Cooling Solutions for Intel Core i7

Written on February 9, 2009 by
Share:

Introduction

The latest powerhouse CPU offering from Intel is here. The Intel Core i7 -- a quad-core processor available in three different speed configurations that is really taking the computing world by storm. Several new features have been added to this processor, such as on-chip DDR3 memory controller, smart cache, and HD boost. Of course, with all the extra features and power comes the issue of how to keep it cool. The Core i7 may be powerful, but it is also very hot running. From the stock heat sinks and fans, to liquid cooled solutions, the cooling possibilities are many. Unfortunately we can't test them all, so in this article we'll take a look at 4 popular cooling solutions and how they fared.

Test Setup

We ran two tests per cooler, one with the case fans on low speed and the other with the case fans on full blast to also help us better understand the effect that case fans have on the cooling of a system. We tested these cooling solutions with the following hardware:

  • Antec P182 (3x120mm tri-speed case fans)

  • Asus P6T Deluxe Mainboard

  • Intel Core i7 Quad Core 920 2.66 Ghz

  • 3 x Kingston DDR3-1333 2GB

  • Corsair 650watt PSU

  • Western Digital 640GB Hard Drive

  • Asus GeForce 9600GT Video Card

  • Windows Vista Ultimate 64-bit

  • Asus 20x SATA DVD-RW Lightscribe (x2)

  • Tuniq TX-2 Thermal Compound
The ambient temperature was climate controlled, always within one degree of 19 degrees Celsius. After installing the cooler, the system was allowed to warm up before our temperature logging was started. To put the system under as much load as possible, we ran Prime 95 Torture Test, along with FurMark 1.6.5. After the stress test was complete, the system was given two minutes to cool down before the temperature logging was stopped.

The Coolers

Stock Intel Core i7 1366 CPU Fan

This is the standard fan that comes with the Core i7 CPU. Designed for quiet operation and enough cooling to get you by at stock speeds. Surprisingly, in our test this cooler didn't fare the best, but it wasn't the worst!



Nexus LOW-7000

The Nexus LOW-7000 cooler combines some of the most advanced technologies in the market used for CPU coolers. Heat pipes, SkiveTek and ultra-thin fins combined with a real silent  PWM controlled fan delivers ultimate performance which has never been this quiet before. The 120mm fan is PWM controlled and the speed can vary between 500 and 2000 RPM creating a noise-output of an inaudible 15 dB(A) up till a still truly quiet 24 dB(A)

We were interested in this cooler for our low profile HTPC setups. I was fairly surprised at the result we saw on this one. At an average of about 90º Celsius, it was the lowest performing solution we tested.

Gelid Silent Spirit

The top-flow cooler “Silent Spirit” follows an open frame structure concept whereby both design and dimensions have been improved to eliminate humming and buzzing noises while maintaining a compact overall size. The 92mm fan has been chosen for silent operation and high airflow and comes with an intelligent PWM control to allow users to set the fan speed in their BIOS accordingly. In addition the anti-vibration fan mounts are made of high quality rubber thus guaranteeing a long life span even during high temperatures.

This one was the second best in our testing, it didn't quite beat the Hydro, but at only 2º C higher than the hydro and almost 20º C lower than the Nexus, and at a retail cost of around $35, it certainly is a very nicely priced cooling solution.

Puget Hydro CL1

The Puget Hydro CL1 is a self-contained CPU liquid cooling solution. It has a lot of the benefits of liquid-cooling: better temperatures (which allows for overclocking) and fairly quiet operation, but also has the benefits of air-cooling: the ease of installation and maintenance. The radiator screws in place over the rear 120mm fan in your case. By default, the rear fan is usually left as an exhaust. Through our testing, we have found that in cases like the Antec P182 (which has a top exhaust fan), the cooler performs much better by reversing the rear fan and turning it into an intake. Since the case has a top exhaust fan, the impact on system temperatures is acceptable.

Ah, the almighty Hydro. It performed better than all the others and kept the CPU an average of 12º C cooler than all the other ones tested. The Gelid wasn't far behind but the Hyrdo still proved to be one of the best cooling solutions.

Results

These temperature graphs are what we normally use internally to examine the temperatures of each new system we ship. While not all that easy to compare to each other, they do provide a wealth of data. If you would like to see direct comparisons between the different coolers, please skip ahead to the next section.

Note: the Nexus temperatures top off at 100C because that's the point at which the CPU throttles its speed. Not good!

