Problems breaking the 4.2Ghz mark on 4670K?

[Boonstick]

I've been messing about, trying to overclock my new i5 4670K and I cant seem to get past the 4.2Ghz mark at a decent voltage. So far I've got it stable, using OCCT to stress test, at 42x on all cores @1.225v - 42x cache @1.2v. Even these voltages seem a bit high for a 4.2ghz overclock. I'm still testing but I'm sure if I go any lower it becomes unstable.

Is there anything that Im doing wrong as I'm a bit new to OCing but I've followed the guide on here and read a few tips on a few other sites and I feel I'm doing the right thing. Is it likely that 4.2ghz is my OC limit?

I cleared the CMOS and changed no other settings in the BIOS (except boot sequence). Updated BIOS version to the latest.

I'm using an ASUS Maximus VI Formular motherboard.

Any help would be greatly appreciated.

 

[airdeano]

what voltage is 4.2 stable at? would it be stable at 70% memory? tried Intel Extreme
Utility? or AIDA64? could be hitting it too hard with OCCT LinPak Mode.
have you tried single core testing?
possible one core is holding your stability hostage.

 

[NeverBackDown]

What are the temps? cooling solution? How well is the case ventilated?

 

[Boonstick]

@NeverBackDown - Cooling wise, I'm using a Corsair H100i and the case it well ventilated, 2x 120mm front intake fans, 1xAF120 base intake fan , 1x AF120 rear exhaust fan. The temps at 4.2ghz are at around the 70*c mark but I will confirm that when I get home from work.

I've read somewhere that Haswell runs a bit hotter than previous gen but not sure how true that is. Even at the 4.2 OC these temps seem a bit high.

@airdeano - 4.2 volts are at 1.225v fully stable with OCCT on 6 hour test.

I haven't tried single core testing yet and I'm still getting used to OCCT. I was doing some more testing last night at 4.4ghz and it seemed that if my pc didn't randomly crash/blue screen, core1 was hitting the temp limit of 86/87*c first. Maybe it is the type of test I'm using in OCCT as I managed to get stable settings using the Intel Extreme Utility and also run futuremark several times with no problems. To memory, the vcore volts were at 1.9v.

Thanks for the replies, everyone! Great help



Right, I've started again, resetting to default settings, and also following TTL's video guide to the letter and it seems to be a lot more stable. I've managed to get it up to 4.5ghz with vcore at 1.275v and core 1 (highest temp) at around 77*c. I'll do an over night test to confirm. The only problems now is, the CPU usage seems to dip to around the 25% mark for about 20/30seconds every 4 mins.

Is this a software or hardware issue?

 

[EmmEff]

I'm no expert by any stretch, but I've encountered the same problem with my i5. Anything over 4.2 was eating volts. I decided to settle for 4.2 @ 1.16. The reason I settled is that the performance boost from stock to 4.2 was excellent, but the boost from 4.2 to 4.3 or 4.4 just isn't worth the hassle / risk.

 

[NeverBackDown]

Right, I've started again, resetting to default settings, and also following TTL's video guide to the letter and it seems to be a lot more stable. I've managed to get it up to 4.5ghz with vcore at 1.275v and core 1 (highest temp) at around 77*c. I'll do an over night test to confirm. The only problems now is, the CPU usage seems to dip to around the 25% mark for about 20/30seconds every 4 mins.

Is this a software or hardware issue?

Hardware. This means temps are hitting 100C(or 95? Can't remember) and in order to keep itself from committing suicide it throttles down the clock speed and volts till it cools down. If your CPU has to do this it is not stable and you MUST lower clock and volts; otherwise risk killing off the CPU and haswell chips are very picky about to much volts.

Also as said above Haswell chips hit a point where an immense amount of volts are needed to get that one extra 100Mhz and that is the point where you should stop and stay under that leap to keep as long of a life cycle and best temp/noise ratio.

 

[theDTP]

@NeverBackDown - Cooling wise, I'm using a Corsair H100i and the case it well ventilated, 2x 120mm front intake fans, 1xAF120 base intake fan , 1x AF120 rear exhaust fan. The temps at 4.2ghz are at around the 70*c mark but I will confirm that when I get home from work.

I've read somewhere that Haswell runs a bit hotter than previous gen but not sure how true that is. Even at the 4.2 OC these temps seem a bit high.

@airdeano - 4.2 volts are at 1.225v fully stable with OCCT on 6 hour test.

I haven't tried single core testing yet and I'm still getting used to OCCT. I was doing some more testing last night at 4.4ghz and it seemed that if my pc didn't randomly crash/blue screen, core1 was hitting the temp limit of 86/87*c first. Maybe it is the type of test I'm using in OCCT as I managed to get stable settings using the Intel Extreme Utility and also run futuremark several times with no problems. To memory, the vcore volts were at 1.9v.

Thanks for the replies, everyone! Great help



Right, I've started again, resetting to default settings, and also following TTL's video guide to the letter and it seems to be a lot more stable. I've managed to get it up to 4.5ghz with vcore at 1.275v and core 1 (highest temp) at around 77*c. I'll do an over night test to confirm. The only problems now is, the CPU usage seems to dip to around the 25% mark for about 20/30seconds every 4 mins.

Is this a software or hardware issue?

Please oh lord tell me that was a typo...

Also I can run my ram at 2400MHz @ 4.2GHz, only 2200MHz @ 4.4GHz and 2000MHz @ 4.6GHz (this is recent, I had it at 1600MHz until I found out BF4 does benefit from higher speed ram) That probably has something to do with me using all 4 lanes, could help if I took 2 sticks out.

I run 4.2Ghz @ 1.15v, 4.4GHz @1.28v and 4.6GHz @ 1.36v which also allows for 4.7GHz but I have to increase in multiples of 200MHz, OCD of mine, that 1.36v also has an offset, it does boost up to 1.375 at times, I'm not too good at overclocking and have no clue how to stop that. That 4.8GHz requires me to use 1.45v which I'm not happy to use even though temps are fine in game. I am idling now (PC has been on for 8 hours doing all sorts, gaming and such) @ 29, 26, 25, 26 with max temps all night of 62, 57, 59, 57.

I also have all that power saving crap turned off even though everyone says it doesn't matter with Haswell, turbo boost disabled. Runs 4.6GHz constantly.

 

[Boonstick]

Please oh lord tell me that was a typo...

Haha, whoops! Yep, typo. I was using my phone and it goes a bit funny on this forum when typing long messages.

1.29v

@Neverbackdown, I don't think it is hardware. Looking at the OCCT graphs the temps are well below the threshold and the intervals between when it happens seems way to exact. It's literally every 4 mins for 20/25 seconds. Everything seems ok at 4.4GHZ now anyway.