is currently sat at 4.5ghz @ 1.33v vcore. I have had it running on 1.3v for 22 hours under aida64 extreme stress test... however it would crash quite quickly in battlefield 4 (sometimes)! After bumping up to 1.33v it seems stable with everything including BF4.

That said, I'm having problems going past 4.5ghz - I can do up to 1.5 - 2 hours of stress testing at 4.650Ghz at 1.375v but I don't want to push the voltage any further. Are there any other voltages/tweaks I can acheive to get a totally stable system at my target of 4.7Ghz?

My system:-

5820k
EVGA FTW X99 board
16GB Corsair LPX RAM (2666mhz) running XMP profile 1
GTX 980 EVGA SC edition

Full custom water loop
XSPC D5 Vario Pump & EK-D5 X-TOP
1x 280MM RAD, 1x 240MM RAD

Temperatures even at 1.375V do not exceed 80C under full AIDA stress testing. My board is set to AUTO for everything else.

cheers!
 
Hey, thanks for the response - It will be configured to AUTO. I've not touched any setting other the the vcore and changed my memory profile to XMP.

Does this setting have something to do with the supply of power to the CPU? I.E how much is available to the CPU?
 
Hi - thanks for the advice. I'll take a look into the vrin setting. Is that the only other important setting? Are we saying that once the vrin is up I should be able to get a higher clock speed?
 
You'll be VERY lucky to get a 5820k stable at 4.7. If you do, you've won the lottery with your CPU, as many struggle to get near that. Mine is stable at 4.4, couldn't really get it beyond that, but I may yet revisit and tinker some more. I'd be happy with 4.5 if I were you though, 4.7 remains a dream for many.
 
What cooler are you running .. and going on that I am ok to up my one a bit more .. is 80 ok as I still have some room ..

I am at 4.375 or 4.4 and stable but my voltage is still low at 1.184

4.5 it boots but will not last long under testing .. might have to have a go with a bit more voltage

Richard
 
Legend said:
You'll be VERY lucky to get a 5820k stable at 4.7. If you do, you've won the lottery with your CPU, as many struggle to get near that. Mine is stable at 4.4, couldn't really get it beyond that, but I may yet revisit and tinker some more. I'd be happy with 4.5 if I were you though, 4.7 remains a dream for many.
Click to expand...

I have a cherry chip. Does 4.4ghz with 1.25v, 4.8ghz with 1.35v
 
What a we all using for stability testing ?

Not really finished playing with mine yet, Currently running at 4GHz at 1.1v stable in IBT.
So far had it up to 4.5GHZ at 1.246v runs cinebench benchmark fine but will crash IBT using 4GB mem(very high) after 5-6 results pushing up to 74c.
 
I've found so far that I can get AIDA64 Extreme to run for around 22 hours (stopped it at that) at 4.5Ghz @ 1.3v. Unfortunately launching and playing BF4 results in a crash anywhere up to 1 hour of playing.

Really weird that it can pass 22 hours of rape via AIDA64 but bombs out gaming. Anyway, increasing the vCore a bit more to 1.33 seems to have resulted in a stable system at 4.5Ghz. Anything above this causes problems 4.55Ghz at 1.35V is a no no and crashes under AIDA anywhere from 5 minutes to 5 hours.

I think 4.5Ghz is the absolute most I'm getting out of mine unless I go silly with the voltages... Shame really as the 5820K ultimate overclocking guide on youtube suggests that 4.5 should be a piece of cake to achieve at 1.3v.
 
Richard-h2 said:
What cooler are you running .. and going on that I am ok to up my one a bit more .. is 80 ok as I still have some room ..

I am at 4.375 or 4.4 and stable but my voltage is still low at 1.184

4.5 it boots but will not last long under testing .. might have to have a go with a bit more voltage

Richard
Click to expand...

I'm running a full custom water loop with a liquid metal thermal interface material.

I think you'll need at least 1.3 volts maybe more to get 4.5 (from my experience anyway).

80C seems high for such a low voltage...
 
spooks said:
I'm running a full custom water loop with a liquid metal thermal interface material.

I think you'll need at least 1.3 volts maybe more to get 4.5 (from my experience anyway).

80C seems high for such a low voltage...
Click to expand...

I am not at 80 yet but seams I can push it a bit more as I have more room before 80

Might have to have a go and up the voltage a bit more and try 4.5 over the weekend ..

I am on a Air cooler . so guess there is only so far I can or could go ..

Richard
 
Well, I'm still attempting to push my cpu to the max. I'm currently sat on 4.6ghz with the following settings:-

vCore: 1.28
vCore offset: 0.8
Total vCore: 1.36

VIN: 2.00

20 minutes under AIDA64 Stress test so far... doubt it will be stable but we'll see.

Do these voltages stack up for long term use with a water cooling loop? Temperatures are fine, mid 60's with a max peak of 78C.
 
Well I tried 4.5 pushed Vcore up to 1.210 my Vin is a 1.95 and it passed 15 mins

but temp did go to a max 80 on one core .. but it only spiked there was below and more like 78.

Not sure why my Core 3 or 2 if you start from 0 is a lot more .. think I might redo the thermal stuff

Do you do a blob or a line ???

Richard
 
Not to be a bumder but do remember all that the max safe temp at Tcase for this chip is only 66c and 1.3v seems to be the max safe voltage from what I can find long term.

IMO for normal, real world use - ie. not just benchmarks, 4.4Ghz seems about the realistic hassle / crash free limit for most I'd say. I don't see much point in going much higher as temps would also get way beyond Tcase max on anything like IBT or P95. Even with full water loop I get up to 72c with 1.3v.
 
Stability - OCCT is the best program to find issues. I have used AIDA/Realbench/Prime (blend and Small FFT)

Prime does a good job, don't get crashes if it all went through. But it causes problem with heat and adaptive voltage on Haswells.

OCCT does as good job as prime, although it tends to BSOD when there is something wrong as opposed to posting errors. and the heat is more distributed.

for gaming, its a good idea to run Rog Realbench - bench test (not the stress test) a few times in a loop as that bench test has some pretty hardcore multitasking bit at the end which will definitely test stability in a gaming environment.
 
You must log in or register to reply here.
Back
Top Bottom