SPCR's 2010 CPU Heatsink Test Platform

Want to talk about one of the articles in SPCR? Here's the forum for you.
jtcb
Posts: 104
Joined: Sat Dec 13, 2003 10:58 am
Location: Vancouver, BC Canada

Post by jtcb » Sat Jan 30, 2010 9:58 am

Why isn't the Noctua NH-D14 in the test? Would Prolima Megahalems still br the best if Noctua NH-D14 is included? Anyone has link to review where these two go head to head?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sat Jan 30, 2010 10:13 am

jtcb wrote:Why isn't the Noctua NH-D14 in the test? Would Prolima Megahalems still br the best if Noctua NH-D14 is included? Anyone has link to review where these two go head to head?
Look, understand what the article is: It's to introduce a new test plaform, and to establish reference points for future reviews by retesting big HSF that we've already tested and still have on hand.

We also did this with many products when the anechoic chamber and high resolution audio test gear was introduced: Re-measure and re-record the noise to establish reference benchmarks for future reviews.

And it so happens that I'm working right now on a NH-D14 review w/ results from the new test platform.
Last edited by MikeC on Sat Jan 30, 2010 6:17 pm, edited 1 time in total.

jtcb
Posts: 104
Joined: Sat Dec 13, 2003 10:58 am
Location: Vancouver, BC Canada

Post by jtcb » Sat Jan 30, 2010 10:36 am

Sounds good. Please post review of the NH-D14 soon. I am debating between these two HSF.

loimlo
Posts: 762
Joined: Thu May 03, 2007 3:58 am
Location: Formosa

Post by loimlo » Sun Jan 31, 2010 1:10 am

K.Murx wrote:
zds wrote:-If you want the very best performance, go for Megahalems.
-If you have very hot CPU and want bang-for-buck, get Scythe Mugen2
-If you have less hot CPU, consider also Ninja Mini and Big Shuriken.
HDT 1283: $23 after MIR / 22 Euros
Scythe Mugen: $35 / 33 Euros
HDT is significantly lighter (less stress on MB), and has better cooling performance (tested on the Pentium D).
Would you like to reconsider?
ekerazha wrote:You should also update your burn-test application, LinX ( http://www.xtremesystems.org/forums/sho ... p?t=201670 ), based on Intel LINPACK, is far better than Prime95.
But... LINPACK has long since been superseded by LAPACK, which has largely been superseded by FLAME in the HPC community.
And why on earth would you use a vendor provided library for benchmarking???
Additionally, all those Libraries rely heavily on the BLAS, so performance and load will vary a lot with whatever (and please, NOT vendor-specific) implementation of BLAS is chosen...
Xigmatek 1283's issue may be the Intel stock mounting kit compared to Mugen 2's superior backplate. That said, I have purchased 775 backplate at a bargain price, 4 dollars, to solve this issue for a friend's computer. It's hard to beat it given its price and performance. Bottom line? I'm also for 1283 testing, especially with DIY backplate/retention kit.

Image

Tzupy
*Lifetime Patron*
Posts: 1561
Joined: Wed Jan 12, 2005 10:47 am
Location: Bucharest, Romania

Post by Tzupy » Mon Feb 01, 2010 9:04 am

@MikeC: I hope I'm not missing something, but: the Thermalright U-120 is discontinued and the new heatsink used in the CPU & mobo reviews is Thermalright MUX-120.
Which MUX-120 is very similar but maybe not identical with the U-120. I couldn't find a SPCR review of the new MUX-120, so I suggest to do it, since you are using it as a reference heatsink / fan.
I'd also like to see the Ninja2 reviewed on Skt 1366 and Skt 1156, WITH bolt-through mounting. I believe that a review of bolt-through kits would be quite useful for the SPCR folk. For CPUs in the 45W-65W range many heatsinks don't come with bolt-through, but with the hated (by me, at least) push-pins, and it can be a hassle to find the right bolt-through kit.

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Mon Feb 01, 2010 12:22 pm

Tzupy wrote:I'd also like to see the Ninja2 reviewed on Skt 1366 and Skt 1156, WITH bolt-through mounting.
I think you'd be disappointed. The newer Ninjas have two deficiencies relative to the original Ninja: the push-pin mount, which as you note can be replaced, and the base plate/heat pipe interface, which was both pressed and soldered on the original. On the newer Ninjas, there is no solder, and there are visible gaps in the press-fit.