Cooler Comparison



These graphs represent the idle and load CPU temperatures over the course of testing. It's easy to see the winners and losers here -- the Nexus ran much hotter than the others while the Puget Hydro CL1 was the best. The comparison between the Intel Stock heatsink and the Gelid Silent Spirit is interesting. Though the Gelid only cooled a few degrees better, it was significantly quieter. While the Puget Hydro CL1 is the clear pick for cooling performance, the Gelid Silent Spirit could also be a good choice in lower power situations when noise level is a concern.

Conclusion

In terms of our personal recommendations, the Puget Hydro gave us the best overall results, the Gelid Silent Spirit ran a close second. Considering price and noise levels the Gelid certainly is a good solution for the i7. The biggest surprise here was the Nexus LOW-7000. At an average temperature above 90º Celsius I wouldn't consider this a good option for Core i7, though it still could be perfectly adequate for lower power 775 CPUs.

Tags: Processors, Cooling, Performance, Products
boomer

I was wondering what the Core i7 runs at with liquid cooling? You indicated the Core i7 is "very hot running." Would a liquid cooling setup, such as your extreme cooling package, match or surpass by a significant factor the cooling solutions in your tests? Thanks, Bob

Posted on 2009-02-11 20:12:20

Yep, the extreme liquid cooling setup will definitely be a better performer than any of the air cooling setups (including the Puget Hydro). The extreme liquid cooling has 4x120mm radiators all dissipating the heat, which is at least 4x more surface area than any of the air cooling setups.

The thing that makes it not so cut-and-dry is that with nearly all liquid cooled setups, the video card(s) are included in the liquid cooling loop. So while the cooling power is significantly greater, the CPU temperatures are about the same, or only 5C or so less than seen with the Puget Hydro. The difference, of course, is that the video cards run a lot cooler.

In the tests we did for this article, a fairly low end video card was used. If we had put this exact system in one of our extreme liquid cooling systems, I'd expect pretty significantly improved CPU temperatures. I just don't have any data to tell me exactly what temperatures to expect, because with all the Core i7 systems we've built, our customers have only gone to liquid cooling when the system also used high end video cards. And that makes sense -- the liquid cooling isn't needed so much when you're talking about just cooling the CPU (the temps seen with the Puget Hydro were great!).

Posted on 2009-02-11 20:32:39

https://www.pugetsystems.com/g...

above for graph....

just wondering if that long of a high temp was a good idea or not? thats 8 hours 39 minutes of 174 farenheit. we ran furmark but only for short time. normally idle temp is 39/39/42/38.

your answer greatly awaited and check out below screen dump for sore temp....shows engineering sample for my 940??? must be an app isssue huh?
===================================

---------------------------------------------------
Core Temp version 0.99.4.65 - 64bit
---------------------------------------------------

Number of processors: 1
Number of cores: 4
Number of threads: 8

Processor information:
---------------------------------------------------

Processor 0
Package 0
Cores: 4 (Max 8)
Threads: 8 (Max 16)
Processor name: Intel Core i7 940 (Bloomfield) (Engineering Sample)
Processor name string: Intel(R) Core(TM) i7 CPU 940 @ 2.93GHz
Platform: LGA 1366 (PlatformID: 0x1)
Processor signature: 0x106A4
Processor revision: C0/C1
Process node: 45 nm

Processor speed: 1603.64MHz (133.64 x 12.0)
Bus speed: 3207.27MHz (6.41 GT/s)
Bus ratio: x24
Current TDP: 37.9 Watts
Default speed: 2933.00MHz
Multiplier range: x12.0 - x22.0
Instruction set: MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T
Feature set: VT, XD, EIST
L1 data cache: 4 x 32 KiB, 8-way set associative, 64-byte line size - 2 thread(s)
L1 instruction cache: 4 x 32 KiB, 4-way set associative, 64-byte line size - 2 thread(s)
L2 cache: 4 x 256 KiB, 8-way set associative, 64-byte line size - 2 thread(s)
L3 cache: 8192 KiB, 16-way set associative, 64-byte line size - 16 thread(s)

Register information:
---------------------------------------------------

Posted on 2009-02-14 15:21:22

jon during testing on my system we ran prime 95 for 8 hours 39 minutes with 0 errors (overnight but i kept an eye on it...not sure how accurate coretemp app is screen dump for it below post .... coretemp app never got above like 79c....is that an excessive torture test for i7 940 on my system? i see you guys only log it for 1 hour on the online logs

It depends on what you're testing. If you're testing to see whether your CPUs were stable under that temperature, then I'd say you ran a good test -- 8 hours seems about right.

If you were testing to see what your cooling system would max out at, then you definitely ran it longer than you needed to. 15-20 minutes would have been enough for that :)

Posted on 2009-02-15 13:00:51

It depends on what you're testing. If you're testing to see whether your CPUs were stable under that temperature, then I'd say you ran a good test -- 8 hours seems about right.

