Is 1.52v really safe for Sandy Bridge?

[tweakboy]

At 5ghz at 1.5v I hit 71C.

At 4.8ghz at 1.45v I hit 65C.

At 4.6ghz at 1.4v I hit 60C.

That is pretty nice temps...for HSF cooling...

 

[chimaxi83]

Yikes, what are the water temps like at that point?

Of that I'm not sure. I haven't plugged a temp probe into my loop just yet. Neither of my CPU's go higher than 85C though, with those healthy ~1.55V overvolts. I also have a 7950 in the loop and it tops out at 39-40C, so I would imagine the water temp is in that range.

Wow, those are some fast CPUs. You're brave giving them that much voltage.

Heh, well my bravery got a +2 once I picked up the Intel Tuning Plan Overclocking/overvoltage insurance ftw.

 

[SickBeast]

That is pretty nice temps...for HSF cooling...

Yeah it turned out I wasn't 100% stable at those voltages, though. My chip actually seems like an average overclocker. I'm settling at 4.5ghz at 1.35v for now. My temps are still quite good.

 

[BallaTheFeared]

Hey Balla, good question!

I thought about doing that for my GTX460 (since I had two application packs and only used one for my 2600k), but the application process and the curing process really looked to be super finicky and I decided I wasn't going to do it and risk destroying my 460.

Instead, I went the "pop the IHS off" route for my 460, and then used NT-H1 on the bare die. Worked great (10°C temperature improvement).

With this experience, I will always be popping the IHS off my video cards until such time that the IHS are welded on like Intel does theirs.

What about using a FPS limiter to keep the temps in check without burning out the card?

The problem with popping the lid is that I use a full cover block for both. It's milled to sit perfectly on the IHS/ram/vrms... If I removed the IHS there would be a gap and it wouldn't make contact.

I could sand down the parts that hit the ram/vrms but I dunno if that's the smartest thing to do nor do I think I'm capable of getting it flush enough not to affect the other parts :'(

 

[Idontcare]

The problem with popping the lid is that I use a full cover block for both. It's milled to sit perfectly on the IHS/ram/vrms... If I removed the IHS there would be a gap and it wouldn't make contact.

I could sand down the parts that hit the ram/vrms but I dunno if that's the smartest thing to do nor do I think I'm capable of getting it flush enough not to affect the other parts :'(

Yeah, no don't sand anything at that point, keep the IHS on. You need to worry about the mount with your block.

You could definitely go with the IX, I can think of a few ways that it could end in disaster since it isn't designed for that specific application but at the same time having used it on my 2600K I do think it is possible/reasonable to expect it to work on your 470.

The thing is it initially intentionally creates an air gap between the IHS and the block. So it will run hot, on purpose, until you go through the thermal cycling.

That means your vrms won't have good contact until the cure is done as well (because the whole block will be lifted).

 

[BallaTheFeared]

Humm do you think it would be possible to downclock the 470's a bit, then use a fps limiter to adjust load on them to control the heat?

I assume I want to run them around 95C sustained for the burn in period correct?

 

[Idontcare]

Humm do you think it would be possible to downclock the 470's a bit, then use a fps limiter to adjust load on them to control the heat?

I assume I want to run them around 95C sustained for the burn in period correct?

I don't know if 95C is warm enough, and would just need more time, or if it really does need 98C, but what you propose is exactly how I would try and do this (if I were trying to do it) so, yeah, I think it'll work.

In my case I had to run my 2600K at TJmax 98C for about 30 minutes for the cure process to complete. So if you do it at 95C then it would presumably take a bit longer still.

 

[BrightCandle]

Two things kill CPUs, voltage and temperature. Personally I go from Intel's word when it comes to the maximums for both, because if anyone is going to know and has actually tested it its them. Intel list a maximum safe voltage of 1.35V and a VAbsolute of 1.4V. Breaching 1.35V should not be taken lightly.

With an i7 920@4Ghz I ran it at 1.35V for 2.5 years before it started to degrade and wouldn't then hold better than 3.8Ghz, all the while under custom water topping out about 55C. At 1.4V it will likely show degredation within the chips lifetime, just be aware it'll happen.

 

[GotNoRice]

I run my 2500k @ 5Ghz, 1.48v. I've run it at that since last September and it's run great.

I turn my system off at night right now though, mainly because of the idle power consumption of my 2x 4870x2.

 

[BallaTheFeared]

I don't know if 95C is warm enough, and would just need more time, or if it really does need 98C, but what you propose is exactly how I would try and do this (if I were trying to do it) so, yeah, I think it'll work.

In my case I had to run my 2600K at TJmax 98C for about 30 minutes for the cure process to complete. So if you do it at 95C then it would presumably take a bit longer still.

