TRUE and Megahalems: Performance with different fan configurations

[muddocktor]

After being a long time user of Thermalright products, I finally got tired of having to lap the base of every heatsink I've bought from them in the last 3-4 years because of substandard base finishes. Then, the Prolimatech Megahalems came out and I started seeing very good results from them being posted here without having to resort to lapping the base straight out of the box so I decided to get myself one and see what it's all about. Since I already own a lapped TRUE, it made for a chance to do a comparison on the performance between the two heatsinks too, using the same fans is several different configurations and see how the performance of both compare.

The fans I am using for this comparison are as follows:

Sanyo Denki - San Ace 109R1212H1011 (X 2)
Specs:

• Fan Size: 120mm x 120mm x 38mm

• Speed: 2600 RPM

• Airflow: 102.5 CFM

• Noise: 39 dBA

• Static Pressure: 64.7Pa (~6.6mm H2O)

• Bearing: Ball bearing

• Power: 6.24 W

Yate Loon High - D12SH-12 (X 2)
Specs:

• Fan Size: 120mm x 120mm x 25mm

• Speed: 2200 +/- 10% RPM

• Airflow: 88.0 CFM

• Noise: 40 dBA

• Bearing: Sleeve

• Power: 3.6 W

Yate Loon Medium - D12SM-12 (X 2)
Specs:

• Fan Size: 120mm x 120mm x 25mm

• Speed: 1650 +/- 10% RPM

• Airflow: 70.5 CFM

• Noise: 33.0 dBA

• Bearing: Sleeve

• Power: 3.6 W

Noctua NF-P12 - NF-P12-1300 120mm Fan
Specs:

• Fan Size: 120mm x 120mm x 25mm

• Speed: 1300 +/- 10% RPM

• Airflow: 54.3 CFM

• Noise: 19.8 dBA

• Bearing: SSO-Bearing

• Power: 1.08 W

Scythe Slipstream (medium) - SY1225SL12M
Specs:

• Fan Size: 120mm x 120mm x 25mm

• Speed: 1200 RPM

• Airflow: 68.54 CFM

• Noise: 24 dBA

• Bearing: Sleeve

• Power: 3.12 W

Some personal notes on the fans:

• Specs for the fans were pulled off of the Petra's Tech Shop website except for the Noctua, which I got off of Jab-tech's website.

• One Yate Loon Medium fan was bought from PTS early this year by me. The other Yate Loon Medium was donated to me for this test by Jab-tech. I would like to thank Jab-tech for sending me the second Yate loon Medium fan when I was ordering extra fan clips for my Megahelams so I could test with 2 Yate mediums in push-pull. Thanks John, I really appreciate it!

• I only had 1 Noctua NF-P12 fan and I wasn't going to cough up another $20 to get another for this test, so I subbed in a Scythe SY1225SL12M Slipstream to take the second fan's place in push-pull configs. Previous testing of this fan showed me similar results to the Noctua fan so I think it's a valid substitution for a second NF-P12.

• I have no measuring equipment to test this, but I think the cfm figure is inflated and the noise figure is deflated for the Noctua fan.

• I believe the cfm figure is greatly inflated for the Slipstream.

• The thermal grease used throughout the testing was some Thermalright white goop they used to package with their heatsinks. It's not the best in the world, but it needs no curing time and gives repeatable results. I was looking for those 2 features primarily in this test for my TIM. Using a better quality TIM would probably net a 3-5 C decrease in temps over this TIM material.

The test configuration is a system made of various spare parts I had laying around, consisting of:

Antec 1040 style case (Actually a Chieftech clone)
Hiper M1000 psu (O_W review sample I bought from him)
Gigabyte P35-DS3R mobo (no i/o backplane panel installed)
Q6600 overclocked to 3600 with the voltage running between 1.52 to 1.48 at load. I wish I could have gotten a little more voltage, but this Gigabyte board has no load line calibration and absolutely horrid vdroop and vdrop. These voltages were what I saw when I had 1.6v vcore set in bios.
eVGA 7900GTX vid card
2 X 1 gig kit of OCZ PC2-8500 Platinum DDR2 ram
old 20 gig 7200 rpm IDE Maxtor hard drive
Lite on IDE dvd burner
OS: Win7 RC, 32 bit version

I used OCCT version 3.0.1 as the heat source since it will run a one hour test automatically and also gives a graphical result as a series of png files at the end of the run for each core. I then went through the pics and pulled the high and low temps off of each core. I maintained room temps as close as I could to a range between 20.55-21.11 degrees C (69-70 F).

