IC Perihelion, Noctua NT-H1, and OCZ Freeze Review


Thermal Testing Methodology/Specification

Methodology

My TIM tests are a derivative of my waterblock tests. I use Dallas One Wire DS18B20 temperature probes at the block inlet and I use the same pump and block on every test, and use good testing practice by performing 5 mounts with all TIMs. Where applicable, I follow manufacturer’s installation procedures to the letter. For my TIM tests, I’ll be plotting temperature vs. time, in the form of a 120 minute moving average (or less for the first two hours of data). A moving average is used to smooth out the noise associated with this kind of measurement and to maintain a very high precision of information. I will be examining two specific components of TIM performance: how long it takes to cure (if within the 10 hour testing time) and, more importantly, what kind of temperatures an end-user can expect.

Specification

A single 10hr test per mount with 5 mounts was completed for each TIM at each contact ‘setting’ then repeated for the other two contact settings. Everything was held consistent between tests and everything was logged.

  • The processor I’m using for this test is an Intel Core i7 920 D0. I’m running it at 3800MHz at 1.40V loaded on an MSI X58 Big Bang. It is unlapped. I’m running 2GB of G.Skill DDR3 1333MHz. The VRMs, northbridge, and southbridge cooling are all stock. The video card is a passively cooled ATi 4350 running in the lowest PCIe 16x slot. The board is sitting on an HSPC Top Deck Tech Station and airflow is provided over the VRMs and RAM.
  • The watercooling loop I’m using is very basic but is enough to allow me to test the way I want to test.
    • I use a single MCR-220Res for my radiator with Gentle Typhoon AP15 fans.
    • I use an MCP350 with an XSPC ResTop with a Swiftech Micro-Res on top of the ResTop.  This allows me to get full bleeding speed without having to refill the loop every block switch.
    • I use a trio of Koolance CPU-360s with VL3N QDCs and a custom mounting system based on the stock mounting system.
  • I do 3 five-mount tests, each with their own TIM application and full cleaning between. I discard the worst mount and average the remaining four. I also present the best mount results as their own comparison as I feel it is indicative of what you can expect on a near-perfect mount. As a reviewer, I feel it is my duty to present the reader with performance numbers of a product that represent what its typical performance is and I also present the average performance data.
  • I have seven temperature probes in use: three Dallas DS18B20 Digital one-wire sensors and four Intel thermal sensors in the processor.
  • For temperature logging, I use OCCT v3.1.0’s internal CPU polling that is performed every second and is automatically output to .CSV files. I also use OCCT for loading the CPU. For water temperature, I use Crystalfontz 633 WinTest b1.9 to log the Dallas temp probe data on my Crystalfontz 633. I have found, much to my chagrin, that programs like RealTemp, CoreTemp, Everest, etc., all have their own massive flaw in temperature logging that prevent them from being used for such a test. These flaws range from data formatting issues, to sensor polling issues, to random, yet common, stalls in the software (especially when logging).
  • For processor loading, I find OCCT v3.1.0 to be extremely competent. With the Small Data Set setting, it provides a constant 100% load (so long as WinTest b1.9’s packet debugger is fully disabled) and is extraordinarily consistent. It allows me to, in one button push, start both the loading and the logging simultaneously, which helps. I immediately also start to log the Crystalfontz data via WinTest b1.9. I run a 10 hour and 5 minute program, the first minute is idle, then I have 10 hours of load, and then 4 minutes of idle.
  • Intel DTS thermal sensors are not perfect.  In addition to the mediocre 1C resolution, their response curve (CPU vs. water temperature with cooling and load held constant) is not 1:1. I mapped the entire range of applicable temperature readouts and have remapped the outputs in my Excel spreadsheets so that every one degree increase in water temperature results in a one degree increase in CPU temperature readouts. This must be done to ensure that any variation of ambient and water temperature do not have any effect on the outcome of the data from testing.

Installation Notes and Procedures

IC Perihelion, Noctua NT-H1, and OCZ Freeze

Considering these are classic pastes, installation of this should be familiar and therefore a little easier. I’m a fan of a bead in the center, varying its size depending on the consistency of the paste.  NT-H1 recommends twisting/wiggling the block or heatsink before tightening down, but this is not possible on our testbed.

Pages: 1 2 3 4 5

Comments

Posted On
Aug 11, 2011
Posted By
il0
Posted On
Aug 11, 2011
Posted By
Eric (Vapor) Hassett

The Coollaboratory liquid metal TIMs and IC Diamond will be tested at the very, very end of this testbed’s life. They’re too likely to permanently alter the surface to be done mid-stream (if they do alter the block/IHS surfaces, they skew performance and it would be best if the number of TIMs affected were minimized).

EVGA Frostbite is upcoming after this batch of Shin-Etsu G751, X23-7762, and X23-7783D. I’ve already done some preliminary tests with it on another testbed and it seems very similar to Nexus TMP-1000, which is not a positive sign.

I think it’s unlikely that thermal pads will be fully explored, at least with this testbed and procedure.

Posted On
Sep 11, 2011
Posted By
il0

i know thermal pads generally suck but i was curious how/if there have been any improvements over the years 😛

Posted On
Jul 28, 2011
Posted By
Church

Hmm, interesting. New shin-etsu versions? Somehow this is first place i noticed X23-7921-5 and X23-7868-2D mentioned. After quick googling X23-7868-2D seems what’s coming OEM with antec integrated LC sys (and as OEM TIM for few other coolers), with a bit improved heat transfer of 6.2 W/mK vs 6.0 for X23-7783D, and X23-7921-5 maybe even better then that. Pity these two new ones seem selling retail only in Japan.

Posted On
Jul 28, 2011
Posted By
Eric (Vapor) Hassett

Yeah, hopefully 7921-5 and 7868-2D get to retail before this is all finished…still 4-7 months away from closing down the testbed and testing ICD and CLP/CLU, so there’s time.

Leave a Reply

You must be logged in to post a comment.

Login