So even if the low-pressure mounting between the base plate and the CPU IHS is fixed, the heat transfer from the base plate to the heat pipes is compromised.

This is why the 1283 does so well: if properly mounted, the heat pipes have outstanding heat transfer from the IHS.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Feb 01, 2010 1:16 pm

cmthomson wrote:
Tzupy wrote:I'd also like to see the Ninja2 reviewed on Skt 1366 and Skt 1156, WITH bolt-through mounting.
I think you'd be disappointed. The newer Ninjas have two deficiencies relative to the original Ninja: the push-pin mount, which as you note can be replaced, and the base plate/heat pipe interface, which was both pressed and soldered on the original. On the newer Ninjas, there is no solder, and there are visible gaps in the press-fit.

So even if the low-pressure mounting between the base plate and the CPU IHS is fixed, the heat transfer from the base plate to the heat pipes is compromised.

This is why the 1283 does so well: if properly mounted, the heat pipes have outstanding heat transfer from the IHS.
Agreed. The Ninja's time has come and gone -- Scythe insists on staying with the weird metal adapter to convert a 775 mount to 478 for "bolt-through" mounting of the Ninja (and a few other HS like the monster Orochi). This ends up being no more secure than the standard push-pins. Reviewing such things encourages them. I want nothing to do with such huge and heavy products until they finally incorporate decent mounting systems. The Mugen2 is a step forward, but it's still a long way from Megahalems or any of the Noctuas.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Mon Feb 01, 2010 6:58 pm

MikeC wrote:
cmthomson wrote:
Tzupy wrote:I'd also like to see the Ninja2 reviewed on Skt 1366 and Skt 1156, WITH bolt-through mounting.
I think you'd be disappointed. The newer Ninjas have two deficiencies relative to the original Ninja: the push-pin mount, which as you note can be replaced, and the base plate/heat pipe interface, which was both pressed and soldered on the original. On the newer Ninjas, there is no solder, and there are visible gaps in the press-fit.

So even if the low-pressure mounting between the base plate and the CPU IHS is fixed, the heat transfer from the base plate to the heat pipes is compromised.

This is why the 1283 does so well: if properly mounted, the heat pipes have outstanding heat transfer from the IHS.
Agreed. The Ninja's time has come and gone -- Scythe insists on staying with the weird metal adapter to convert a 775 mount to 478 for "bolt-through" mounting of the Ninja (and a few other HS like the monster Orochi). This ends up being no more secure than the standard push-pins. Reviewing such things encourages them. I want nothing to do with such huge and heavy products until they finally incorporate decent mounting systems. The Mugen2 is a step forward, but it's still a long way from Megahalems or any of the Noctuas.
Is the Orochi rev B new mounting system any good?
http://www.scythe-eu.com/en/products/cp ... rev-b.html

Image

danimal
Posts: 734
Joined: Mon Jun 08, 2009 2:41 pm
Location: the ether

Post by danimal » Mon Feb 01, 2010 10:47 pm

PlanetOfTheApes wrote:[Is the Orochi rev B new mounting system any good?
http://www.scythe-eu.com/en/products/cp ... rev-b.html
that looks rather similar to the mugen 2 mount... if so, it's pretty solid; the screws bottom out, to put a preset load on the cooler.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Feb 01, 2010 11:12 pm

danimal wrote:
PlanetOfTheApes wrote:[Is the Orochi rev B new mounting system any good?
http://www.scythe-eu.com/en/products/cp ... rev-b.html
that looks rather similar to the mugen 2 mount... if so, it's pretty solid; the screws bottom out, to put a preset load on the cooler.
But it's really not that good still. Total pain to install, even if you do this kind of stuff all the time, some risk to motherboard due to user error caused by awkward installation.

