10-Way heatsink roundup

[WhipperSnapper]

It looks like you've collected a large amount of data. I agree with the other posters that your chart, while packed full of information, takes some effort to decipher. I think it would be good to make bar graphs of the CPU termperatures at various noise levels as another poster suggested. For example, you could have a bar graphs for every HSF combination that came in at noise levels of 34-38, etc.

I also would have liked to have seen some data for some of the more established and more popular heatsinks included, especially the Scythe Ninja, and the Thermalright line (XP-90, XP-90 copper, SI-120), and maybe add that Gigabyte HSF that looks like an XP-90, and the Big Typhoon. Of course, that's a lot of work. You've done more than enough already, so thank you.

It sure does look like the Arctic Cooling Freezer 64 Pro made an excellent showing, perhaps beating out the Zalman 9500. Hadn't heard of the Noctua's before. Too bad they aren't available for purchase at the online stores I browse.

 

[jmke]

Noctua is also known as Coolink, those brands seem to carry the same products.
I've checked the Ninja and XP-90(c) and XP-120 earlier (check this page for all heatsinks I've tested: http://www.madshrimps.be/gotoartik.php?articID=389))
I've got Thermalright HR-01 & SI-120 incoming (confirmed) and awaiting approval for Thermaltake Big Typhoon

about the graph: do you mean to split things up according to noise levels?

the chart is not that hard to read actually, the worst HSF combo's are at the top (near the Stock A64) from there on, everything downwards is better; and as you'll see, those heatsinks with silent fans which perform good are near the bottom.

 

[blah8899]

Just wondering what "stock" AMD heatsink you used? Is this the new heatpipe version people have been talking about or the old one AMD used to include? Judging by the fact that your Athlon runs at 1.5V stock, I'm guessing it's the old one?

The system I'm planning to build includes: Sonata II case, AMD X2 3800, Artic Cooling Freezer 64 Pro, evga 7800gt oc. What do you suppose the noise/heat levels for this setup would be?

Also, I recall in your Sonata II review that you got better temps w/o the duct system and w/o an intake fan. But in the heatsink roundup, an intake fan was used?

Lastly, some people have stated that Artic Silver 5 performs better than the MX-1 paste. Your thoughts?

BTW, great review and personally I didn't find the graphs all that hard to decipher. Keep up the good work.

 

[jmke]

Hi there Blah, thanks for the comments; I'm using the vanilla A64 alu HSF from AMD which came with 3200+, and also comes with 3000+ and Opteron 1.8Ghz S939

does AMD have a stock cooler with heatpipes? link?
that would be cool

Originally posted by: jmke
Got around to finishing a roundup with these coolers:

# AMD Stock Cooling (from 3200+)

AS5 vs MX-1: the MX-1 is designed for the base of their Freezer products, if you LAP the base of the freezer to a mirror finish, than AS5 will perform better than MX1, otherwise MX1 will take the lead. But when I say better & lead.. we're talking about 1-2°C difference.. at most. and I would not be surprised if it were <1°C.

Sonata II: yes the duct seems to be counterproductive, in my testing the duct worsened performance and a fellow reviewer in the US made the same conclusion;

ABout the intake fan: I did not do extensive testing with different fan configurations, so I can not comment on that any more than the findings in the review done by "Lazyman". His CPU temperature dropped 2°C when removing the intake fan, but HDD temp went up. In my tests it seems that the in-take fan does not do much but disrupt airflow inside the case and not actually work to a benefit for CPU cooling; but it helps PWM and System cooling as well as HDD. so it's not useless.

About your system, the 7800GT OC and PSU will most likely be the sources of noise; the Freezer 64 Pro can easily tame the X2 3800+, one of our forum members used X2 4400+ with the Freezer Pro @ 1400rpm in a case with 120mm PSU and 120mm outtake: 44°C
http://www.madshrimps.be/forums/showthread.php?postid=114978#post114978

Getting a 3rd party VGA cooling and maybe a larger "fanned" PSU will help you most, if you decide that the noise coming from your PC is "too much"

/me personally.. I'm not really bothered by noise that much, I swapped an Athlon XP stock cooling with a GlobalWin FOP38 (Delta 60mm on there) easily producing 50-60dBA noise
that was on my first custom build PC (Abit KT7 + AMD Tbird900).. since then I have been choosing more silent-wise solutions. Currently at 44-45dBA @ 50cm with my system.

It's noticable in the room, but when I play music/dvd/game I can't hear the fans.. which is good enough for me

 

[blah8899]

Thanks for all the info jmke.

Here's a link for the AMD heatsink with heatpipes:
AMD Heatpipe HSF
Seems like a pretty good stock cooler. Made me question the purchase of a Freezer Pro 64.