I decided to do this comparison inside a case instead of open air to see what actual performance differences might be seen in something that the normal person would actually use (instead of enthusiasts like us who will run them nekkid ). The case design is an older one that uses 2-80mm fans on the back side for ventilation and while it's not as good as newer cases in this regard it's still not bad, just a little noisy. And I don't have the i/o backplane panel installed around the i/o ports of the mobo so that gives extra room for the push-pull configurations to exhaust air.

Here are the results I got with the TRUE and various fans:

TRUE with 2 San Ace fans in push-pull:
Max temp - 73 C
Min temp - 36 C
CPU1 max temp - 73 C
CPU2 max temp - 73 C
CPU3 max temp - 69 C
CPU4 max temp - 69 C

TRUE with 1 San Ace fan in push config:
Max temp - 73 C
Min temp - 37 C
CPU1 max temp - 73 C
CPU2 max temp - 73 C
CPU3 max temp - 68 C
CPU4 max temp - 68 C

TRUE with 2 Yate Loon Highs in push-pull:
Max temp - 75 C
Min temp - 37 C
CPU1 max temp - 75 C
CPU2 max temp - 75 C
CPU3 max temp - 71 C
CPU4 max temp - 71 C

TRUE with 1 Yate Loon High in push config:
Max temp - 77 C
Min temp - 40 C
CPU1 max temp - 77 C
CPU2 max temp - 77 C
CPU3 max temp - 72 C
CPU4 max temp - 72 C

TRUE with 1 Yate loon Medium in push config:
Max temp - 78 C
Min temp - 38 C
CPU1 max temp - 78 C
CPU2 max temp - 77 C
CPU3 max temp - 72 C
CPU4 max temp - 72 C

TRUE with Noctua in push and Scythe in pull config:
Max temp - 77 C
Min temp - 38 C
CPU1 max temp - 77 C
CPU2 max temp - 77 C
CPU3 max temp - 72.5 C
CPU4 max temp - 71.5 C

TRUE with Noctua in push config:
Max temp - 81 C
Min temp - 42 C
CPU1 max temp - 81 C
CPU2 max temp - 81 C
CPU3 max temp - 75 C
CPU4 max temp - 75 C

• NOTE:I didn't have the second Yate Loon Medium when I was running these series of tests on the TRUE. I did try to retest after I received the second YL Med but ran out of time before I had to leave for work offshore.

Now the Prolimatech Megahalems test results:

Megahalems with 2 San Ace fans in push-pull:
Max temp - 72 C
Min temp - 35 C
CPU1 max temp - 72 C
CPU2 max temp - 72 C
CPU3 max temp - 66 C
CPU4 max temp - 66 C

Megahalems with 1 San Ace fan in push config:
Max temp - 73 C
Min temp - 34 C
CPU1 max temp - 73 C
CPU2 max temp - 73 C
CPU3 max temp - 67 C
CPU4 max temp - 68 C

Megahalems with 2 Yate Loon Highs in push-pull:
Max temp - 73 C
Min temp - 37 C
CPU1 max temp - 73 C
CPU2 max temp - 73 C
CPU3 max temp - 67.5 C
CPU4 max temp - 68.5 C

Megahalems with 1 Yate Loon High in push config:
Max temp - 75 C
Min temp - 35 C
CPU1 max temp - 75 C
CPU2 max temp - 75 C
CPU3 max temp - 70 C
CPU4 max temp - 70 C

Megahalems with 2 Yate Loon Mediums in push-pull:
Max temp - 74 C
Min temp - 37 C
CPU1 max temp - 74 C
CPU2 max temp - 74 C
CPU3 max temp - 69 C
CPU4 max temp - 68.5 C

Megahalems with 1 Yate loon Medium in push config:
Max temp - 76.5 C
Min temp - 37 C
CPU1 max temp - 75 C
CPU2 max temp - 76.5 C
CPU3 max temp - 71 C
CPU4 max temp - 71 C

Megahalems with Noctua in push and Scythe in pull config:
Max temp - 77 C
Min temp - 36 C
CPU1 max temp - 77 C
CPU2 max temp - 75 C
CPU3 max temp - 71 C
CPU4 max temp - 71 C

Megahalems with Noctua in push config:
Max temp - 89 C
Min temp - 46 C
CPU1 max temp - 89 C
CPU2 max temp - 89 C
CPU3 max temp - 87 C
CPU4 max temp - 87 C

NOTES:

• The Noctua fan by itself is totally inadequate to handle the Megahalems heatsink. It's static pressure and cfm are too low to provide adequate airflow through the thicker cross section of this design compared to the TRUE. All other configurations of fans showed equal or better performance than a lapped TRUE. I didn't monitor for it, but from looking at the posted temps I am sure that the processor had throttled due to high heat while making this run with the Noctua. The temperature difference between cores 2 and 3 are too close to the max temps on cores 0 and 1. This processor has always shown a definite separation between the 2 series of core temps like all the other test runs have shown.

