Core i7 4770 w/ stock cooler - 100 C during Intel CPU stress test

[CNelsonPSU]

Hi all,

I recently built a new machine with an i7 4770 and the stock Intel cooler that came with the processor - I am not overclocking anything. My motherboard is also Intel - the H87-based DH87MC, and I'm running on Windows 8 x64 w/16 GB 1600 MHZ RAM. Power supply is a 1000W Thermaltake.

The cooler is quiet and such during normal use (internet browsing and such), with the processor hovering around 38-40 C, but I noticed after testing out Crysis that the fan got fairly loud and temps were 80 C or so.

To see how the cooler was doing under load, I downloaded the Intel Extreme Tuning Utility and ran the CPU stress test. After about 3 seconds (!) into the test, it showed my CPU temp hitting 100 C even with the fan cranking away at what seemed like full speed, and I started getting warnings (critical temp seems to be 90 C).

I bought a Noctua NH-L12 (1) cooler, but I mistakenly bought RAM with high fins so it won't fit. I am returning the RAM and getting lower profile stuff / no heatsink fin RAM so I can put it on, but of course I wanted to try my machine so I just put the stock cooler on for now. But now I'm worried that maybe the processor, MB, or PSU are faulty. Should this thing really be shooting to 100 C so quickly?

Thanks for any help you all can provide!

(1) http://www.newegg.com/Product/Produc...irtualParent=1

 

[Soccerman06]

Did you use any TIM or does the stock heatsink come with it preapplied? Either way reseat the heatsink

 

[CNelsonPSU]

Did you use any TIM or does the stock heatsink come with it preapplied? Either way reseat the heatsink

Yeah, on the stock cooler they have some pre-applied. So even if the CPU is OK most of the time at 38-40 C, reseating the heat sink could still help it under load?

I'm also hoping that when I put on the Noctua, it will be much cooler. If not does that mean something else in my system could be causing problems and/or the chip is faulty?

 

[Idontcare]

Yeah, on the stock cooler they have some pre-applied. So even if the CPU is OK most of the time at 38-40 C, reseating the heat sink could still help it under load?

I'm also hoping that when I put on the Noctua, it will be much cooler. If not does that mean something else in my system could be causing problems and/or the chip is faulty?

It would be unlikely that anything else in your system would cause the elevated temperatures (like ram or PSU), but there are a few things that likely to cause it and you should check them out.

For starters, your ambient temperatures (how hot is the room which the computer is in?) as well as the air flow within the computer case are critical in determining how hot your CPU will get under load.

One way to check case air flow is to do an extreme test in which you take the side off the case and position an external fan (those big ones you buy to cool a room down) to blow large amounts of air straight into the case. Then run your stress test. See if the CPU temperature is much better, or not. Then you will know how much (if at all) your case air-flow is the problem and you can choose to proceed from there (new case, more case fans, etc).

A second thing is definitely the TIM (too thick or if it gets an air-bubble) and the fact that the stock cooler is notorious for only ending up with three out of the four latches actually being fully latched. The only way to really confirm that all four latch pins on the stock cooler are fully secured it to attach the stock cooler before you install the motherboard into the case (because you can visually inspect the backside of the mobo to confirm all the pins are latched).

Even with as much experience as I have in swapping heatsinks I still have trouble getting all the push-pins to latch, I really have to be careful about it. And when three of them are latched you can't just wobble it back and forth and know the fourth is not latched, it is physically secure but not good enough.

In my mind the most likely culprit for your temperatures is that the stock HSF is not fully latched for all four push-pins.

 

[CNelsonPSU]

In my mind the most likely culprit for your temperatures is that the stock HSF is not fully latched for all four push-pins.

Thanks, I'll check that out. Luckily the Lian-Li case I have has a little window cut out of the motherboard tray so I can see / mount CPU coolers and brackets. They looked all fully latched, but I'll check it out.

I have some low profile memory coming on Saturday, so I may just wait to install the Noctua NH-L12 cooler and see how that performs before continuing troubleshooting.

Good to hear it's probably not other components. Thanks!

 

[Idontcare]

Any particular reason you went with the low-profile Noctua cooler? (thus necessitating the low profile ram)

Not a critique, I'm just curious.

