Fast stress test failure

[jbarcel]

I just assembled a new build, using the i5-2500k (and an Asrock Extreme4 Z77 motherboard, Radeon HD 7870 GPU, and 650W PSU). I'm fairly new to this, and I wanted to do a few preliminary tests to see if the CPU etc. were performing all right. The case is still open and a couple 120mm fans are running. I'm not overclocking, I'm using default BIOS settings, and I'm using the default CPU cooler.

When I loaded up OCCT's CPU: OCCT and CPU: LINPACK tests, they each fail within about 30 seconds. All four cores quickly and pretty equally climb up from the idle 25 C or so to about 85 C during that time period, and I get a "Core #2 over maximum value! Value Reached: 87, Max Value: 85" message at about 30 seconds. Or sometimes it's Core 1 that gets the message, although they're all about the same temperature. Running Prime95 results in the same thing happening in about the same time period, maybe even a little faster. I cut it off quickly when it reached 90 C. After a test is over, temps drop pretty quickly to the thirties, then a few minutes later the cores are down to about 25 C each.

Is this something to be highly worried about, considering that I'm not currently taking any special measures to get extra cooling? What do you think my next step should be?

 

[rolli59]

Try it under normal load and check the temps. That is really what matters.

 

[jbarcel]

If I'm doing a couple medium-level things at once like installing a game from a DVD and running a full-system antivirus scan simultaneously, I get upper thirties on all cores, with Core 0 a little higher at maybe 40 C.

I installed a game I had lying around, Alan Wake, since it's a couple years old but I remembered it was supposed to make pretty good use of the cores. When moving around and such in windowed mode, the cores were all upper sixties, with an occasional lower-seventy. The game was running in windowed mode at 1366x768 (so I could see CPUID HWMonitor in the background) on high settings, while the computer is 1920x1080 on a 55" HDTV at the moment.

What do you think? Based on these temperatures, is the CPU acting abnormal? Am I in for trouble, particularly when gaming?

 

[rolli59]

I would check the cooler install since that is on the high end.

 

[sykozis]

The stock Intel cooler isn't designed to handle the types of loads that software like OCCT, Intel Burn Test or Prime95 can generate. Using these programs with the stock CPU cooler can result in damage to the processor.

 

[jbarcel]

If I remove the cooler temporarily to try to reseat it, do I have to wipe off the thermal compound the cooler came with and apply new thermal paste? Or will the compound still be fine for when the cooler is put back on?

 

[Scott_D_Bowen]

Your Asrock Extreme4 Z77 motherboard is probably automatically overclocking and/or overvolting the CPU too much.

Disable that in the BIOS and you should never get over 85C at max load.

If it continues then the CPU cooler should be looked at next.

 

[azathoth]

Your Asrock Extreme4 Z77 motherboard is probably automatically overclocking and/or overvolting the CPU too much.

Disable that in the BIOS and you should never get over 85C at max load.

If it continues then the CPU cooler should be looked at next.

No options have been tinkered with.

He/She is using the Stock Intel Cooler, which is only meant for normal loads at normal clocks.

jbarcel, I HIGHLY recommend you purchase an aftermarket cooler such as the Hyper 212 EVO.
This air cooler is under $30 and you can easily hit at least 4.2Ghz with it. At those speeds your temperature will still be significantly lower then your current stock Intel Heatsink.
http://www.newegg.com/Product/Product.aspx?Item=N82E16835103065
Currently on sale for $20!

Whenever you remove a heatsink, you need to reapply thermal paste too.
I recommend Arctic Silver 5, or IC Diamond.

 

[bigcyco1]

Did i read that correctly?Your trying to overclock on a stock cooler? Overclocking is the act of changing the clock speeds or frequencies of a certain component in a computer. It is usually done in the BIOS for CPUs/Memory, or using a specialized tool in Windows for video cards (such as MSI After Burner). Overclocking a component increases its speed, but if it is not done correctly, can make the computer run unstable, or can even damage components. Temperatures are the main thing to look out for while overclocking. If you do not have adequate cooling, the components will at some point overheat, which can cause failure. You also have to keep the voltages in check. Overclocking a component too far with stock voltage can cause instabilities. If the BIOS lets you, voltages can be changed for the CPU and other hardware. Increasing the voltages of a component will increase its heat output significantly.


Long story short, don't overclock if you are not comfortable doing so. Because doing something wrong can turn out to be very bad. Including giving a component far too much voltage, and literally frying said component.
Some manufactures have more lenient restrictions on their warranties when it comes to overclocking. Read your warranty well.

 

[jbarcel]

I'm not overclocking, just trying to make sure the CPU works. The BIOS is using default settings and it doesn't seem to be automatically overclocking the CPU, so I don't think that's it. Oddly, I either forgot or never heard that that the CPU stress tests don't work at all on stock coolers.

Still, based on the video game performance (mentioned in my 2nd post), it seems like the CPU should be running cooler than it is; I don't know if that means something's wrong or not. I'll buy the Hyper 212+ and try that out.

 

[sykozis]

The stock Intel cooler is not designed to handle the types of loads generated by stress testing software. Stress testing software is designed to simulate "worst case scenarios"...which are the most unlikely scenarios during normal usage.

At default settings, ASRock Z77 Extreme4 will increase CPU core voltage under load unless you specify a core voltage manually. Running Intel Burn Test, I've seen my CPU core voltage increase to 1.45V and 85C within a few seconds of starting the stress test.

 

[jbarcel]

That's interesting, I'll have to look into that. Should I even change that core voltage option within the BIOS, if I'm going to get a Hyper 212+ but not overclock? If so, which setting under CPU Voltage would have to be changed, and what's a good thing to set it to?

 

[azathoth]

That's interesting, I'll have to look into that. Should I even change that core voltage option within the BIOS, if I'm going to get a Hyper 212+ but not overclock? If so, which setting under CPU Voltage would have to be changed, and what's a good thing to set it to?

Under CPU Voltage, there should be an option simply called fixed voltage. You only need to raise it if your system begins to crash, or fails a stress test such as Prime95.

Basically in short it goes Raise Multiplier, Run stress test, if it is stable. Raise again.
If your system begins to crash, or prime95 throws errors, raise the Fixed Voltage by a slight degree and run the stress tests again.

Repeat until you hit a clock speed you are comfortable with, and ensure your temperatures do not exceed thermal limitations.

Should be able to hit 4.0 to 4.2 Ghz on the stock voltage setting, I would recommend aiming for 4.5Ghz if you can which seems to be a sweet spot.

 

[sykozis]

If you plan to run at stock clocks, change the CPU Vcore option to offset voltage and change the value to -.005v and you should be good. Or you can just set fixed voltage to 1.24V and be safe.