Asetek 570 LX - 240 mm Radiator (Push-Pull Configuration) Review

MingtheMerciless

Distinguished
Oct 11, 2010
24
0
18,520
Taming the Core i7 w/ Asetek 570LX 240 mm Radiator
To break the 4 GHz threshold, a slight modification was performed on the Asetek 570LX 240 mm radiator to help bring down temperatures on my particular CyberPowerPC Build. The system previous configuration consisted of dual 120 mm fans placed on the top of the radiator, in pull-exhaust configuration, which relies on a pressure differential to pull air through the heat exchanger – grill. However, the dual stock fans did not provide sufficient CFMs for overclocking beyond 4GHz. Therefore to further induce maximum air flow, the stock fans were replaced, and a push-pull configuration was adopted, by placing additional set of dual 120 mm fans on the front of radiator along with the ones on the rear. The temperature effect was most pronounced at 100% load, temperatures dropping by nearly 10 C, while idle temperatures seem largely unaffected. For those interested, in a similar push pull fan configuration w/ the Asetek570LX 240mm radiator, you’ll require 6-32 screws. For typical 120mm (35mm ~1” – thick fans) you’ll need 1 3/16” length screws, which isn’t a standard length. Get a 1 ¼” and dremel it down to your length. For those wanting to use the Scythe Ultra Kaze 3000 (38mm ~1.5”) you’ll need 1 11/16” length screws, but again find what you can get and Dremel them down.

Stock Dual 120mm Fans Replaced By Scythe Slipstream 110 CFM Fans
IMG_2584.jpg

Scythe Ultra Kaze 3000 Fans Removed From Custom Northbridge and SLI Fan Fixture and Placed On Bottom Grill of 240mm Radiator
IMG_2574.jpg

Push-Pull Configuration
IMG_2582.jpg

Barely Enough Clearance
IMG_2591.jpg

IMG_2595.jpg

IMG_2603.jpg

Before and After Temperatures (Drop by 10 C)
Asetek570LXPushPull_Chart.png



Having dropped maximum operating temperatures by roughly 10 C, I wanted to see how far I could take it down by reseating and replacing the TIM with Arctic Silver 5. The Asetek 570 LX comes with some type of thermal pad, what looks like some circular thermally conductive adhesive that cures with use, but I opted during the CP ordering process to use CoolMaster ThermalFusion 400 TIM. Below are some pictures that show what the interface looks like, and the CoolMaster viscosity is comparably lower compared to the Artic Silver 5. But the CP technicians did a proper job prepping the Asetek copper plate by applying a light coat first, but put just a tad bit too much on the processor.

IMG_2635.jpg

IMG_2636.jpg


Results show negligible difference, if at all any drop was present, or rise, it would be masked by the inherent error of +/- 1 to 2 C from either incorrect calibration of the sensors or by thermal cycling. Then again, Artic Silver 5, possesses a curing time of roughly 200 hours, so I’ll report back in a week if the temperatures do indeed drop significantly.

IntelCorei7950TIMComparison.png
 

MingtheMerciless

Distinguished
Oct 11, 2010
24
0
18,520
Those load temps are horrendous...

Agreed, some contend go-full water cooling or not at all, however I never liked the concept of cooling on air with a large heat exchanger hanging cantilevered from the processor/motherboard - concerned about bearing stress, shear, bending moment and torsion. Considering it's performance w/ others coolers in it's class, I say the temperatures are its spot on, if not slightly better - especially running under the maximum stable over clock for this CPU, Core i7 950 4.0GHz @ 1.312 V (167X23/24-TurboBoost Enabled, HT Enabled).