 

[CNelsonPSU]

Any particular reason you went with the low-profile Noctua cooler? (thus necessitating the low profile ram)

No real reason...I guess it was just the one I started finding reviews for, it looked good, and people seemed to universally like it. I didn't realize its heat sink would be so low or that the RAM I bought was so tall! I like the look of the LP RAM better anyway

 

[JimmiG]

100C doesn't sound unreasonably during stress testing with the tiny stock Intel cooler. Linpack, Prime95, Aida64 etc. will produce temps at least 20C over what you get during normal use. This is why many overclockers choose to ignore those temps and call their overclocks a success anyway. Wouldn't surprise me if Intel also were to ignore those stress tests and just provide a cooler that is only good enough for gaming and normal use.

 

[ShintaiDK]

The stock cooler shouldnt go much beyond 85C with stock settings.

Tjmax is 100C and the CPU will throttle there.

 

[CakeMonster]

My Gigabyte Z87-D3HP would just throw vcore at the cpu with the BIOS it came with. I noticed it pulling more than 1.4v at times without changing the Auto setting. I've reflashed and oc'ed and locked it way lower now. Check your voltage and adjust down in the BIOS if necessary. At stock speeds you should be fine with 1.15 ish.

 

[CNelsonPSU]

The stock cooler shouldnt go much beyond 85C with stock settings.

So I took off the stock cooler, cleaned it and the chip, put some Noctua thermal compound on, and put the stock cooler back on. My temps are now acting normal - the CPU stress test shows a more normal temp ramp up to around 83-86 C.

I'm assuming I didn't have the heat sink sitting on the CPU correctly. Still putting on the Noctua when I get it this weekend, but at least I know nothing is wrong with my CPU or mobo.

Thanks all!

 

[Cardio]

"Even with as much experience as I have in swapping heatsinks I still have trouble getting all the push-pins to latch, I really have to be careful about it. And when three of them are latched you can't just wobble it back and forth and know the fourth is not latched, it is physically secure but not good enough."

Couldn't agree more! Why Intel continues to use this AWFUL method of connection is hard to understand. With as many talented designers and engineers as they have and year after year ignore this problem is just stupid!

 

[Remobz]

Good read....

 

[Idontcare]

So I took off the stock cooler, cleaned it and the chip, put some Noctua thermal compound on, and put the stock cooler back on. My temps are now acting normal - the CPU stress test shows a more normal temp ramp up to around 83-86 C.

I'm assuming I didn't have the heat sink sitting on the CPU correctly. Still putting on the Noctua when I get it this weekend, but at least I know nothing is wrong with my CPU or mobo.

Thanks all!

Glad to hear it was something that was a "quick fix" to resolve, much better than just living with it or having to send it back to Intel for an RMA.

 

[crashtech]

Still hating the HSF retention systems in use today. LGA 771 used a very nice system with large through holes in the mobo, mounting the HSF directly to the tray, as today's heavier units should be. A simple and inexpensive adapter could screw to the stock ATX standoff locations and provide mounting standoffs for the HSF. I'd be thrilled if they would return to this method.

 

[bryanW1995]

Still hating the HSF retention systems in use today. LGA 771 used a very nice system with large through holes in the mobo, mounting the HSF directly to the tray, as today's heavier units should be. A simple and inexpensive adapter could screw to the stock ATX standoff locations and provide mounting standoffs for the HSF. I'd be thrilled if they would return to this method.

The stock intel HSF system has sucked for several generations. How much would it cost them to run 2 skus, 1 just oem (no hsf at all) and 1 with something like a hyper 212+? Maybe $10 each? They could charge $25 extra for the cpu + hsf combo and make extra profits, and coincidentally stop pissing off system builders.

 

[james1701]

They did have the OEM tower cooler that was included on the 980X. That would work really good for OEM CPU's out now compare to the junk that is out now.

 

[Machiavegli]

The stock intel HSF system has sucked for several generations. How much would it cost them to run 2 skus, 1 just oem (no hsf at all) and 1 with something like a hyper 212+? Maybe $10 each? They could charge $25 extra for the cpu + hsf combo and make extra profits, and coincidentally stop pissing off system builders.

Intel doesn't care because it doesn't have to. It knows that almost anyone who knows anything about heatsinks will install their own aftermarket even if Intel invested significant money into making their heatsinks better or selling package deals of good heatsinks. In fact, looking over my collection of Intel stock heatsinks, it looks like they are just getting worse with every year.

 

[mlody]

I had exact same issue with also i7-4770 and H87 board (Asus though).
As per some advices in my original thread, I re-seated the Intel cooler and also double checked all the push pins, but none of that improved the performance. My CPU in under 5 sec would shoot over 100C during Prime95 test.
Today, I installed Cooler Master 212+ and I immediately noticed significant improvements. So far I think the highest temp I noticed was around 75C after 15 minutes of Prime 95 testing.