If you look at this page of our review, you'll see that there are several parts you have to align -- the heatsink upside down on on table, the mobo, upside down on heatsink, the backplate, atop the trace side of the board, and finally the screw which must go in to three holes in the previous 3 parts. Nothing is automatically aligned, all three parts are complete free to move around, themal paste wants to slide everywhere. And you have a screwdriver to handle too. Larry & I call this a total PITA.

danimal
Posts: 734
Joined: Mon Jun 08, 2009 2:41 pm
Location: the ether

Post by danimal » Mon Feb 01, 2010 11:31 pm

the mugen 2 is at the max size/weight limit for me, but the price/performance ratio makes it worthwhile... i wouldn't want to hassle with the orochi.

i'm able to get the mugen 2 mounted pretty well with clear packing tape holding the screws, it doesn't make contact with the paste until the final drop, but there is a lot of room for error... the indigo thermal interface is really sticky, where you set the cooler down is where it stays.

edit: i just looked at the 1366 mounting instructions for the orochi, it looks the same as the mugen 2; the replacement cpu cover bolts up to the backplate, so the only thing that's loose are the four screws, which should be taped in from the backplate side, with each screw hanging over the table, so that it can be started, then taped back up... flip the entire assembly over, then drop the mb onto the cooler... still a hassle, tho.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Tue Feb 02, 2010 1:23 am

Ok sounds like the Orochi is a pain to install. But I don't want to dismiss the Orochi until I find another cooler with a better mounting system and the same or better passive performance.

Vibrator
Posts: 94
Joined: Sun May 31, 2009 9:46 am
Location: British Columbia

Post by Vibrator » Tue Feb 02, 2010 10:53 pm

Interesting.

Wouldn't it be better to use a thermal paste that is accessible to consumers though? These reviews are mostly read by consumers, after all.

Something like OCZ Freeze or MX-3?

Jadukey
Posts: 20
Joined: Wed Oct 28, 2009 5:30 pm
Location: USA

low power

Post by Jadukey » Wed Feb 03, 2010 7:00 pm

95W CPU coolers tested at max TDP is great and all, but what guidance does this provide for low power / true silent systems?

I was inspired by the true 'silent' pc SPCR built with fanless CPU and SSD. I built my own using a ninja2 thinking the wider fin spacing would be superior for fanless operation. My PC pulls 45watts from the wall and i have a big vented case so I dont need fans and the noise they bring. I went ahead and installed a switch + fan in case I ever taxed my 65W TDP CPU to high temps, but my usage never hits the thermal envelope of the components, and the fan + switch is used for entertainment purposes only.

after having a silent pc, i don't plan to revert to fan usage, and will buy passive cooling hardware in the future. I look forward to seeing my silentpcreview website review CPU coolers that best accomodate passive / silent (0db) computer builds.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Feb 03, 2010 7:34 pm

Jadukey --

Good for you, nice that SPCR inspired you to a silent PC.

We've never tested heatsinks w/o a fan, tho, not in any consistent way. Not even back in the days when a P4 was our test platform. Fanless cooling is fairly easy to accomplish today if you choose low power components, think clearly about convection cooling, and choose an appropriately large & open-spaced fin heatsink.

But for us to review HS fanlessly -- this will not happen. There's no point. Whether a passively cooled CPU/system works depends so much on the thermal envelope of the components, operating conditions/demands. These vary... way too much. Your post is a perfect example of users not needed a formal fanless HS review to accomplish this. It won't be done by a mainstreamer, only by those who have the time, interest and make the effort to pull it off.

Our focus will continue on testing heatsinks at very quiet reference levels mostly with the same fan. For the vast majority of readers, this information is extremely useful, and not available elsewhere.

In my own builds, I shy away from fanless: I've had too many issues over the years with components that have had shortened lifespans due to heat damage from running fans too slow or w/o any fan. A little airflow is just as quiet for the most part, and far safer...

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Thu Feb 04, 2010 8:36 am

What about a "Passive Systems" forum, similar to "Green Computing" ?

Maybe a poll can be done to gauge interest.

Jadukey, why not show your rig in the General Gallery, if you haven't already done so.

