i7 4790k at 40C idle. Is that high?

[sakete]

Hi guys, recently built a i7 4790K rig with Noctua NH-D15 cooler.

Currently it's idling at around 40C. Is that normal? Or should I reseat the cooler and redo the cooling paste? (used Arctic Silver 5).

By the way, I just played a game that took the temps up to about 70C, but it stays idling at ~40C. I did do a system OC to 4.6 GHz at 1.3V, but have that as the Turbo (adaptive) voltage. At idle it's at 0.71V.

The case I'm using is a Fractal Design R5 with two intake fans (front and bottom) and one exhaust fan (top rear).

See screenshot here:

 

[ignatzatsonic]

What's your ambient temp when it's idling at 40?

The 4790K is known to have heat issues in some situations, but I can't say about yours with an NH-D15 at that clock speed and voltage.

There's certainly nothing wrong with 40 idle for eternity.

 

[alcoholbob]

You should be able to hit 4.6 at under 1.25V. I have a 4.8GHz overclock at 1.22V.

My idle is 26-27C, although during gaming it will hit around 65C. Your load temps are fine but idle is a little high for a Noctua D15. In fact even on the Intel stock cooler I idle around 33C. It may be voltage related or it could be high ambient temps.

 

[ShintaiDK]

Its completely normal depending on cooling and fan control.

Idle temperature as such doesnt matter at all.

Mine idle at 40-45C with stock cooler and ~1000rpm in idle.

 

[Fox5]

I'd gone through hell trying to cool my 4790k.

As it is, it idles around 40C, but it'll reach the threshold temperature in OCCT (85C) in about 5 minutes. That's better than where it was when I started, where it instantly reached 85C, and later only took 1 minute. I haven't been able to reproduce those temps in real world or in other stress testing, so I suppose it's really just turbo speeds + 4 core AVX workload makes the cpu dump out ridiculous heat, way more than the 88W TDP.

 

[Ketchup]

Yeah, she (4790k) gets pretty hot in occt and the Intel burn test at just about any speed (using cm 212 evo). Regular gaming doesn't get as hot, but idle is about what everyone else is seeing. You might want to check background apps.

 

[ButtMagician]

That case doesn't have the best airflow to say the least. I have the same fan config as you, and after I went from an old Antec 300 to Fractal R5 my idle and load temps both increased noticeably.

The price you pay for silence...

 

[sakete]

Well I've decided to lower the OC voltage to 1.25V and OC'd it an additional 100MHz to 4.7GHz (turbo speed). Did an AIDA64 / OCCT / Intel XTU stress test and no immediate crashes or hangs (did 10 mins each). Will run a longer stress test overnight. But temps didn't go above 79C when running those tests (I made sure voltage was manually set to 1.25V, but changed it to adaptive voltage after the stress test).

Still getting about 40C idle though. Plan on attaching the case fans to the motherboard instead of using the case controller and will setup a profile for the rear exhaust fan to spin up a bit quicker so it can blow out all that hot air.

 

[videogames101]

It will idle at 40°C without a problem until the heat death of the universe, absolutely nothing to worry about.

 

[bbhaag]

This thread makes me wonder if there is a better solution to cooling. As dies get smaller the ability to dissipate heat seams to be a greater challenge and thermal limits seem to stay steady with every shrink.

I know this is a little bit OT but does anyone know how the industry is adapting to this? I know water cooling can do better but it's not for everyone and certainly not something OEMs want to take on.

 

[BonzaiDuck]

I'm always a bit put off by the stylish-looking cases like the Fractal that don't "look like" there's much front intake. I'm sure it sucks air through the front, but the front panel looks obstructive compared to some other cases.

You can HAVE silence, but with better intake airflow. You will also attend to more frequent duty with an ED500 blower or other dusting assistance.

Like I said -- another member got IBT or similar load temperatures around 56C at the regular 4.4 turbo clock, and he was showing around 70C+ at maybe 4.6. He was using an H80 cooler ducted to rear exhaust in the small C70 case with good 140mm intake fans wherever he could put them.