As for the FOP38 with Delta... You = psycho!! ... or possibly deaf

I had a Thermaltake Volcano 6Cu with a 4550rpm fan rated at 31dBA and I found that noisy. Granted I had a crappy generic case with holes punchered in the side panels.

It's too bad my order already went through, as I would change the case/PSU and add a VGA cooler. Any recommendations?

 

[jmke]

only have experience with Arctic Cooling's Silencer's, and they are not too bad, good value for money.

about the AS5 vs MX1, found somebody who tested it in Germany http://www.madshrimps.be/forums/showthread.php?s=&threadid=20474


thanks for the Heatpipe AMD HSF linkie, looks interesting

 

[jpinard]

I think what a lot of poeple say when they state "I don't want to read it all" is that it begins to get overwhelming.

So, when I look at all the data... as somone ewho wants to overclock "just a little" I'm looking for the best stock cooling solution. Not messing with fan speeds nor sorting through a bunch of different fans... just...plus it in and go... or plus in the HS and add that fan... and go... What performs best, then how much noise does it generate?

In this respect a "conclusion" would be excpetionally helpful. What stock solution was best? (and when I say stock, I mean "out of box"). What was best when you factor in noise?
Something I'm concerned about (with the side-mounted fans), is how does that affect motherboard overclocking? One of the primary benefits of the P-120 is the system-wide cooling it produces.

So for all the peoeple like myself and are techies... but not in search of high overclocsk nor the perfectly silent pc... well we're a bit overwhelmed. By the way, can you asnewer the questions I bolded up above?

And thanks a million for all your hard work!

 

[jpinard]

A couple more questions...

1. What's PWM stand for?

2. Why do people in this thread keep saying the Arctic Colling Silencer is best? From what I see, isn't the Scythe Shogun + Papst best? Ack, so confusing...

3. Do you do a video cooling comparison too?

 

[jmke]

Originally posted by: jpinard
I think what a lot of poeple say when they state "I don't want to read it all" is that it begins to get overwhelming.

well.. if you want to make an informed purchase.. some research is required; if you can't be bothered to read everything.. then you are on a way one trip to a bad(der) purchase


to answer your question
1) it's on the first page (near bottom)
2) both are very good, but with AC you get higher PWM values
3) not an updated one; but one coming along, check this section of the site: http://www.madshrimps.be/?action=articles topic "cooling other"

and the PWM also answers your MOBO OC question BTW

 

[Diasper]

I think if it's 10 or under variables on a scatter graph it should be fine. The other graphs you linked are just too crazy. What the other guy should have done is perhaps make bar charts of all of them at various noise levels and then afterwards perhaps made a scatter graph of a few select heatsinks - say the top 5 combined with the stock heatsink? Regarding using a scatter graph I think if the actual lines could be labelled more closely eg the name of the heatsink (or abbreviation but enough so its clear) just next to the line say towards the higher dB end would make it much clearer.

On the Vapochill I was saying the mounting method for DFI should be used on *all* boards to provide much better contact. If the Vapochill can be rotated without giving it a proper push it indicates it would benefit from the mounting method used with DFI boards as it'll improve contact and temperatures.

 

[jmke]

but then that mounting will deviate from standard mounting, as that method is not documented in the manual, thus rendering the results not quite usable. What could be done is a comparison between the 2 different mounting methods (normal and the altered one) and mention the difference; but for reference and comparison use the normal mount.

 

[Diasper]

Guess so - taking two would be best. Though anyone reading the review would then given good results use the same mechanism and so would still be as relevant for them.. . Ok I'll stop being a pain now : )

But, back onto the main topic, if you could find a clearer more easy way of representing data that would be great and hopefully good for your site as you should get more visitors. I have more ideas I guess I could say but are perhaps less relevant. Anyway all the best with it.

 

[jmke]

In my latest article with dBA/CPU graph I added these comments
http://www.madshrimps.be/?action=getarticle&number=4&artpage=1566&articID=391

"How to read the charts below: We have had some inquiries about how we displayed our temperature and noise results using the charts in Excel, we put dBA and °C on the same bar to improve the easiness of reference between noise and performance. Never would we consider adding °C to dBA to obtain a valid number which we use to deduct overall performance from. However our aim is to give some kind of relation between the two and we color code them differently so you can easily make out what FAN is making what noise, and at the same time, know how much air it is pushing."

a scatter chart I tried, but you end up with 1001 different colors for each heatsink, I know enough people who have trouble with colors (included /me ) so that won't make it easier IMHO
suggestions are always welcome; I have some professional and personal experience with Excel, but the charts generation in that application remains quite basic and not very customisable with my not so extensive working knowledge

 

[Diasper]

Scatter would never work - only scatter with lines would I guess unless each heatsink is limited to a single result - 10 dots on a charts with labels next to the point would probably be workable if you ever felt like taking it in that direction.