Heatsink testing overview - bit-tech.net Forums

Tim S · 16th Jun 2008, 08:56

http://www.bit-tech.net/hardware/200...ing-overview/1

Reliably testing CPU heatsinks can be a tricky business, with so many variables capable of jeopardising accurate results. For this reason, we've re-assessed our testing methodology, and put some of the most popular and high-performing heat sinks through their paces in our new standard test setup.

The boy 4rm oz · 16th Jun 2008, 09:14

The vendetta did quite well under load. I thought it would perform worse.

wuyanxu · 16th Jun 2008, 09:20

oh yeah! TRUE is still the top one.

i bet it can drop a lot more degree if you use one or two high CFM fan.

so what heatsink are you planning to test next? one of those newer direct contact heatpipe ones? many test sites say they are better than TRUE....... me wants a reliable source.

ViPPeR_666 · 16th Jun 2008, 11:22

Nice review, as always.

Btw there's a Noctua instead of a Thermalright 120 on last page.

Tim S · 16th Jun 2008, 11:23

Quote:

Originally Posted by ViPPeR_666

Nice review, as always.

Btw there's a Noctua instead of a Thermalright 120 on last page.

it's a thermalright with the noctua fan on it

sotu1 · 16th Jun 2008, 11:45

typo error on the noctua title

dam i'm getting anal in my senility. surprised there's no bit-tech badges applied here to highlight the best of the punch, editors pic, recommended etc. i like those badges. makes reading a non-necessity. or was that the point?

Sark.inc · 16th Jun 2008, 11:53

wth is a delta-t stand for?!?!

i use the Coolermaster CM Sphere and, i was blown away with it's performance, 22c idle and 43 on load.

I am running the fan passively, as the fan that comes with it was to loud :P, with the fan is was 35 or so, but there are 2 fans blowing on to it :P

:edit: mind you, i am only running a E6600..

kenco_uk · 16th Jun 2008, 12:04

I should be receiving a Xigmatek HDT-S1283 from Scan in the next couple of days - from a couple of reviews, it looks to be pretty good, if not better than a TRU. See if it fits my mobo first though

Tim S · 16th Jun 2008, 12:05

Quote:

Originally Posted by cnvrt

wth is a delta-t stand for?!?!

It's the difference in temperature (a delta) between two values (in this case the CPU and ambient temps).

zoot2boot · 16th Jun 2008, 13:49

i think that's a miserable testing methodology. Ambient temps to the nearest whole number? if you're going to test coolers at differing ambient temps then you shouldn't even be displaying the temperatures. it's just misleading.

this, despite being very old, is the way to do it.

http://www.dansdata.com/coolercomp.htm

cjmUK · 16th Jun 2008, 16:12

I know we aren't trying to compete with SPCR, but some sort of metric for noise would be good; perhaps as simple as marks out of 10. Subjective is better than nothing.

Shielder · 16th Jun 2008, 16:26

Quote:

Originally Posted by zoot2boot

i think that's a miserable testing methodology. Ambient temps to the nearest whole number? if you're going to test coolers at differing ambient temps then you shouldn't even be displaying the temperatures. it's just misleading.

this, despite being very old, is the way to do it.

http://www.dansdata.com/coolercomp.htm

Good link. Very interesting. I disagree with the not displaying ambient temps though. How else are you going to prove that you didn't stick the coolers in an oven while testing?

Why do you think it's misleading anyway?

Andy

Baz · 16th Jun 2008, 17:49

Testing in the same ambient temperature is one of the very expensive options we just don't have. In an ideal world we'd test in an air conditioned, climate controlled room, where we can control the temperature to the nearest degree, but the sad fact is we can't, and so have to publish the ambient temps vs highest load temp as the key performance value.

For the same reason, we round to the nearest degree, as consistantly accurate thermometers are not cheap. While I admit it's not 100% accurate, I feel our test methodology is better than most, and gives a solid indication of real world performance, that you'd get if you fitted the heatsink in your own case, not just if you fit the heatsink on a test bench.

Although if anyone wants to donate a laboratory grade thermometer, feel free

liratheal · 16th Jun 2008, 17:54

Quote:

Originally Posted by Baz

Testing in the same ambient temperature is one of the very expensive options we just don't have. In an ideal world we'd test in an air conditioned, climate controlled room, where we can control the temperature to the nearest degree, but the sad fact is we can't, and so have to publish the ambient temps vs highest load temp as the key performance value.

For the same reason, we round to the nearest degree, as consistantly accurate thermometers are not cheap. While I admit it's not 100% accurate, I feel our test methodology is better than most, and still gives a solid indication of performance.