Jadukey
Posts: 20
Joined: Wed Oct 28, 2009 5:30 pm
Location: USA

Post by Jadukey » Thu Feb 04, 2010 6:06 pm

While I'd argue that a real "mainstreamer" is simply buying his e-machine/hp/dell from walmart, I don't think a passive or nearly passive computer requires much more effort than anyone already buying their own CPU heatsink or other components. From my perspective, the passive cooling market has a longer product cycle with fewer products / vendors to compare. I understand the reasons against reporting this market segment, but had to ask anyway :wink:

PlanetofTheApes: I put a picture of my PC setup in the gallery.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Thu Feb 04, 2010 7:54 pm

Jadukey thanks for putting your rig up, here's the link

viewtopic.php?t=57460

colm
Posts: 409
Joined: Tue Jan 31, 2006 8:22 am
Location: maine

article discussion

Post by colm » Thu Feb 11, 2010 11:20 am

Hey that was great read. the honesty keeps me lurking.

I may be wrong, but I still think the prescott 3.4e is written as 103w and should be 130w..in fact, I could have sworn the early revisions had to be changed because the world never saw so much heat ever, from any cpu, and the stepping changed. I have an early, known as "the hot one". also multiple cores stop energy , the prescott has 43 stage pipeline, there isno getting away...especally as pin grid array. (That is a good thing)

5 years after starting, I am yet again upgrading the dreaded vid editor..for just a few more facts that work with a giant cpu. It is not old , to get to the point..
Also not old is memory controller hubs with active ECC, I am running one now, the 875p chipset. with ecc memory installed. it is Colassal.
I joked of a 2800 prescoot sounding like a jet when loaded, this 3400 is the largest worker I have ever heard (pleasant sound)

Both mentioned can add to a stronger testing demanding integrity.

You know you got a big system, when reverse cooling is the ONLY answer as minimum requirement.

other than my humble findings, the article is spot on correct, and one can simlpy apply it to thier own for slight differences. I still use xp90 with a strict duct, because of this place.

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Sat Mar 06, 2010 2:10 pm

I did a quick comparo of load test software running on Core i7 @ stock speed with stock Intel cooler. Did not measure CPU temp actively, did log power usage (at wall socket):

- Prime95 (latest version: 155W
- 8 instances of K7Burn: 167W (linkie: http://www.madshrimps.be/files/mirrors/K7burn.rar)
- LinX: 200W !

quite a noticeable difference... 45W!

Bat
Posts: 279
Joined: Sun Jan 05, 2003 12:09 pm
Location: U.K.

Post by Bat » Fri Mar 12, 2010 6:12 am

It strikes me as very odd that the rankings can change when you change nothing but the heat power input, as you do when going from stock i7 processor to overclocked.

I'm having trouble believing it's real.

Either there's something odd about the physics that I've failed to consider, or the apparent change is due to random variations between test runs, perhaps due to mounting the heatsink slightly differently between the stock and overclocked tests.

Thoughts? Mike?

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Fri Mar 12, 2010 6:23 am

It strikes me as very odd that the rankings can change when you change nothing but the heat power input, as you do when going from stock i7 processor to overclocked.
each heatpipe has a certain capacity to transfer heat, at a certain heat level it's possible they fail or work less efficient. take into account the fin, orientation, mounting method, and change in thermal load will cause in change of ranking;

Bat
Posts: 279
Joined: Sun Jan 05, 2003 12:09 pm
Location: U.K.

Post by Bat » Sun Mar 14, 2010 4:12 am

Ah, yes, that's plausible.

Heatpipes are going to have nonlinear behaviour in terms of power carried as a function of temperature difference... and absolute temperature is going to matter too.

So for a given heatsink if you were to draw a graph of the temperature of the processor as a function of power (for fixed room temperature and fan speed) it would start off shallow and then grow steeper.

The graph for a different heatsink might start off shallower (better performance) at low powers, but then grow steep sooner as most of the liquid in the heatpipes turns to vapour. Then it would cross the first heatsink's graph and end up worse at higher powers.

mdrumt
Posts: 62
Joined: Tue Jul 21, 2009 8:55 pm
Location: Australia

Post by mdrumt » Sun Apr 11, 2010 5:00 pm

Great review MikeyC.

Lots of info here, together with other sites reviews i really appreciate the time you guys put into testing these things out! I think you did a great job, cant wait for more tests on newer heatsinks on your new platform :)

There was another excellent review i think @ Benchmark Reviews demonstrating the difference in cooling performance across platforms (AM3, 775, 1366 etc) and how this can make a difference to each cooler esp the heat-pipe direct touch coolers.

