Testing methodology recommendations?

[davidthemaster3]

Hi guys,

I recently bought a Noctua NH-D14 as an upgrade to my Coolermaster 212+.

So far I thought of testing with the 212+ overclocked/not overclocked

I will also be changing my fan configuration because I bought new fans (2 x Cougar Cooling Turbine 140mm Hyper Spin Bearing Fan 1000RPM)

Here is a list of my specs:
Intel 3570K @ stock
Gigabyte 670 OC @ stock
4 x 4GB Samsung green RAM @ stock 1600 Mhz 11-11-11-28 2N @ 1.35V
Asrock Extreme4 z77 Bios version 2.00
Crucial m4 256 GB firmware 000F
Western Digital Caviar Black 2TB
Creative X-Fi Titantium Sound Card
Fractal Design R3 case with 2 x Gentle Typhoon Front Intake, 1 x Silverstone AP121 bottom intake, 1x rear exhaust Fractal case fan, 1x middle top exhaust Fractal case fan

Here is what I'm aiming for in the end:
Intel 3570K @ 4.5 GHz
RAM @ 2000 Mhz 9-10-10-28-1N @1.45V (I have gotten this tested and stable already)
GPU OC maybe don't know how much yet
Fractal Design R3 case with 1 x Gentle Typhoon Front (lower) Intake, 1 x Silverstone AP121 Front (higher) intake, 1x Cougar Cooling Bottom intake, 1x rear exhaust Fractal case fan, 1x middle top exhaust Fractal case fan, 1x Cougar Cooling top exhaust (the other space).

What I want are recommendations on a testing methodology to see how my temps have changed and recommendations on my fan placement

I will be able to post pics in needed

 

[T_Yamamoto]

Run prime95 and check temps

 

[davidthemaster3]

Run prime95 and check temps

That and LinX are tests I'm gonna do at least

 

[T_Yamamoto]

That and LinX are tests I'm gonna do at least

Then that's all you need. Make sure to run furmark or something GPU intensive. Fan placement should be so air flows from bottom front to top back

 

[dkm777]

I'd say for a total peace of mind a furmark+Linpack run should be done. In the past I made my PC overheat and BSOD when I forgot to close my Folding client and started gaming . Replacing/rearranging the fans fixed that and now it's all gravy.

 

[Nickel020]

Set a fixed FFT size in Prime, e.g. 1024. When you choose one of the normal options Prime will cycle through different sizes and different sizes cause slightly different heatloads. Also use the logging feature in CoreTemp/Realtemp and calculate an average over a few minutes in Excel, that's much more accurate than reading a set of values at one given moment.