Although if anyone wants to donate a laboratory grade thermometer, feel free

I always get the feeling that I'm being lied to when things are tested in climate controled rooms. I can't speak for everyone, but I'd be quite willing to bet that the large majority of users do not have the ability to control their rooms climate either. As users we're more likely to get similar results to compare with your results, which IMO, is a more realistic view of performance.

pistol_pete · 16th Jun 2008, 19:12

Good to see my TRUE120 is still king of the coolers.

One test I'd like to see is how the cheapest and most expensive coolers (say a TRUE120 and a Freezer 7) compare for different chips. Eg, test bother with a low power dual core, and both with a heavily overclocked quad.

I'd image the TRUE120 would pull out well ahead for the quad and be much quieter, but there would be little difference between it and the Freezer 7 Pro for the low power chip.

Some actual numbers to back this up would help people in choosing a cooler suitable to their CPU.

ViPPeR_666 · 16th Jun 2008, 21:23

Quote:

Originally Posted by Tim S

it's a thermalright with the noctua fan on it

ohhhh hahaha! my bad

Sark.inc · 17th Jun 2008, 08:10

Quote:

Originally Posted by Tim S

Quote:

Originally Posted by cnvrt

wth is a delta-t stand for?!?!

It's the difference in temperature (a delta) between two values (in this case the CPU and ambient temps).

cheers that _b

MiNiMaL_FuSS · 17th Jun 2008, 14:51

Those load temps seem a good 10c under what most other reveiws of the TRU seem to achieve?! 44C LOAD is very low!

impar · 17th Jun 2008, 18:07

Greetings!

Quote:

Originally Posted by MiNiMaL_FuSS

Those load temps seem a good 10c under what most other reveiws of the TRU seem to achieve?! 44C LOAD is very low!

They are too low because:
"All CPU temperatures are recorded from the MSI PC Alert 4 software in the operating system"

Temperatures are not from CoreTemp or RealTemp. One needs to add another 10-15ºC to compare to other reviews.

ParaHelix.org · 19th Jun 2008, 17:36

Water cooling cheesecake!

16th Jun 2008, 08:56	#1
Tim S Pewlius Caesar bit-tech Staff Join Date: Nov 2001 Location: Ascot, Berks Posts: 18,021	Heatsink testing overview http://www.bit-tech.net/hardware/200...ing-overview/1 Reliably testing CPU heatsinks can be a tricky business, with so many variables capable of jeopardising accurate results. For this reason, we've re-assessed our testing methodology, and put some of the most popular and high-performing heat sinks through their paces in our new standard test setup. __________________ Tim on Twitter \| Tim's Blog \| Tim's Flickr The bit-tech Blog \| bit-tech on Twitter \| Join the bit-tech & Custom PC folding team (#35947)

16th Jun 2008, 09:20	#3
wuyanxu quad fuelled, GTX200 powered Join Date: Aug 2007 Location: in a room near Soton Uni. UK Posts: 3,943	oh yeah! TRUE is still the top one. i bet it can drop a lot more degree if you use one or two high CFM fan. so what heatsink are you planning to test next? one of those newer direct contact heatpipe ones? many test sites say they are better than TRUE....... me wants a reliable source. __________________ Death to consolification !!!Core i7 860 @ 4Ghz Cooled by Corsair H50 // Asus P7P55D Deluxe // 8GB of Corsair Dominators @ 1528Mhz 8-8-8-24 // BFG gtx260+ OC2 MaxCore 216SP 896MB // Samsung 64GB SSD // WD Raptor 74GB + Black 1TB + Green 1TB // Corsair Hx620w // Antec p182 // Samsung 24" T240

16th Jun 2008, 11:22	#4
ViPPeR_666 Minimodder Join Date: Jul 2006 Location: Maceió (beach city) - BR Posts: 32	Nice review, as always. Btw there's a Noctua instead of a Thermalright 120 on last page. __________________ "The Christian god can be easily pictured as virtually the same as the many ancient gods of past civilisations. The Christian god is a three-headed monster; cruel, vengeful and capricious. If one wishes to know more of this raging, three headed beast-like god, one only needs to look at the calibre of the people who say they serve him. They are always of two classes: fools and hypocrites." John Adams

16th Jun 2008, 11:45	#6
sotu1 Mini Love Join Date: Aug 2007 Location: England, South of. Posts: 1,169	typo error on the noctua title dam i'm getting anal in my senility. surprised there's no bit-tech badges applied here to highlight the best of the punch, editors pic, recommended etc. i like those badges. makes reading a non-necessity. or was that the point? __________________ Death to consolification !!! [size=1]Give me my Windforce! I'm coming back!

