AM2 Heatsink Roundup

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,701
AM2 Heatsink Roundup - We bring together 7 CPU coolers on AMD’s new AM2 platform and put them to the test to find out how they stack up against each other.

The other coolers reviewed here did not stand out overall when compared to these four. The either had glaring flaws or were just out performed for the same price. Surprisingly, out of all of these products, none were bad products, simply some shined over others when it comes down to doing the job of cooling. As noted by the tightly graphed temperatures above, air cooling is about more than cooling now days.
Click to expand...
 
Nice Article, but I have a question:

Do all the AM2's have the heatpipe cooler, or only certain higher-end processors?

Thank you for another great article. I'm looking forward to more in the future.
 
Good article, however these are all mid-range aircooling heatsinks(except for the Zalman), it would have been nice to have some heatsinks from the upper echelon of aircooling, like the Tuniq Tower, Scythe Ninja/Mine/Infinity, Thermalright Ultra-120, etc. Maybe for a future article?

Aside from that, the evaluation did a good job of simplifying the standout aspects of each heatsinks.
 
You guys should have waited for Thermalright to release the AM2 clip for the Ultra-120 (or done a mod job on another clip like I did so you could test it). I put a small bolt in the center of the Zalman HSF clip and then used it to install the U-120...

I have the CNPS-9500 AM2 and the Ultra-120, and the Ultra-120 is giving me even lower temps than the Zalman (by about 3 degrees C). Combine that with its easier than the XP-90 installation, the ability to use any 120mm fan, etc...

Its the easy winner. The clip should have been released sometime this month (July), but even w/o it, its easy to adjust another one to fit.
 
aren't the scythe mime AM2 compatible? @ 34.99 shipped at newegg

They perform way better than samurai Z. and are quiet

The heatsink round up seemed out of sync...
 
Good article. I just use the AMD 4-pipe from a socket 939 rig. It gets the job done.
 
The stock coolers coming from AMD are amazingly great, and i almost always upgrade to an XP-90 in all my rigs, easiest for me to install, people don't get out of whack when i offer that to them as a more quiet and cooler solution (if they ask for one), fan is painless to obtain, and by simple design it happens to perform around the best

you just cannot beat that
 
The XP-90 hasn't been sufficiently able to cool anything in a long time. The best air coolers weren't even included - TT Big Typhoon and Scythe Ninja. Even the Arctic Cooling Freezer Pro kicked the shit out of my XP-90, the Big Typhoon was 1 degree lower at load than the Freezer for me. I don't understand people loving the XP-90, it's just not even a competitive cooler anymore. I don't think the review was bad I don't agree with the results having done my own comparison testing.

Using multiple mounts and AS5 TIM I found load temps with an opty 148 at 3.0 (1.44v) to be the following:
Big Typhoon idle 28, load 38
Arctic Cooler Freezer Pro idle 30, load 39
Ninja with tt 120 mm fan idle 29, load 38
XP-90 with 60 cfm 92mm fan idle 32, load 48
Stock Cooler idle 31, load 54

As for the XP-90 - it wasn't even close to being a good performer next to the likes of the Ninja and the Big Typhoon.
 
I don't know how or why the XP-90 got the award, people are so stuck to that cooler like a following cult gang.
 
Ockie said:
I don't know how or why the XP-90 got the award, people are so stuck to that cooler like a following cult gang.
Click to expand...

Exactly - its not even a good performing cooler anymore. I dont get it.
 
Wow...memory lane. :eek: I remember when I used to get most of my hsf reviews from here but that was many years ago. Anyway, good mini-review and my Freezer 64 pro performed like I expected. It's still the best blend of price, performance, looks, low noise and easy install imo.
 
Zalman is still impressing me as a really great choice for cooling products. I hope they
continue to succeed. I will probably use a Zalman heatsink fan on my next rig while I gather
the funds necessary for watercooling.
 
Nice article, good for some opinions since there is so many options out there for heatsink/fans.

And did I really see a non-zeroed graph? :rolleyes:
 
Very informative.

I have an Arctic Pro64 on an FX-53 and it dropped the load temp 10 degrees from the stock. Very,very easy to install.

Would it have made a difference putting the selected set-ups inside a case, rather than leaving them outside surrounded by cool air???? I realize the case choice then factors in and you are distracted by any case fan noise. It just seems more real-world.

Great article. I'm going to be in the market soon.....this really helps alot.
 
magoo said:
Very informative.

