Air or Water?

Add Your Comments

A perspective on choosing between the two – Joe

SUMMARY: Air cooled heatsink are competitive with water cooling kits, with some caveats.

I routinely get emails remarking on the similarity of C/Ws between air cooled heatsinks and water cooled kits, so I thought a more detailed look was in order.

First, direct C/W comparisons between water cooling kits and P4 heatsinks are not valid – the air cooled heatsinks are tested using a P4 test bed and the water cooling kits use the die simulator. The die measures 10 x 13 mm and as such is not comparable to the P4’s IHS.

However, there have been times when I tested P4 heatsinks using both methods. I looked back to these tests and on average, the difference between the two is about 0.10 +/- 0.02, with the P4 results being lower than the die simulator. I want to caution that this is an average difference – please view what follows with that caveat.

The table below ranks P4 heatsinks and water cooling kits using the conversion factor above – if, for example, a water cooling kit C/W was 0.20, it adjusts to 0.10 to compare against P4 air cooled heatsinks. I have indicated the adjusted C/W in parenthesis in the water cooling kit results (water kits in BLUE); I have limited the selections to P4 heatsinks with C/Ws of 0.20 or better and water cooling kits with adjusted C/Ws of 0.20 or better.

Air Cooled P4 Heatsinks and Water Cooling Kits C/W Ranking

C/W     0.10

0.20

0.30

0.40

0.50

0.15 (0.05 est)

Swiftech H2O-Apex “Extreme Duty” Watercooling Kit, 2 120 mm fans

0.16 (0.06 est)

Asetek WaterChill – Radiator with 3 120 mm fans, 12 volts

0.16 (0.06 est)

1A-Cooling Blacklord 240 – External radiator with 2 120 mm fans

0.17 (0.07 est)

Asetek WaterChill – Radiator with 3 120 mm fans, 7 volts

0.18 (0.08 est)

1A-Cooling SET-5Z280V1 – Fans high; 0.26 fans low

Thermalright XP-90C Heatpipe: 0.11

90 mm fan, 4870 rpm, very loud; 1960 rpm quiet, C/W 0.13

Corsair HydroCool 200: 0.22 (0.12 est)

External Kit – Turbo fan speed

Thermalright XP-90 Heatpipe: 0.13

90 mm fan, 4821 rpm, very loud; 1500 rpm very quiet, C/W 0.18

CoolingKing: 0.23 (0.13 est)

Kit with EHEIM 1048

Innovatek: 0.23 (0.13 est)

Kit – EHEIM 1046, maxXpert radiator

Koolance EXOS: 0.23 (0.13 est)

Fans at 100% power

Thermalright SI-120 Heatpipe: 0.14

120 mm fan @ 2994 rpm; 984 rpm = 0.18 C/W, virtually soundless

Thermalright SP-94 Heatpipe: 0.14

80 mm fan, 4838 rpm

Corsair HydroCool 200: 0.24 (0.14 est)

External Kit – Low fan speed

Thermalright SLK-800U: 0.15

80 mm fan, 55 dBA

Q-Power: 0.25 (0.15 est)

Case & Kit – EHEIM 1048

Asetek Micro ™ Extreme: 0.16

92 mm fan, moderate noise

TTIC-NPH-101 P4 Heatsink: 0.16

2 70 mm fans, low noise

Cooler Master IHC-H71: 0.16

70 mm fan, 51 dBA

Thermalright-947U: 0.16

Delta 80mm 4800 rpm; loud

Thermal Transtech TTIC-NPH-1: 0.16

Two 70 mm fans

Alpha PAL 8942: 0.17

Delta 4800rpm

Thermalright 120mm P-4 Heatpipe: 0.17

Delta 120mm fan @ 2500rpm, 54 dBA (low noise)

CoolWave Stormcool III: 0.27 (0.17 est)

Best case results.

Koolance EXOS: 0.27 (0.17 est)

Fans at 45% power

maxXxpert: 0.27 (0.17 est)

Kit – EHEIM 1048, spiral waterblock

CpuMate Heatpipe: 0.18

moderate noise @ 5000 rpm

Scythe Kamaboko : 0.18

Rheostat controlled fan

Vantec Aeroflow P4: 0.18

TMD fan – low noise

Zalman CNPS7000-Cu: 0.18

Very low noise

CoolerMaster Aero4: 0.19

Adjustable fan

Evercool WC-201: 0.29 (0.19 est)

Fan on High

Koolance: 0.29 (0.19 est)

Fans on HIGH; Proprietary Case

Dynatron DC1207BM-X: 0.20

70mm fan, relatively quiet

Swiftech MCX4000: 0.20

TMD fan, 39 dBA

Thermalright AX478: 0.20

Sanyo 4420 rpm

Asetek WaterChill: 0.30 (0.20 est)

Watercooling Kit

Evercool WC-201: 0.30 (0.20 est)

Fan on Low

Space2000: 0.30 (0.20 est)

Watercooling Kit

While not a perfect comparison, results indicate that there are indeed air cooled heatsinks that are competitive with water cooling kits. However, note that for extreme performance, high-end water cooling kits dominate – not a surprise by any means. While some air cooled heatsinks attain top results, many times it’s with very aggressive (noisy) fans – not necessarily the case for water.

Air cooled results must be tempered by taking into account case temps – an excellent air cooled heatsink in a poorly ventilated case will NOT deliver the C/Ws shown in this table (for more on this, go HERE.) The higher the case temp, the larger the discrepancy between test results and performance on screen.

Water cooling kits, in comparison, tend to pull cooling air from outside the case, so that case temps have much less impact on performance.

One major difference between air and water is the amount of heat which can be handled; the “overhead” with water is much higher compared to air cooled solutions. It is not uncommon to see water cooling systems with TECs used to cool CPUs to below-ambient temps, a situation not feasible for air cooling.

CONCLUSIONS

All my desktops are water cooled, first and foremost for low noise. Newer heatpipes can effectively compete on this basis and deliver extremely good cooling performance, assuming excellent case airflow. Water will always trump air for the load it can handle.

Users pondering air vs water must first consider their objectives in CPU cooling – low noise with high heat loads tips the balance to water, while low noise and “normal” heat loads may tip the balance to air, although other factors, such as case airflow, must also be taken into account.

Finally, the cost difference between the two solutions is significant – expect water to cost about five times more than air for high-end kits.

Email Joe

Leave a Reply

Your email address will not be published. Required fields are marked *