Intel P4 Socket 775 Heatsink Roundup

Cooling/CPU Cooling by JNav89GT @ 2005-01-05

Looking to cool that small nuclear reactor Intel has sold you in the Prescott P4. Well we have for you today 6 samples of aftermarket LGA775 coolers that might just let you squeak a little more MHz out of your CPU, or at the very least allow the CPU to run just a bit cooler.

  • prev
  • Go to mainpage

Conclusion

TTIC Revisited

During testing, the TTIC cooler seemed to be performing below my expectations. While still a bit better than stock, the noise of the supplied fan was quite loud in comparison to a stock Intel cooler and I believe a 1°C gain at load in exchange for the added noise is not a good trade. I had been running the TTIC unit in another computer for a couple of weeks, and as you can see a bit of dust had been "blown" onto the top of the unit from the fan. By placing my hand near the unit while running, it was apparant that quite a bit of air was blowing around the heatsink and not through it. I did exchange the supplied stock fan for an 80mm 38cfm unit I had around, as an 80mm fan would fit up against the fan bracket without gaps. This did improve performance a bit, but I suspect a stronger fan would be needed for improved performance. I also question whether the addition of a fan shroud to "duct" air through the heatsink design would improve performance, but that I will leave to TTIC to discover. I did remount the unit several times, in an effort to improve performance, but results were similar.

Madshrimps (c)Madshrimps (c)



Conclusion

Most of the heatsinks tested today did fairly well with appropriate fan speeds. The Swiftech MCX775-V (high fan speed), Evercool HPC925 (high fan speed) and the CoolJag JAF8025H2 units were among the best during testing. I would recommend either of these heatsinks should you wish to air cool your overclocked or even non overclocked Prescott P4 LGA775.

My assumption about tower coolers was off base, as the Evercool HPC925 basically proved me wrong and tied for best temperature scores.

The Swiftech MCX775-V continues Swiftech's commitment to excel in heatsink design, construction, and implementation.

The CoolJag JAF8025H2 unit surprised me as well, furthermore, the differences in temps between basically two identical units is confusing.

What's grossly apparent was the ability of the overclocked P4 to overwhelm the coolers tested that offered a low fan setting. The Prescott just pumps out too much heat to dissipate with such low speed fan flows. With higher flows things improved, and since we are dealing with larger fans of 80/92mm diameter, the noise levels were in most cases tolerable.

The other units did well and should warrant consideration depending on your needs and price points you are considering.

I would like to thank Swiftech , Evercool, TTIC and Cooljag for the samples provided.

Question/Comments: Forum thread



UPDATE: We had a reader request data on the Evercool HPC925 tower cooler oriented as it might be in a "normal" users case. I had to respond, "we have no normal readers, they are all maniacs just like us", but I digress. Anyway, the point was valid, so I put system into a spare case, and this time oriented the fan on the side of the heatsink so that the hot air would be blown back towards the exhaust fan of the case(see pic below). I then re-tested the temps with case lying on it's side, as it had been during bench testing. The case side panel was on, for those who care. I then re-tesetd data to see if there would be any changes from prior bench test data. After I collected side lying data, I stood case up to simulate a normal orientation the HSF would be in most users cases. The concern voiced in the comment of the reader was if the heatpipes would be more effective in our original testing. This could be logical as a heatpipe could be more efficient where there was a verticle orientation of the heatsink versus a normal type horizontal orientation in relation to the ground as in most cases.


Madshrimps (c)

Madshrimps (c)


Well, after revisiting this heatsink in different orientations, I can say there was basically no difference for me whether the heatsink was vertical or horizontal. Yes idle temps went up a bit, but this could be due to higher case temps as well, as the horizontal/upright tests were done after the vertical/side lying. This is actually good news, as I was afraid the data would go to hell in a handbasket and I'd be eating my shoe or something. We'll hope this helps! Any other questions please visit our Forum thread!
  • prev
  • Go to mainpage