TR Forums

I finally got my i7-4770k upgrade installed with an ASRock Extreme6 board. I was able to use my Cooler Master Hyper 212 Plus HSF (with push and pull fan setup) with the Haswell chip. I set up the computer to take advantage of all power saving methods it can via the UEFI so no manual overclocking at all so far. Im getting idle temp readings at around 29 C using Core Temp. I ran Prime95 for a couple of minutes and the CPU temp went up to 72 C immediately. Is this on par with what some of you other guys are getting?

Don't get freaked out by those temperatures. I'm OC'd to 4.7GHz and my hottest core can hover just below 85C during the worst parts of the Linpack torture test (mprime is a little cooler in the low 80's). The chips are stable in that temperature range, although I wouldn't go out of my way to run them that way 24/7 either. If your experience is similar to mine, you'll see that the chip idles at much much lower temperatures too.


Coming from a Core 2 Duo where the chip would idle in the mid-40 C range but only get up to about 65C under high load, it's a change, but it works OK for me.

Mine runs at 55°C to 57°C running Prime95 with 2 CPU fans and 1 chassis fan on high (even after 45 minutes). With my normal noise-optimized fan profile, it runs closer to 70°C.

JustAnEngineer wrote:

Mine runs at 55°C to 57°C running Prime95 with 2 CPU fans and 1 chassis fan on high (even after 45 minutes). With my normal noise-optimized fan profile, it runs closer to 70°C.

OC'ed?

Overclocked 4770k @ 4.9Ghz using watercooling (ok it's not actually water but antifreeze.) - run to a fridge unit.

I am stable at 36C max running the cpu at 100% during a Linpack torture test. Idle is 15C. Room is between 23 and 22C. I'm pretty damn happy with it.

keltor wrote:

Overclocked 4770k @ 4.9Ghz using watercooling (ok it's not actually water but antifreeze.) - run to a fridge unit.

I am stable at 36C max running the cpu at 100% during a Linpack torture test. Idle is 15C. Room is between 23 and 22C. I'm pretty damn happy with it.

Those temps are impressive. Why the need for that much cooling?

I have i5-4670 at stock speed with stock cooler and it runs with exactly the same temps as yours - 29 to 72 C.

keltor wrote:

Overclocked 4770k @ 4.9Ghz using watercooling (ok it's not actually water but antifreeze.) - run to a fridge unit.

I am stable at 36C max running the cpu at 100% during a Linpack torture test. Idle is 15C. Room is between 23 and 22C. I'm pretty damn happy with it.

I bow to your refrigerated machine! It has amazing performance, temperatures, and doubles as a beer cooler!

OP - I am guessing you have PWM or other motherboard-controlled fans on your cooler? The CPU is just heating up to its standard thermal envelope before the fans start increasing in speed is all. If you want to see, go into BIOS and set the CPU fans to permanent 100%, not 'auto', and your temps should be lower.

I tweaked the fan profile to spin up my chassis fan a bit more aggressively so now I get around 60°C in Prime95 (without overclocking) and I still get a very quiet system (fans at 270-300 rpm) when just browsing the web.

Topic moved to OC/Cooling forum.

Can you help me understand the i7-4770K temperatures I'm seeing?

For example, I just ran Prime95 (Small FFT's Torture Test) and the Asus Probe II Sense app reported 79°C for the CPU and 31°C for the motherboard (an Asus Z87-A, stock cooling, no overclocks).

79°C sounds like a normal CPU temperature to me, but the official Intel specs only mention temperature to say that the T_CASE max temperature is 72.72°C.

T_CASE is defined as: "Case Temperature is the maximum temperature allowed at the processor Integrated Heat Spreader (IHS)."

Is the Prime95 temperature of 79°C likely to damage my CPU?

No.

OK but what temperature can damage my CPU? Or do modern CPU's do fancy throttling to make temperature damage impossible?

sparkman wrote:

Or do modern CPU's do fancy throttling to make temperature damage impossible?

Yes. At the very worst your PC will simply reboot itself.

JohnC wrote:

sparkman wrote:

Or do modern CPU's do fancy throttling to make temperature damage impossible?

Yes. At the very worst your PC will simply reboot itself.

If it happens once or twice then it's not a problem.

However, sparkman, keep in mind that even though the CPU protects itself, it's still running hot and will die sooner rather than later. And if that doesn't happen, then the motherboard goes because it's also subject to a lot of heat around the CPU area.

I upgraded my 4770K from the stock heatsink to a Hyper 212 EVO.

The same Prime95 (Small FFT's Torture Test) as above now peaks at 52°C. If I add Unigine Valley on top of Prime95, the CPU gets up to 59°C. All of this with only a quiet whirring from the three fans in the case (2 case fans, 1 fan on the Hyper 212 EVO).

Sweet. I love this build! The quietest, fastest, screamer of a gaming PC I've ever owned.

The GTX 770 GPU got up to 77°C during the Prime95+Unigine Valley benchmark. I have two more case fans on order to improve total airflow and hopefully bring that 77 down.

With the automatic throttling built into modern chips it's nearly impossible to actually damage them from heat without really trying.

A 4770K will throttle aggressively once it hits 100 C; anything below that and it'll happily run all day.

Waco wrote:

With the automatic throttling built into modern chips it's nearly impossible to actually damage them from heat without really trying.

Keep in mind that even if it's not throttling, when running very hot it will heat up the area around the CPU socket, damaging VRMs and other power circuitry, the board and socket itself, etc.

Also, the lifetime of the CPU will probably be severely diminished.

morphine wrote:

Keep in mind that even if it's not throttling, when running very hot it will heat up the area around the CPU socket, damaging VRMs and other power circuitry, the board and socket itself, etc.

Also, the lifetime of the CPU will probably be severely diminished.

I doubt the heat from a 4770K running in the 90s is enough to damage anything at all especially since the VRMs on the motherboard will run up past that temperature with ease. In fact, I'd be surprised if it really heated up anything other than itself (and maybe the socket) to any real extent. It's not like the backside of the CPU is ever cooled at all even with the best of coolers.

I guess we'll see what the long-term effects are for Haswell chips but I know there are a great many servers and HPC systems that sit with their CPUs running near throttling temperatures 24/7 for years on end without many issues. Intel didn't set (or at least I wouldn't expect them to have) the throttling limit at a temperature that would cause them to fail in less than the warranty period even if run full tilt at throttling for the entire warranty period.

Waco wrote:

but I know there are a great many servers and HPC systems that sit with their CPUs running near throttling temperatures 24/7 for years on end

That's true, but that's part of why server chips cost more. Intel qualifies the server silicon at a higher level suited for 24/7 operation. Desktop chips don't necessarily meet that standard.

In general I think you are both right. Heat can degrade the components over time, eventually causing failure. But that doesn't mean running a modern chip at 90C is necessarily a bad thing. You just might be shaving off the tail end of the chip's lifetime.

Sure, it causes some degradation, but I don't believe that degradation is worth being concerned about on a chip that you'll likely use for less than a decade.

To be completely fair, this is only a concern in Average Joe's case that will be poorly ventilated to begin with (so damage to the mobo is more likely), and/or on a laptop, for obvious reasons.

On a well-maintained and ventilated desktop case, it should be okay.... ish.