"Heatsink Roundup"
Joe Citarella
The following table ranks CPU cooling performance based on CPU Die Simulator Test Results¹ for various CPU cooling solutions. The Die Simulator is a small die (130 mm surface area) pure performance test which excludes secondary heatpath effects found on motherboards.
Also note that these rankings use air cooling heatsink fans as they are sold unless a heatsink is sold without a fan; some stock fans are underpowered and replacing them with a Delta 38 or more powerful fan will improve results. However, noise levels will noticeably increase, as will cost, if another fan must be purchased. Factors such as noise, motherboard performance and the impact of higher speed fans may be found in the reviews.
Sidewinder Computers has sound files for most of the fans you'll encounter HERE. "All fan sounds were recorded in MP3 audio format (128kbps bit rate; 44.1kHz sample rate) and are ranked by apparent loudness. To hear them, you must have an MP3 player installed, such as Winamp, Sonique, Musicmatch and others." A very nice resource!
Interpreting Results
The relative rankings in this table indicate the difference, in degrees Centigrade, among the CPU cooling options tested. For example, at 100 watts, the difference between Swiftech's MCX462 and the Millennium Glaciator is three degrees C; at 50 watts, 1 ½ C, and so on. Closely ranked heatsinks may, in fact, show little or no difference, or even reverse positions, when compared in a specific system and especially at heat loadings of 50 watts or less.
Viewing results as +/- 5% is more indicative of what a user may find than relying on the absolute number.
Air Cooling DIE Simulator C/W TEST RESULTS
(click on name for full review²)
C/W 0.20 |
0.30 |
0.40 |
0.50 |
0.60 |
| Swiftech MCX462: 0.31 |
PIV, Socket A/370, Delta 80mm |
| Antec Jet Cool: 0.62 - Intel PIV |
|
| Dynatron DC1U-B01: 0.64 |
|
What Can I expect In My System?
First, understand what C/Ws are telling you: The difference between a heatsink with a C/W of 0.30 and 0.35 is 5 C at 100 watts under stress. At 50 watts, it's half that. Closely ranked heatsinks may show little or no difference on a motherboard, due to secondary heatpath effects which are not captured in these rankings.
Second, I get emails from some readers who buy heatsinks and find performance does not match what they read in our and other reviews. The first question I ask is "What is the temp inside your case??" Invariably the answer is something like "30/35/40 C" - sometimes higher!
The air temp going INTO the heatsink or radiator fan is the floor - all other temps you see are added onto this temp. If the air temp is 35 or 40C going into the fan, then there is NO WAY this heatsink will give you 40C at the CPU. The following table lays out what to expect for a 100 watt CPU:
CPU Temps @ 100 Watts
| Case Temp / C/W |
0.15 |
0.20 |
0.25 |
0.30 |
40 C |
55 C |
60 C |
65 C |
70 C |
35 C |
50 C |
55 C |
60 C |
65 C |
30 C |
45 C |
50 C |
55 C |
60 C |
25 C |
40 C |
45 C |
50 C |
55 C |
20 C |
35 C |
40 C |
45 C |
50 C |
NOTE: Assumes CASE TEMP = FAN INLET TEMP!
As the table shows, the higher your case temp (and fan inlet temp), the hotter the CPU temp. It's no wonder that many are finding fresh air ducts to be a very effective solution at lowering CPU temps. Effective CPU cooling requires BOTH a good heatsink and airflow through the case.
Note that AMD's guideline max temp for CPUs is about 70 C! (Go HERE for details)
To test out how much case temp impacts your system, open the side of the case, aim a house fan at the motherboard and record temps under stress. This is "best case" for airflow through the case.
To calculate what to expect for other CPUs, for every watt the CPU radiates, the heatsink will cool the core by the (C/W x watts) plus ambient temp. For example, at a fan inlet temp of 25 C, a C/W of 0.25 with a CPU radiating 50 watts means that the CPU temp will be 50 x 0.25 = 12.5 C over ambient temp, or 37.5 C.
Last, remember that in-socket thermistors are not accurate measures of CPU temps. Every heatsink affects an in-socket thermistor differently, as does airflow over the motherboard. There is no consistency between motherboards as well. At a minimum, you can factor +/- 5 C to these temps.
For an overview of what AMD recommends, see
AMD CPUs: How Hot Is Too Hot?
¹Heatsinks tested prior to September 2001 were newly tested during December 2001 on the Die Simulator. Heatsinks were tested at about 75 watts.
²In some instances, Die Simulator test results may be reported before a full test report is posted.
Email Joe
<< back to tips