If you were testing to see what your cooling system would max out at, then you definitely ran it longer than you needed to. 15-20 minutes would have been enough for that :)

we were testing i assume to rule out heat as what was causing the few shutoff/lockups that would also blowout the bios settings... (i believe i cc'd you the email).

dan was great working thru this with me and updating my bios to asus most recent one which was like 4 revisions since my xmas build. i am crossing my fingers it was bios related and not power supply so it doesn't have to travel cross country again.

i found it weird that we never went above a certain temp 75-79celcius. my room is like 75-80 farenheit this time of year... i hate the cold. would that threshold be related to your asetek/puget cpu cooling you wrote about?
is there a limit these i7's cpu's take? if there is i guess we can rule out heat for my problem.

either way nothing could have been damaged running for that long rite? i mean all seems fine with video card (updated those drivers also and processors). just never ran any tests other than hd tach and siSandra in the past niether of which ever ran anywhere near that amount of time. nor does it seem that you all run prime 95 that long either for a torture test.

and i will add richard was awesome sending me new side panel (which like you told me at xmas time was a piece of cake to replace) if you check it out you will see the beating this took from fedex.... ups for me always ;) but you had already viewed pics of box so you knew this.

Posted on 2009-02-15 17:04:05
jeffmorris

Should I buy the Puget Hydro CL1 CPU cooler?

Posted on 2009-02-16 02:43:50

Should I buy the Puget Hydro CL1 CPU cooler?

your graph showed all was okay so i wouldn't know but if you needed it they probably would have recommended it as they are on top of things here.

Posted on 2009-02-16 05:05:02

Should I buy the Puget Hydro CL1 CPU cooler?

You'd see an improvement, but you don't have a problem. Your system is cooled adequately as is. With a Puget Hydro CL1, it'd be cooled even better, but the bigger change would be that it would be quieter.

Posted on 2009-02-16 09:29:45
amityeric

Yep, the extreme liquid cooling setup will definitely be a better performer than any of the air cooling setups (including the Puget Hydro).

Is the hydro an air-cooled setup, or a liquid-cooled setup?

Posted on 2009-04-13 06:18:31

You can think of it either way :) I tend to think of it as an air cooler, because it installs like one, but it technically is a liquid cooled setup. It is a closed loop liquid cooler, with a mini pump and reservoir right on the CPU block, and a 120mm radiator. It needs no maintenance because it is closed loop, and we've never once seen one leak.

Posted on 2009-04-13 10:30:23
davisthediver

I have read and worked with I7 cpus that were cooled by the Asetek LCLC. IMHO this unit is THE unit for this cpu in a full tower case. The only problem I have run into so far is the lack of ANY mfg. manual, especially pertaining to reseating the LCLC when changing out Mo Boards.

I am not sure of the quality or lifespan of the original thermal compund but am wondering if AC-mx-2 or similar product would improve the effective cooling on this unit?

I have found that cooling is better when using the fan as a pusher instead of the more commonly seen puller.
I have had to hand cut light foam seals to make sure there is 100% of the pushed air going through and not leaking around the radiator regardless of the way the fan is set.

To uncloud my questions- 1. Manual for LCLC? and 2. Thermal paste improvement recomendation?

Posted on 2009-05-26 20:49:37
gnustatjadfa

Hello, I'm planning to build a pc with an i7 quad core processor but I'm not planning to overclock it or anything so is using the stock fan good enough? I'm struggling to keep the budget low.

Posted on 2012-06-03 01:49:17

In general, we consider most stock coolers for i7 CPUs to be inadequate, although it somewhat depends on the CPU. Socket 1155/1156 i7 CPUs might be OK with stock cooling as long as the airflow in the chassis is very high, but Socket 2011 i7 CPUs we have found to overheat often with stock cooling. Technically, they usually keep the CPU below Intel's thermal threshold of 100c at full load, but we always like to keep CPUs below 80c whenever possible. It helps prolong the life of the CPU as well as give it plenty of thermal headroom for turbo boost. To be clear, we really stress our systems, so in normal everyday use, stock cooling might be OK, it just depends on what you are doing with the system.

As for cooler recommendations, if you are on a budget I would recommend the Cooler Master Hyper TX 3 for a socket 1155/1156 CPU, and either a Gelid Tranquillo Rev2 or Gelid GX-7 for a socket 2011 i7 CPU. Sure,  they will cost a little bit of money, but in the long term it will be well worth it.

Posted on 2012-06-04 19:55:55