A bigger AiO would need the advantage of the ducting he'd done with the H80, and the H80 is just compact enough to make it possible in that case. You could probably put an H110i at the case front of a C70, but it would alter the need for more intake fans, so something else would be in order. I think the C70 would allow for a dual-radiator setup, though.

I still can't make up my own mind about whether to go Devils Canyon or 5820K "E" with an X99 board this year. Since I can't make up my mind, I keep putting it off while I spend my time gaming on my 2700K system.

I just feel like I can take all the time I want. Meanwhile, I took 1st place against the other AI cars in GRID2's Hong Kong Peak Road descent -- day and night.

There are more tweaks I can make in this system yet, just for funsy-wunsies.

 

[Ketchup]

The 4790k is a very popular chip among enthusiasts due to it's high default speed and decent price. Most have found that in normal applications it will maintain decent temps and even when it does get hot, Intel has a good system built in to prevent meltdown. Most other users are not going to be running the Intel burn test, nor overclocking the chip. The next chip will most certainly be running cooler than a 4790, so any heat problem should take care of itself.

 

[LTC8K6]

I believe mine idles at 34-35C with a 212evo and good case air flow-thru.

 

[crashtech]

My 4790K ides at 29°C with an H110, 22°C ambient.

 

[sakete]

I'm going to re-apply the thermal paste and make sure it's very evenly and thinly spread on the CPU heatspreader. By the way, do you guys recommend I use the Arctic Silver 5 (that's what I used this go around)? Or should I use the paste that was included with the Noctua?

Will also setup a better fan profile for the case fans. Especially the rear exhaust.

 

[crashtech]

I'm going to re-apply the thermal paste and make sure it's very evenly and thinly spread on the CPU heatspreader. By the way, do you guys recommend I use the Arctic Silver 5 (that's what I used this go around)? Or should I use the paste that was included with the Noctua?

Will also setup a better fan profile for the case fans. Especially the rear exhaust.

If the Noctua paste you have is NT-H1, you should use it.

 

[sakete]

If the Noctua paste you have is NT-H1, you should use it.

Yep, it's the NT-H1. Guess I'll try that instead.

 

[BonzaiDuck]

The 4790k is a very popular chip among enthusiasts due to it's high default speed and decent price. Most have found that in normal applications it will maintain decent temps and even when it does get hot, Intel has a good system built in to prevent meltdown. Most other users are not going to be running the Intel burn test, nor overclocking the chip. The next chip will most certainly be running cooler than a 4790, so any heat problem should take care of itself.

No doubt. Even so, I'd done some stress-tests with an SB and IB CPU recently at their stock voltages -- adding a couple hundred Mhz to the clocks, and using the Intel stock cooler. Even OCCT:CPU will push them toward the mid-80's C.

For OC'ing, there's an opinion that tuning up a 4790K to 4.7 Ghz doesn't offer much of a performance improvement. But it's a great processor. As you say, the default turbo speed is one factor, but you can't just measure performance in Mhz. Finfet technology and other improvements in processor design also add something in performance.

 

[Ken g6]

This thread makes me wonder if there is a better solution to cooling. As dies get smaller the ability to dissipate heat seams to be a greater challenge and thermal limits seem to stay steady with every shrink.

I know this is a little bit OT but does anyone know how the industry is adapting to this? I know water cooling can do better but it's not for everyone and certainly not something OEMs want to take on.

Half the problem is Intel using cheap thermal paste instead of solder. Besides, most OEMs are using lower-end CPUs and aren't overclocking, so this really isn't a problem for them.

For OEMs using 4790k's and/or those overclocking their CPUs, why wouldn't a closed-loop water cooler be an option?

 

[BonzaiDuck]

Half the problem is Intel using cheap thermal paste instead of solder. Besides, most OEMs are using lower-end CPUs and aren't overclocking, so this really isn't a problem for them.

For OEMs using 4790k's and/or those overclocking their CPUs, why wouldn't a closed-loop water cooler be an option?

