Or at least mine seems to, from what I've seen of the voltages others are achieving. Seems I have a chip that needs voltage.

 

I let Asus Turbo V Evo do it's thing, and it gave me 4.7 GHz, but at a very high voltage of 1.46 Volts. Temps in Aida were right up to 93 degrees on one of the cores, I shudder to think how high the temp would have been in prime. Yikes! Thanks Asus, seems the algorithm you cliam to be first rate, isn't first rate at all. The other three cores were 10 degrees cooler. Such a variance in temps across the cores, the garbageTIM Intel used perhaps? De-lidding required perhaps?

 

I figured that Asus Turbo V might be overvolting, so I fine tuned the voltage, even OC'd in the UEFI, but no, at 4.7 I certainly seem to need voltage well bove 1.4.

 

Temps are too high at 4.7 of course, so I backed off to 4.6, and managed 1.32 volts. Temps at 4.6 were 80 degrees on the hot core and more like 70 on the other three. That's in Prime after 30 mins. No lengthy stress tests yet.

 

At the moment I have it at 4.5 GHz, and 1.265 volts, temps at 74 on the hot core.

 

Any tips you guys can give me, in order to clock higher without excessive voltage?

 

Would de-lidding even the temps accros cores?

 

 

I use an NH-D14 at the moment.

de-lidding is the must if you want high overclock!

 

Here is the great guide: http://forum.avsim.net/topic/387006-ivy-bridge-pictorial-de-lidding-guide/

Hmmm... seems so for me. The high temp is no doubt necessitating even higher voltage.

 

My initial goal was 4.7. I don't mind the 4.5 I'm at currently, I mean it's only 200 MHz, but something tells me I may be tempted to de-lid very soon. I don't see why I should put up with Intel's incompetence Or was it incompetence I guess the small cost they save by using cheap Tim instead of solder, amounts to a lot of money when you sell thousands of CPU's.

Hmmm... seems so for me. The high temp is no doubt necessitating even higher voltage.

 

My initial goal was 4.7. I don't mind the 4.5 I'm at currently, I mean it's only 200 MHz, but something tells me I may be tempted to de-lid very soon. I don't see why I should put up with Intel's incompetence Or was it incompetence I guess the small cost they save by using cheap Tim instead of solder, amounts to a lot of money when you sell thousands of CPU's.

 

You probly need delidd to get that 200mhz , costa rica or malay and batch? please

Batch is 3226C073 Costa Rica I believe Westman. :smile:

Yep, need to delid to make temps go down.

 

As for the voltage, yeah, you are in the same boat like me. I also have one that needs (needed) 1.48V for 4.8, and about 1.55V for 5.0. 5.1Ghz was no-go. Didn't try 1.7V lol...

 

I am contemplating of getting a 2700K with hope to be able to OC better.

 

2700K is better for OK, no doubt about that. But, i think there is something important about fsx and overclock stability. My 2700K was able to achieve stable 4.9GHz with 1.4 vcore. Stable in all stress tests - IBT, Prime95 etc. But there was occasional BSOD in fsx, approx. 1-2 weeks interval beetwen BSOD's.

My solution to this was to reduce GHz by 100 and keep the other settings. So, i reduced GHz to 4.8 and I kept 1.4 vcore. No BSOD so far(2 months), and fsx looks more stable. I can set lover value for vcore, but i have some theories about stability, and i think that higher voltages means better stability if you don't have problems with temperatures. But that's just me

I wrote about this becouse i think that your initial 1.48vcore for 4.8GHz was not really so high

Didn't try 1.7V lol...

 

 

The thermal paste might combust! :lol:

I don´t feel comfortable running the 3770 at high voltages 24/7 so I oc it to 4.9/1.46v just before starting fsx via the msi control center.

That's ok, you can OC before start fsx or use offset mode

That's ok, you can OC before start fsx or use offset mode

 

I use offset mode and have the OC at 4.9 after delidding. So far so good. I will say that when I did the initial tests after delidding 1 core was much higher than the others (15+ degrees) so I went back and reapplied the thermal paste and now under load the cores are never more than 4 degrees apart. I know that the OP has not delidded but the fact that you are getting higher temps on one core may be due to a bad application of TIM under the IHS (or just bad TIM in general, thanks Intel) or between the IHS and heatsink.

Running now at 4.5 GHz. Using an offset voltage of 0.030. LLC's giving me 1.304 volts under load. Temps in prime are 65,73,68, 68. Been stable in Prime for an hour. I'll test more thoroughly tomorrow if I have time.

 

Definitely seems to be 4.6 my limit. If I go beyond 1.33 volts, [required for my chip at 4.6] the temps shoot up, as other Ivybridge owners have testified. Once it hit's that brick wall, the garbage Intel TIM has had enough and can't transfer the heat.

 

@Jordan. What sort of temps are you getting now at 4.9, after de-lidding? How did you find the process, easy enough or any problems? [Other than the initial variation in core temps you already mentioned]

 

I am using an NH-D14, I do wonder if the weight of the cooler might be causing higher temp in that hotter core. I might give it a nudge with my finger tomorrow and see what happens. :biggrin: More likely to bIntel'sls TIM but you never know.

Your temp´s should stay in the 60´s @4.9GHZ and about 1.45v (depends also on your cooler) if you follow exactly @TechguyMaxC´s de-lidding.guide (Coolaboratory Liquid highly recommended).

 

@Jordan. What sort of temps are you getting now at 4.9, after de-lidding? How did you find the process, easy enough or any problems? [Other than the initial variation in core temps you already mentioned]

 

I am using an NH-D14, I do wonder if the weight of the cooler might be causing higher temp in that hotter core. I might give it a nudge with my finger tomorrow and see what happens. :biggrin: More likely to bIntel'sls TIM but you never know.

 

For cooling I have the H100; I also lapped the IHS and base of the H100 to a mirror finish while I had my processor out for the delidding. I replaced the Intel TIM with CoolLaboratory Liquid pro (CPU DIE - TIM - IHS - TIM - H100) Max temp was 88 during 30 runs of Intel Burn Test at max memory (14ishGB) and 81 during a 12 hour Prime95 Blend. My goal was to keep my stress test temps below 90. FSX temps never get out of the 50's.

 

As far as the delidding goes it was pretty easy, but you do need to be patient. Also, a sharp razor is a must. I got a pack of 10 scraping razors from Home Depot for about $2 (USD) and ended up using a new blade for each edge. As long as you follow TechGuys guide that is sticked in this section you should be fine. A q-tip and rubbing alcohol is enough to remove the Intel TIM (and the first application of Liquid Pro in my case). A little pressure with a fingernail is good to remove the epoxy that kept the IHS in place.

Thanks Jordan.

 

From what I've read, the type of TIM makes a difference, but tests are demonstrating that it's the thickness of Intel's TIM that's the primary issue.

 

I'll see how my stress testing goes at 4.5, but I can see myself de-lidding soon.

On the other hand...

 

Just done some testing in FSX. With the NGX, in the worst weather scenario I could muster, courtesy of REX, the frame rate isn't dropping below 35. In reasonable weather I'm seeing 45 frames per second in the cockpit at EGLL and 60 plus in the air.

 

That makes me think, why do I need to delid to gain a couple of hundred MHz, when my frame rate is now excellent. 300MHz would do very little to increase performance, and my temps in FSX are only mid 50's.

 

Thinking now there's no point in bothering to de-lid.