So, when we gonna see some new silent (orinetated) case reviews from Antec!? ;)

dev
Posts: 43
Joined: Sun May 10, 2009 12:47 am
Location: Here

Post by dev » Sun Apr 11, 2010 10:01 pm

Its funny you mentioned the inconsistency in results across a longer period of time around the same time my friend Ramiro noticed the same thing. We came to the conclusion that even if the temperature is the same there are other factors that affect the results which are not temperature dependent and can't really be controlled. He is doing a massive HDT vs regular heatsinks roundup and he was surprised how well some of them are going at lower voltages. After retesting the likes of Noctua and TRUE Copper in the same session the situation went back to normal.

I think the settings you are using won't affect the results that much but someone using higher loads might get bigger differences.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Apr 12, 2010 6:34 am

dev wrote:I think the settings you are using won't affect the results that much but someone using higher loads might get bigger differences.
.
Oh, the differences were big enough, when the up/down power/temp effects occurred -- 6 degrees C or more is an enormous difference. The big swings in power/temps seemed to occur only after extended high load testing at OC'd/OV'd settings without any cooling off periods in between. So one thing we've established as standard procedure is to use a honking big fan between test runs to ensure the board always has a chance to cool off before starting another test run.

If you are referring to the last comments on page 5, they're just a note that simply reinstalling and re-running a thermal stress test on the same HS will give you slightly different results. But I really can't see this variance being so dependent on CPU power (except when it is really low); ours is already pulling somewhere over 120W.

Camacha
Posts: 41
Joined: Tue Nov 17, 2009 7:40 pm
Location: The Netherlands

Post by Camacha » Mon Apr 19, 2010 5:10 pm

MikeC wrote:Why isn't the Noctua NH-D14 in the test?
And it so happens that I'm working right now on a NH-D14 review w/ results from the new test platform.
I am very curious to see the results. Most websites just focus on performance, while silence is also quite important to me, even though I do really like performance. I was, after seeing what SPCR thought of it, highly interested to buy a Megahalems, but now I'm waiting to see what the NH-D14 is doing.

It might be a tad too massive anyway, but I like to make informed choices :)
Last edited by Camacha on Mon Apr 19, 2010 5:13 pm, edited 3 times in total.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Apr 19, 2010 5:56 pm

Camacha wrote:...but now I'm waiting to see what the NH-D14 is doing.
The review has been up for quite a while...

echn111
Posts: 161
Joined: Thu Nov 02, 2006 2:15 pm
Location: UK

Post by echn111 » Fri Apr 23, 2010 9:42 am

I recall the previous best in class low airflow cooler for noise oriented enthusiasts, was the Thermalright HR-01 Plus, as reported by SPCR.

I don't mind not including in the recommendations if it was merely 3rd or worse. But it was the #1 low airflow cooler, clearly beating the Thermalright Ultra-120 eXtreme and others. The SPCR recommendation was "if you want the very best low airflow cooler, the HR-01 Plus is unrivaled today".

I don't mind not including it if it won on a non-quiet PC type criteria like overclocking or high airflow. But it won as the best low airflow cooler, which is a very relevant criteria for a noise oriented site. No one cares about high airflow with 2000rpm+ fans. SPCR writes: "The Thermalright HR-01 Plus is better suited for ultra-quiet cooling than either the Thermalright Ultra-120 eXtreme or...due to its unmatched performance with extremely limited airflow."

I'm amazed that it wasn't included, on this site, of all sites.

btw: SPCR reviews have the HR-01 Plus beating the TRUE by a significant margin and also the Prolimatech Megahalems by a fraction at low airflow (i.e. 7v). The Megahalems does better at high airflow. According to SPCR anyway.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri Apr 23, 2010 10:11 am

echn111 wrote:I recall the previous best in class low airflow cooler for noise oriented enthusiasts, was the Thermalright HR-01 Plus, as reported by SPCR.

I don't mind not including in the recommendations if it was merely 3rd or worse.
Not sure what you're referring to. The HR01 is still at the top of our recommended heatsinks list: http://www.silentpcreview.com/Recommended_Heatsinks

Post Reply