Yeah the basic principle is there, I believe Fermi is ok up to 105C so I'll bump it up to the 98C if I can... Going to be pretty touch and go there though I'll need to use a game that doesn't change anything and let it stare at the same image the entire time ;o



My chip still hasn't degraded, but it only needs to last until Ivy Bridge

 

[tweakboy]

I think 71c at 1.5v is actually kind of nice. I think if I had a SB I'd try that . . . along with the Intel insurance plan.

I wonder if they made that plan so they could gather data -- on our dime -- about what it takes to make their cpu's blow up?

71c is fine if your going to upgrade in 2 years or soo.

 

[ehume]

71c is fine if your going to upgrade in 2 years or soo.

That 71c ought to represent a maximum. Inless you run FAH you wouldn't expect to see that temp 24/7.

 

[ehume]

How did I answer post #62 with post #61?

 

[Idontcare]

How did I answer post #62 with post #61?

That's impressive

 

[atm743]

i have my 2600K at 4.5ghz and it set on offset mode. (forgot what i have load line set to) but on full load never goes over 1.29V

the gain from 4.5 to 5.0ghz is min and not worth the extra voltage and heat imo

 

[Idontcare]

I know mine can take 2.6V, at least momentarily that is.

I forget I was in offset mode, thought I was still in manual mode, so I plugged in what I thought was going to be a manual Vcc setting of 1.35V and it turned out to be an offset of +1.35V to the existing 1.25V value.

When I saw the Vcc reading of 2.6V I nearly fainted in a panic to turn off the PSU. But it doesn't seem to be any worse for the wear, no degradation to speak of at this time. But I'm sure I rapidly accelerated the demise of the CPU. If it dies in a year I'll know why.

 

[LOL_Wut_Axel]

1.52V is definitely not safe for 24/7 use. Hell, even the often-touted 1.4V is only good if you don't want to lower your overclock in two years from having the chip. And after 1.35V power consumption seems to rise a lot.

I use my chip for folding@home, and as I've seen it's the most stressful program I've run. For everything else, including Prime95 and IntelBurnTest, it's completely stable at 1.26V at 4.3GHz, but for f@h I need 1.30V. Electricity here is also 26 cents/kWh, which is extremely high. Needless to say, my PC has caused our electric bill to rise by a lot, which is why I've cut running f@h by half. I hope 4.2GHz is stable in f@h at 1.25V or less given I need 1.30V for 4.3GHz. It would make power consumption lower by some 20W, and I've also changed and started to fold from ~11:00PM-8:30AM and not during 7:00AM-4:00PM because from what I've heard electricity rates are lower during off-peak hours and ambient temps are about 5C lower, meaning lower power consumption.

Well, at least I know my chip won't degrade much over the coming months and years even if at 1.3V. Temps in f@h are also around 65-70C.

 

[Puppies04]

1.52V is definitely not safe for 24/7 use. Hell, even the often-touted 1.4V is only good if you don't want to lower your overclock in two years from having the chip.

While I don't necerssarily disagree with you I would love to see some proof of this. You state it as fact when in reality it is your educated guess.

 

[tweakboy]

Too much noise for me,, heatsinks are.....

 

[Kenmitch]

I know mine can take 2.6V, at least momentarily that is.

I forget I was in offset mode, thought I was still in manual mode, so I plugged in what I thought was going to be a manual Vcc setting of 1.35V and it turned out to be an offset of +1.35V to the existing 1.25V value.

When I saw the Vcc reading of 2.6V I nearly fainted in a panic to turn off the PSU. But it doesn't seem to be any worse for the wear, no degradation to speak of at this time. But I'm sure I rapidly accelerated the demise of the CPU. If it dies in a year I'll know why.

That's incredible!...Kinda not really but in a way it is.

 

[Idontcare]

That's incredible!...Kinda not really but in a way it is.

I wouldn't have believed it, or even realized I was doing it, if it weren't for the fact I had my voltmeter plugged into the mobo and sitting right in front of me.

I thought for sure something would have gone pop at that voltage. But nothing is different, it OC's to the same levels (no loss in peak OC's) at the same voltages and same temps as before, all IBT tested.

It is NOT an experiment I'll be repeating though

 

[Phynaz]

How did I answer post #62 with post #61?

Quarks.

 

[Joseph F]

I wouldn't have believed it, or even realized I was doing it, if it weren't for the fact I had my voltmeter plugged into the mobo and sitting right in front of me.

I thought for sure something would have gone pop at that voltage. But nothing is different, it OC's to the same levels (no loss in peak OC's) at the same voltages and same temps as before, all IBT tested.

It is NOT an experiment I'll be repeating though

Now, that's a testament to Intel's robustness in their fabrication facilities.
2.6V...

 

[Thor86]

Use 1.42vcore on my SB. No issues so far, but 1.52 would be too much even with temps under control.