16th Jun 2008, 11:53	#7
Sark.inc Multimodder Join Date: Dec 2007 Posts: 148	wth is a delta-t stand for?!?! i use the Coolermaster CM Sphere and, i was blown away with it's performance, 22c idle and 43 on load. I am running the fan passively, as the fan that comes with it was to loud :P, with the fan is was 35 or so, but there are 2 fans blowing on to it :P :edit: mind you, i am only running a E6600.. Last edited by Sark.inc; 16th Jun 2008 at 12:00.

16th Jun 2008, 12:04	#8
kenco_uk I feel your presents. Join Date: Nov 2003 Location: Bourne, Lincolnshire, UK Posts: 6,316	I should be receiving a Xigmatek HDT-S1283 from Scan in the next couple of days - from a couple of reviews, it looks to be pretty good, if not better than a TRU. See if it fits my mobo first though __________________ MSI X58M - i7 920 4Ghz - 6GB Corsair - GTX260 648/2520 ::: Vote for bit-tech T - theevilelephant - "it sounded like bouncing around on a pogo stick, screaming in chinese while occasionally throwing a cat at a door."

16th Jun 2008, 13:49	#10
zoot2boot Multimodder Join Date: May 2006 Location: aus Posts: 75	i think that's a miserable testing methodology. Ambient temps to the nearest whole number? if you're going to test coolers at differing ambient temps then you shouldn't even be displaying the temperatures. it's just misleading. this, despite being very old, is the way to do it. http://www.dansdata.com/coolercomp.htm

16th Jun 2008, 16:12	#11
cjmUK Old git. Join Date: Feb 2004 Location: UK Posts: 1,946	I know we aren't trying to compete with SPCR, but some sort of metric for noise would be good; perhaps as simple as marks out of 10. Subjective is better than nothing. __________________ P180B, Q6600 [with Thermalright HR-01+/Noctua NF-P12], 8GB OCZ Reaper PC2-8500, EVGA GTX 260 SSC, 3.75TB worth of Samsung F1s, running on Windows 7 RC

16th Jun 2008, 17:49	#13
Baz Benchmarketeer bit-tech Staff Join Date: Jan 2005 Location: Ruislip Posts: 589	Testing in the same ambient temperature is one of the very expensive options we just don't have. In an ideal world we'd test in an air conditioned, climate controlled room, where we can control the temperature to the nearest degree, but the sad fact is we can't, and so have to publish the ambient temps vs highest load temp as the key performance value. For the same reason, we round to the nearest degree, as consistantly accurate thermometers are not cheap. While I admit it's not 100% accurate, I feel our test methodology is better than most, and gives a solid indication of real world performance, that you'd get if you fitted the heatsink in your own case, not just if you fit the heatsink on a test bench. Although if anyone wants to donate a laboratory grade thermometer, feel free

16th Jun 2008, 19:12	#15
pistol_pete Air Cooled Fool Join Date: Jan 2008 Location: Derby, via Scotland Posts: 709	Good to see my TRUE120 is still king of the coolers. One test I'd like to see is how the cheapest and most expensive coolers (say a TRUE120 and a Freezer 7) compare for different chips. Eg, test bother with a low power dual core, and both with a heavily overclocked quad. I'd image the TRUE120 would pull out well ahead for the quad and be much quieter, but there would be little difference between it and the Freezer 7 Pro for the low power chip. Some actual numbers to back this up would help people in choosing a cooler suitable to their CPU. __________________ - Q6600 G0(9x333) -- Asus P5k Premium -- 2GB Ballistix DDR2-800 -- BFG 8800GT 512MB -- TRUE120 -- 500GB Samsung Spinpoint -- Antec Quattro 850W -- Samsung Syncmaster 206BW -- Razer Lycosa, DeathAdder, and Destructor -- Antec Nine Hundred -- Pistol Pete's Project Perspiration Project Log

17th Jun 2008, 14:51	#18
MiNiMaL_FuSS I Mod, Therefore I Own Join Date: Dec 2003 Location: Norwich, Norfolk, UK Posts: 3,079	Those load temps seem a good 10c under what most other reveiws of the TRU seem to achieve?! 44C LOAD is very low! __________________ -Yesterday is History, Tomorrow is a Mystery, Take the Present as a Gift-

19th Jun 2008, 17:36	#20
ParaHelix.org Supermodder Join Date: Jun 2008 Posts: 270	Water cooling cheesecake!