Introduction

My neighbor had an easy bake oven when we grew up; I remember thinking how silly it was to cook brownies under what amounted to the heat produced by a light bulb. Well she would probably have a good laugh today thinking of all the dishes we might cook, powered by the high heat output of our computers processor. The most infamous example of high heat output used to be the Athlon Thunderbird 1.4GHz CPU, well known to roast the insides of people's cases. The torch has passed though, yes pun IS intended here, to the Intel Prescott P4. Even the most ardent Intel supporters can't look the other way when it comes to the heat signature of the Prescott CPU. So while we can debate this all day, in the end we as consumers must learn to deal with the heat in the best way possible, with improved cooling solutions.

Today we are looking at six aftermarket air cooled heatsinks which hope to improve the end users cooling situation. Whether the goal is higher overclocking, reduced temperatures, or both; these heatsinks hope to have what you're looking for. Samples were sent from well known manufacturers such as Swiftech , Evercool, TTIC and Cooljag, all compared against the boxed supplied Intel retail cooling.

Swiftech MCX775-V

Evercool PT01A

Evercool HPC 925

TTIC / Thermal Transtech NPH-775-1

CoolJag JAF8025H1 & JAF8025H2

Test Setup & Methodology

JNav89GT's Intel S775 Test Setup
CPU	Intel P4 530 Retail-3.0GHz LGA 775
Mainboard	Albatron PX865PE7 Pro
Memory	Mushkin PC3200 Level II V2 1024/2x512mb
Video	ATI Radeon X700Pro
Power Supply	Fortron 520W ATX
Operating System	Windows XP SP2

For testing purposes, the motherboard was placed outside of a case on a small pedestal, and all testing was done on bench top. Ambient temps were recorded with a Radio Shack digital thermometer and scores later standardized to accommodate variations in ambient temps.

Each Heatsink was tested with the CPU running at full load by running two instances of CPU burn with the LGA 3.0GHz CPU at 3.6GHz and 1.3825v. This was the best possible way for me to stress each of these heatsinks independently.

Smart fan was disabled in bios to allow fans on heatsinks to run at full speed so as not to confound readings. In the case of the Swiftech unit where a fan is not supplied, I choose a variable speed fan and included results for low and high speeds. If a supplied heatsink offered variable control of fan, I did high and low speed testing, otherwise fans were run at full speed. There was one unit I chose to replace fan on and retest; I will describe the rationale and results later in article.

So now onto the contestants ->