CPUCoolerComparison.png

*Data represented from H50 were taken from Intel Core i7-920 3.8 GHz @ 1.366 V (200 X 19) *Assumes 23 C Ambient
http://www.tomshardware.com/reviews/h50-fort120-cogage,2401-6.html

*All other coolers were taken from Intel Core i7-870 4.0 GHz @ 1.388 V *Assumes 23 C Ambient
http://www.tomshardware.com/reviews/lga-1156-heatsink,2535-14.html


 

MingtheMerciless

Distinguished
Oct 11, 2010
24
0
18,520
Almost all of those large air coolers come with some kind of backplate or other mounting mechanism to reduce strain on your motherboard. 1000's of people wouldn't be using them if they didn't work and they broke PCB.

That's a good point, regarding my reservations structurally for these large air coolers. The back plates on many designs do distribute the load, of these large air coolers, across a larger surface area of the motherboard, therefore reducing the corresponding stresses, so perhaps my concern regarding the shear stress is a little unfounded, especially considering the number of people using such coolers as you mentioned. Moreover, it's not like the cooler is being mechanically fatigued, though thermally of course, but that shouldn't effect creep rates as long as the back plates and fasteners are structurally sound.

1. How much did the 570LX cost? I could not find the unit cost. Seems like this is only specific to CyberPower. If so, how much extra did it cost to get it(unless it already came as part of the build)?

During checkout the cost was $43 extra for the Asetek 570LX, from a NZXT 120mm liquid cooler. It seems it's only offered through CyberPower, Inc, as searching the Asetek website pull any specifics, though there is an Asetek 570 LC (Corsair H70).

http://www.asetek.com/products/oem-standard-products/570lc.html

2. Those temps are way too high. What CPU and what vCore?

Intel Core i7-950 4.0 GHz (167X23/24-TurboBoost & HyperThreading Enabled) @ 1.312 V

The processor performance varies from lot to lot, and for this particular one, I could only obtain a stable OC at ~3.5 GHz without any voltage increases, however had to boost them up considerably to push past 4 GHz. From what I have read though, 1.375 V is an upper limit, and definitely requires a rather effective cooling solution whether that be on air or water. If I have some time, I'll post a chart with load temps at different frequencies and voltages, to better understand the dynamics involved with the over clock settings. I also plan on letting the Arctic Silver 5 cure, to what the manufacturer recommends as 200 hours, before, removing it, and reevaluating the CoolMaster Thermal Fusion 400.
 

MingtheMerciless

Distinguished
Oct 11, 2010
24
0
18,520
If you can get those voltages down, you should see a reduction in temps.

Unfortunately, to run stable, I had to maintain the current voltage at 1.312 V.

Interestingly you mentioned this, as running at 1.298 V I found the system was stable with 4 hour run of Prime95, passes 5 Intel Burn Test Runs at Max, but leaving Prime95 run overnight for failed around the 8 hour mark, so upped the voltages to 1.312 V and will pass 12 hours. After a number of stress tests, the longest stress test ran was roughly 15 hours, I believe the system is stable. Then again I am not entirely sure when one can say with great confidence their system is stable, it seems arbitrary sometimes what people argue as to the length of time, why 12 hrs, 24 hrs, or 48 hrs. Is there some general consensus here as to declaring your system stable?

As for reducing temperatures. From what I have read, base clock changes will effect temperatures linearly, whereas changes in voltages effect temperatures quadratically. I also understand disabling Hyper-Threading could also reduce temperatures. So I'll add that to the list of things to examine when I compile a final chart showing temperatures from stock to over clock with Hyper-Threading enabled or disabled.
 

rubix_1011

Contributing Writer
Moderator
I rarely run those tests longer than an hour or 2...4 would be tops. This would be beneficial if you intended to run the system at full bore for days on end, but how often do people do this when you intend to overclock? If someone is folding or performing some other long-term task, you'd be better off to go with a much milder overclock in the interest of stability and duration of uptime. This is the reason servers aren't overclocked.

To each his own, though. I see the benefits of running benchmarks for many hours on end, but ultimately it rarely impacts end-user applications such as gaming, video editing/ripping/encoding, photo editing, folding or other 'cloud-based' computing. I'm not knocking it, I just don't make part of my routine.