• I believe the lack of meaningful temp differences with both heatsinks and the San Ace config runs is due to the extremely tight space between the back of the heatsinks and the case back (and exhaust fans) constricting airflow out of the heatsink and case. Any gain you would have gotten from the push-pull configuration is lost due to flow restriction between the case back and heatsinks.

Whenever I get in from work this time I plan to do some more messing around with these heatsinks. Thermalright has come out with a new mounting system for the TRUE that I am going to experiment with and I might even try using the Prolimatech mount on my TRUE and see how it does too. The new TR mount looks very promising and it holds the heatsink much more stably than an unmodded old style "X" mount, but I didn't see the temps I was expecting either from it. I was extremely rushed though due to having to get ready to go offshore so I'm not sure it just was a problem with me mounting it in a rush.

 

[baditude_df]

This is a kick *** thread mud. Thx for taking the time to test, record and post.
As you know, I've done a bit of work comparing these two coolers as well, and it was really interesting seeing these results and compare with mine(ie: cased vs caseless). Even though we're comparing 65nm Conroe with 45nm Nehalem, the parallels between the two coolers are quite similar.
I personally love the mounting on the Mega, but am also fine with the original mount on the TRUE.

 

[muddocktor]

Thanks for the compliment, bad. Your thread is pretty kick *** itself and our threads also complement each other. Yours is on the LGA1366 platform and mine is on LGA775 and your are tesing in free air and I'm in a case.

BTW, here's a shot of the new TRUE mounting system next to the old TRUE mount for you to look at. It shows a resemblance to the Mega's mount.

 

[Brando]

It's about time they fixed the mounting on the TRUE. Finally got it right. The old mounting type was slowly adding to my hate of the TRUE each time I removed and reinstalled it. This makes it legit imo.

 

[burebista]

Thanks mud. Perfect test and at last a test in case closed.
May I put here a picture from my friend with 2 x Delta PFB1212UHE – 252 CFM, 5500 RPM in open stand?



Otherwise your testing matches his testing.
Thanks again, I really like to see this kind of tests in forums.

 

[bing]

Excellent thread Mudd, thanks !

Just fyi, I'm using the old Tuniq Tower 120 thick H steel bracket on my TRUE instead of that old crappy scissor bracket, and managed to squeeze out 2 to 3 C more.

Also if you don't mind, before you put the mobo back to the case, please take a shot of the close up on the middle bracket pressing on the HSF, really curious how good those new revision mounting mechanism on true.

For example shot like this :

Looking forward & eagerly waiting for your result using that new TR mounting.

 

[Zonnza]

I posted a reply to you post in my thread muddocktor, might wanna check it out. This thread is definitely awesome. One thing that I ask in my thread is which way I should orient my megahalem if i pick it up, I don't know if it will make a difference in temperatures but if it does I would definitely want to hear other's opinions on it.

 

[muddocktor]

@Zonnza
I'll reply to you in your thread on my thoughts on your setup. That Antec case you are using does give you some good orientation options for the megahelams.

@bing:
Bing, all I have of the new mount in use so far is 1 pic that's halfway close up. Here's a crop of the mount. Since I'm at work presently I can't take any more pics or do any experimenting until I get back in. Interestingly, the box for the new mount is labeled for the Ultra series, plus something I've never heard of called a "Venomous X". Maybe that's the name of a new heatsink they plan to market soon.

@burebista:
Thanks, I always like to see something gone to the extremes! I can't even imagine how loud that was when he was testing it.

 

[bing]

@Zonnza
@bing:
Bing, all I have of the new mount in use so far is 1 pic that's halfway close up. Here's a crop of the mount. Since I'm at work presently I can't take any more pics or do any experimenting until I get back in. Interestingly, the box for the new mount is labeled for the Ultra series, plus something I've never heard of called a "Venomous X". Maybe that's the name of a new heatsink they plan to market soon.

Thanks for that shot, just take your time, and if you had the chance again, please take a close up shot horizontally like my example.

Yep, the name "Venomous X" is the upcoming new replacement for the current TRUE, links -> HERE OR HERE.