It would -- be an option. Air-cooling with heatpipes is still viable for that chip, but I'd do my homework to find the best-performing cooler out there. The budget choice seems to be the 212 EVO -- not the best, but small enough. It does the job.

I see a lot of folks here who offer up cooler recommendations which don't score the best in comparison reviews. You only need a few independent sources of reliable comparison reviews to rank order coolers by performance. You can assess what sort of mods or adjustments you would make to enhance their effectiveness.

And a person's budget is likely an important factor, making compromises to performance acceptable. It depends on what you want to do, whether you'll accept any temperature limitations to your over-clock achievements, and live in happiness with the cooler you chose to buy.

 

[ClockHound]

Room temp and cooler intake temps aren't the same thing. A cooler can can't cool any lower than the temp of the intake air (For Air and Water - sub-zero coolers get a pass). Most often the intake temp is much higher than the room temp.

A great, cheap case airflow tool is a thermometer with a remote probe. Put the probe in the case near the cooler intake and then compare the intake temp to room temp. If the intake temp is more than 5 degrees above the room temp, there's case airflow gains to be had.

I have my OC'd 5660 Xeon in a Ghost which has a closed front door like the Fractal - when first tested with the stock case fans load temps were in the mid 70s at 4.3Ghz with a Mugen 4. Intake to room temp differential was 8-10 degrees.

After optimizing intake airflow - replacing the intakes with stronger fans and creating a clean front to back air path, the intake and room temp differential is now only 2-3 degrees. Temps at load about 10-12 degrees lower. Idle temps are 3 degrees above intake temps.

Best case cooling diagnostic tool ever and costs less than 10 dollars.

 

[ignatzatsonic]

Clockhound:

What particular remote probe thermometer did you use?

 

[ClockHound]

Some cheap drugstore digital model called: TempMinder. Not looking for the most accurate or blingie - just the differential. Seems like this one would do:
http://www.newegg.com/Product/Produ...utdoor_thermometer-_-9SIA1PC2V59960-_-Product

Pretty much any indoor/outdoor model with a probe should work.

 

[KingFatty]

I also have the Fractal Design R5 case. It has a lot of room for hard drive storage. Do you need all of that storage?

I removed the lower hard drive case. It's really easy. That frees up more airflow path from the front and gives more space inside. Maybe consider moving your floor fan up to the front panel after removing the lower hard drive cage? You might get better airflow from two fans in the front instead of one front fan and one floor fan? But definitely take out the hard drive cages you aren't using.

 

[BonzaiDuck]

Room temp and cooler intake temps aren't the same thing. A cooler can can't cool any lower than the temp of the intake air (For Air and Water - sub-zero coolers get a pass). Most often the intake temp is much higher than the room temp.

A great, cheap case airflow tool is a thermometer with a remote probe. Put the probe in the case near the cooler intake and then compare the intake temp to room temp. If the intake temp is more than 5 degrees above the room temp, there's case airflow gains to be had.

I have my OC'd 5660 Xeon in a Ghost which has a closed front door like the Fractal - when first tested with the stock case fans load temps were in the mid 70s at 4.3Ghz with a Mugen 4. Intake to room temp differential was 8-10 degrees.

After optimizing intake airflow - replacing the intakes with stronger fans and creating a clean front to back air path, the intake and room temp differential is now only 2-3 degrees. Temps at load about 10-12 degrees lower. Idle temps are 3 degrees above intake temps.

Best case cooling diagnostic tool ever and costs less than 10 dollars.

I'm confusing myself, trying to figure out the post to which you're responding. But I think I agree with all of your conclusions there.

The Mugen scored well, but not well enough for me to choose it as a cooler over some very few others. Maybe I need to look again.

I was using a similar digital thermometer in process of setting up systems, cages and fans still in operation. I think the device "went on the fritz" soon thereafter. But once one sees what is happening with those factors, you'd hardly need to take temperature measurements when making yet another DIY computer build. You would already know.