Temp Check

[james1701]

Here is my 3770K at 4.5ghz. I have my fans only running 1200 RPM.

That is my highest temps, I had to pause for a screen shot, but caught the wrong window. I had to restart it just seconds later to get the whole desktop. The downside is with my chip, any more voltage and my temps sky rocket.

What do you think?

 

[BonzaiDuck]

Here is my 3770K at 4.5ghz. I have my fans only running 1200 RPM.

That is my highest temps, I had to pause for a screen shot, but caught the wrong window. I had to restart it just seconds later to get the whole desktop. The downside is with my chip, any more voltage and my temps sky rocket.

What do you think?

Well, a coupla things. What's your cooling strategy (air or water) and what type of cooler are you using -- especially for air?

And why are you stress-testing with the built-in AIDA-64 test? It might be more revealing to see what the temperatures are with IntelBurnTest, and at least a setting of "High" -- if not "Maximum."

You know, of course, that your system isn't trusted to be "stable" unless you can get through 25 or 50 runs with either IBT or LinX, or a 16-hour run with Prime-95 small-FFT test?

The Ivy Bridge cores were notorious for showing higher temperatures than their Sandy Bridge predecessors, owing to the fabrication process that eliminated use of Indium solder between the processor die and heat-spreader. And for that reason, people began removing the IHS to either replace the TIM/thermal-interface-material or to install their heatsink/fan directly to the processor die.

I could only guess from your temperatures using the AIDA stress program that your temperatures are at least likely to reach into the mid-80's Celsius with IBT. Maybe higher.

With my SB-K chip, I chose to limit the voltage and clock speed to whatever brought me to an 80C average @ 77F room-ambient with LinX or IBT "Maximum." Then I was able to improve (air) cooling to bring that down to an average of ~73C. Others will tell you that you can set your clock and voltage for stability in the 90's C range, and they justify it with the explanation that your processor will never come close to that under normal usage.

 

[MongGrel]

Looks respectable to me.

My X5650 does about 68-69C at 4.35 at 1.48 volts on air, running Prime on blend.

It's old these days, but 6X12.

Maybe I didn't look at everything close enough, but doesn't look bad here keep tweeking

 

[Deders]

or a 16-hour run with Prime-95 small-FFT test?

I'd go for at least 24, I've had errors after 23 hours in the past.

 

[BonzaiDuck]

I'd go for at least 24, I've had errors after 23 hours in the past.

I've only speculated that the odds of an error decline as the number of error-free hours increases -- maybe . . . a Poisson distribution. But if you got an error after 23 hours without any, we would be cautioned against just "playing the odds."

Even so, with that sort of time-frame, I'd be inclined to bump up the VCORE by three or four millivolts and call it a day.

 

[james1701]

I get the same temps and power draw from Aida as I do with P95. I passed a 24 P95 with no trouble. IBT gets me an average of 82C with 30 runs.

The one test that requires more voltage than any of those torture test is BF4. I can pass the torture tests at 1.225 but anything less than 1.25 and I get WHEA errors while playing BG4.

The good news is even with 2 7970 Windforce cards dumping heat inside my case I'm only running an average of 40C in games. I am using a XSPC Raystorm copper block with an AX360 rad.

Later when something comes out that is a worthwhile upgrade in case I mess something up, I will delid. Then I can see how far I can really push this chip.

 

[guskline]

Is your 3770k in your loop or are you only water cooling your gpus? Since you have a Corsair 900D case you can easily add another 360 rad.

 

[james1701]

The 3770K is the only thing in the loop right now. There are no full cover water blocks for my nonreference 7970 cards, so I am holding off till next gen to replace them.