 

[muddocktor]

OK, looks like this Venomous X has been renamed the TRUE rev.C, after reading the HardOCP review and looking at their pics of the Rev. C TRUE. Not really a new heatsink, but rather a tweaking of the design. But if they have improved their base finishes then I'm all for it.

 

[baditude_df]

Base finish, Mount, and removal of the silly hole through the middle like with the RevB. Are the confirmed upgrades with the RevC by all accounts.

My TRUE Black is an original, but it has the base of a RevC. I'm VERY happy with it, and actually someone else is going to be happy with it come Dec 1st. .

 

[Zonnza]

here's a question that just crossed my mind... i thought there was only 1 megahalem out; is there multiple versions of it? and which one is best for a i7 920?

 

[baditude_df]

Mehahalems, mega shadow and megahalems Rev B.

The original is what you would want for an i7. No need to shell out for anything else.
Not sure that the RevB would have any performance gains based on what I've seen at their site. Just a Scratch and mark resistant top plate and support for 1156.

 

[Kuroimaho]

Thanks for sharing it, will help with finding the right fan. The new mounting system seems like what they used for Socket 771.

 

[mattspalace]

Hey mud - I'm using a prolimatech Mega Shadow with a pair of the Noctua P12 fans and have idle temps around 36c and load of 73c - this is with my i7 920 CPU at 4.2GHz with 1.28v. Case airflow and ambient temps are going to play a major role in any temp comparisons.
I do agree with you that the Noctua fan is not 20dB, but more like 22-23dB and probably about 49-50CFM. But in my HAF 932, they work excellent and I have a nice quiet system to boot.
I'm not cool with 40dB fans, and will also never buy sleeve bearing fans again. I had two I used with my TRUE (CoolerMaster versions) and both started rattling within about six months of use. Sleeve bearings are quiet but they don't last long - at least not in my case.

Comparison looks very good. I also hate the mounting system for the TRUE with the scissor bracket - my last HSF. It performed excellent though. I LOVE the mounting systems for the prolimatech HSF. Very good design - maybe a bit overdone, but it's outstanding. I like that new mount for the TRUE - might have to look into that!

 

[Jolly-Swagman]

Very nice write up Mud, and the new TRUE bracket looks allot better

 

[muddocktor]

mattspalace, yeah I don't doubt that the old case I used affected my results somewhat compared to a newer and better designed case. And I had purposely jacked my vcore as high as I possibly could at this speed for the most heatload I could run too as that Q6600 normally runs that speed at around 1.4v for 24/7 use. The problem I have with testing with a more modern case is that all the other systems in my house are rigged with loops and are under water and this is the only other case that's available to test with.. As for the Noctua fans, I imagine you can get them much cheaper than us since you are in Germany and they are an Austrian based company. They are well built fans, but for the price you pay for them here in the US I can get the S-Flex fans and have the shipping paid for what the Noctua fans cost without shipping. And the S-Flex uses a fluid type bearing too, just like the Panaflo and Noctua fans do.

As for the new TRUE brackets, I have a few quick observations I made before going to work. The new backplate included has longer threaded holes that protrude through the mobo more than the original bracket does. When you mount it to the mobo it actually is loose and not made up tightly to the mobo. I think the new mount would probably work even better with the original TR backplate or an even better choice would be a Swiftech backplate, which will make up tightly to the mobo when tightening down the mounting bracket. That would eliminate some of the slop I saw in mounting the new TR mount on the mobo. Also, since they use stamped steel components for the mount instead of milled aluminum like the Mega uses, I noticed some flexing and deflection in the side rails when tightening down the heatsink. That might affect mount quality too, but I didn't have a chance to look into this much yet.

 

[mattspalace]

heh, you'd think they would be cheaper in Germany, but once you do the conversion to EUR from what you pay in USD, they aren't. And they are hard to find - actually all high-end computer gear is hard to find in Germany, so I end up ordering from the states and having it shipping to my APO address versus my actual German address.
I think I found the Noctua fans for $17 or so from either Sidewinder or SVC - it was between these and the S-flex you mentioned. Since both are fluid bearings it was a toss up for me, but I liked the cabling used on the Noctua's so that's what I chose. They'll do; I'm not going to break any records using air cooling.

 

[burebista]

mud, bad, I see that you like to test new things. I just spoke with my crazy friend (with that thorough 23 coolers roundup) and he's astonished by Noctua NH-D14. It's all that TR IFX-14 is not and sliiightly better than Megahalems.
Some preliminary tests here.

 

[bing]

Burebista, don't worry, my gut feeling is telling me he will review this new beast in near future, don't you Mudd ?






.