Can I passively cool an Athlon II X4 620 w/ Scythe Ninja 2?

[Leggie]

Hi all,

I need some expert input here.

I have had good results in the past with the Scythe Ninja. It did a good job of cooling my old Socket 939 3200+ Venice core; it then also, much to my surprise, passively brought down the temperature of a Q6600 by about 10°C. So I had high hopes when I bought the Scythe Ninja II Rev. B for a new build. The CPU in question is an AMD Athlon II X4 620 (Propus) on an Asus M4A785TD-V motherboard. I have an Enermax MOD82+ 525W PSU (fan speed: 505RPM), which I bought on the strength of the SPCR review.

It was a real job to get it on. The pressure on the CPU must be quite considerable. I could only just about get the clips on. It's now currently running with the 12cm Scythe fan @ 1073RPM with SpeedFan temperatures of Temp1 49°C and Core 43°C under Prime95 load.

My main problem, though, is that I really want to run it fanless. That was the whole idea of buying it, after all! However, without the fan, it fairly quickly goes up to the 70°C mark, at which point I hurriedly stop the test.

Now I will undervolt at some stage, but I expected that I would see better performance before fiddling around with undervolting.

I read that "Tom." in this thread (SPCR 453259) was able to run a Phenom II 720 X3 passively with a Ninja and a similar PSU...so I'd have thought it would be able to manage the lower-clocked Athlon II X4 620.

Has anyone any experience of cooling the Athlon II X4 620? Could I be missing something obvious?

Thanks,

Leggie

[KadazanPL]

Hi Leggie,

I would check the possibility of ducting the Ninja to the exhaust case fan (you do have one, don't you?). A shroud of some kind should tunnel air throught the Ninja's fins and cool it down considerably.

Another thing to consider is the thickness of the thermal paste. Maybe it's too thin or too thick? Have you allowed it to burn in for some time?

Good luck.

[Leggie]

Hi Leggie,

I would check the possibility of ducting the Ninja to the exhaust case fan (you do have one, don't you?). A shroud of some kind should tunnel air throught the Ninja's fins and cool it down considerably.

Another thing to consider is the thickness of the thermal paste. Maybe it's too thin or too thick? Have you allowed it to burn in for some time?

Good luck.

I confess that I don't have an exhaust case fan. I thought it tended to be a secondary concern to having a CPU fan. How essential is it to have one?

What do you mean by burning in the thermal paste? How long for? What sort of thickness of paste would you recommend?

These pictures should show what sort of thickness of paste I use:

i189. photobucket .com/albums/z267/AndrewSpinner/DSC00312.jpg
i189. photobucket .com/albums/z267/AndrewSpinner/DSC00313.jpg

Thanks,

Leggie

[Vicotnik]

If your PSU fan is the only exhaust fan then all the heat from the system will pass through the PSU, causing it's temperature controlled fan to spin faster.
Usually much better to have an exhaust fan close to the CPU. That's usually all the fans you need unless you have some really power hungry components.

Regarding the thermal paste, on a chip with a large heatspreader I would go for a rice size little glob in the middle.

[Leggie]

If your PSU fan is the only exhaust fan then all the heat from the system will pass through the PSU, causing it's temperature controlled fan to spin faster.

Right, I see. So far, however, even under load the PSU fan is only spinning at about 500RPM.

Regarding the thermal paste, on a chip with a large heatspreader I would go for a rice size little glob in the middle.

OK, so is my method of spreading it out over the chip not such a good idea?

[Vicotnik]

Right, I see. So far, however, even under load the PSU fan is only spinning at about 500RPM.

You have a great PSU. But 500RPM is not enough to cool both the PSU and the rest of the system. A case fan close to the Ninja would fix that.

OK, so is my method of spreading it out over the chip not such a good idea?

Whatever works. I prefer the "Arctic Silver method" - thin even layer on naked chips, rice size glob in the middle on chips with a heatspreader.

[Leggie]

OK, so is my method of spreading it out over the chip not such a good idea?

Whatever works. I prefer the "Arctic Silver method" - thin even layer on naked chips, rice size glob in the middle on chips with a heatspreader.

OK, I've removed the heatsink (again!) and tried that method of applying the paste. Fanless temperatures still aren't pretty. Hmmm...will think about it more tomorrow.

Thanks,

Leggie

[dhanson865]

Depending on your thermal paste it can take a day or even weeks to fully cure. Until then you should keep extra fans going and lower the airflow over time until you hit a level you are comfortable with.

Every time you wipe the CPU clean and reapply paste you are starting the clock over again and won't know how it truly works out unless you wait.

They make paste that cures quicker so reviewers can do their thing in a reasonable amount of time but you have to be sure what the cure time is of whatever you are using before you measure temps.

http://benchmarkreviews.com/index.php?o ... mitstart=1 shows this set of footnotes

* (0) No Curing Time or Special Application Suggested
* (1) Antec Formula 5 Application Instructions (no curing time recommended)
* (2) Arctic Silver II Application Instructions(48-hours minimum curing time recommended)
* (3) Arctic Silver 3 Application Instructions (up to 200-hours recommended curing time)
* (4) Arctic Silver 5 Application Instructions (up to 200-hours recommended curing time)
* (5) Arctic Silver Ceramique Application Instructions (25-hours minimum recommended curing time)
* (6) IC Seven Carat Diamond Application Instructions (10-minute evaporation time, 2-hour curing recommended)
* (7) CooLaboratory Liquid Pro Application Instructions (no curing time recommended)

Unfortunately Arctic Silver II has a very poor rating in the tests and Arctic Silver 5 tests well meaning you want the slower curing paste in the long run (at least in this one brand specific comparison).

As to applying the compound, You might want to try the two lines method shown here

[Esben]

I'd recommend you first try undervolting the CPU before trying to run it passive. The CPU has an 95W TDP. I'm using the little brother of your heatsink (a Ninja Mini) to passively cool a 3.2 GHz 45nm Core 2 Duo, which I estimate has a power consumption around 35-40W. The PSU fan in my case sits about 2-3 cm from the heatsink, is 140mm and runs at 900 rpm all the time. When I run small fft calculations in Prime95, the temperature stabilizes around 66C.

From this article at Tomshardware, it looks like your CPU has a very big undervolting headroom. If you shave off 40-50W of the power consumption it will be much easier to cool. A rough guess is that your heatsink is 25 % more effective because of the larger size, but on the other hand your PSU fan is less effective because it only runs 500 rpm. I'm very interested in hearing what you end up with. Apart from undervolting the CPU, you might also has to reduce clock speed a bit to get the TDP down enough for passive cooling.