I have an Arctic Pro64 on an FX-53 and it dropped the load temp 10 degrees from the stock. Very,very easy to install.

Would it have made a difference putting the selected set-ups inside a case, rather than leaving them outside surrounded by cool air???? I realize the case choice then factors in and you are distracted by any case fan noise. It just seems more real-world.

Great article. I'm going to be in the market soon.....this really helps alot.
Click to expand...
The reason why many reviewers don't do that is because of the staggering number of different case/fan configurations that are out there. With an open test bench anyone at home can replicate the numbers with the same setup and ambient temp.
 
Nice review, very informative about the installations. (Something I really look at, cuz damn I hate it when I need to switch out the stock bracket).

but, I would like to see if the thermal pastes that come with some of the HSFs (probably just the Zalman) would perform..

I mean, if I'm spending $65, I probably don't have the extra cash for arctic silver 5 or something. I'd like to know how the Zalman thermal grease does.
 
I fail to see why a physical fan speed controller is relevant. If you really want to be able to control the speed, get a motherboard that allows for it. Do people actually prefer to change fan speed manually, rather than allowing the computer to ramp up the fan as needed? Also, I just installed my AC Freezer 64 Pro on my AM2 motherboard, and at full speed, it is not quiet at all.
 
Mazgazine1 said:
Nice review, very informative about the installations. (Something I really look at, cuz damn I hate it when I need to switch out the stock bracket).

but, I would like to see if the thermal pastes that come with some of the HSFs (probably just the Zalman) would perform..

I mean, if I'm spending $65, I probably don't have the extra cash for arctic silver 5 or something. I'd like to know how the Zalman thermal grease does.
Click to expand...
The stuff that comes with heatsinks is just average, however Zalman's new STG-1 that brushes on like nail poilsh is starting to cause a buzz. I use nothing else now. :cool:
 
Yaden said:
Nice article, good for some opinions since there is so many options out there for heatsink/fans.

And did I really see a non-zeroed graph? :rolleyes:
Click to expand...

For the information presented and the amount of deviation in the results starting the graph at
room temperature was the best choice. 0 degrees Celsius is not technically zero anyway in
terms of temperature. Maybe the graph should have started at 0 degrees Kelvin?
 
i got a question - the stock AMD cooler, was the fan shorted to run at 5000rpm? since its most definately the budget HSF there since it comes with retail cpus it would have been nice to show what it can do at that speed. i remember reading a review on madshrimps where with the fan running at that speed it kept up with the XP-90.

and why the hell do so many people use that piece of crap?? its hardly worth the money if you have the stock amd heatpipe hsf. as this review showed, theres much better out there!!

as for me i'm stickin with my stock heatpipe hsf and replacing the fan with a 120mm using one of the adapters off ebay. beat that for noise/performance!
 
My big question for these and any other heatpipe coolers: Why are they designed for a horizontal motherboard? When the heatpipes lay horizontal (or even slope downward), how much effectiveness do they lose?

In my own personal use, I have never seen a heatpipe cooler achieve the same results for me as in testing unless I have a horizontal motherboard instead of the vertical positioning most of us use.

I understand why testers use an open-air system, but could they test in the vertical rather than the horizontal?
 
mentok1982 said:
For the information presented and the amount of deviation in the results starting the graph at
room temperature was the best choice. 0 degrees Celsius is not technically zero anyway in
terms of temperature. Maybe the graph should have started at 0 degrees Kelvin?
Click to expand...
After some more thought, I agree. Starting the graph at ambient was the best option. I just assumed that the starting value for the graph was selected just to show the most variance on the graph.
 
Yaden said:
And did I really see a non-zeroed graph? :rolleyes:
Click to expand...

It was my decsion to see a non-zeroed graph. We were one of the first sites that pushed the importance of zero-scaled graphs many years ago, but when putting together this data, I got to thinking about it in terms of temp. If the chart was zeroed in degrees C that would represent 32F, and if it was zeroed in degrees F, obviously that would have been -18C. So depending what unit you are using, it would greatly vary the scale of the graph.

Our constant here was 24C, the ambient temperature we used and was noted on the graph. I felt as though temp scaling over ambient is more important in this case as I doubt many of us will be using a computer system at 0F or 0C.

If you think our logic on this is terrible flawed, I would like to hear exactly why. Please share. But yes, on the surface I can see where it might make you go WTF?
 
Good points on the graph. Now that you mention it and I go back to look I can see merit in both methods. The stock cooler was run without shorting it out. What some articles failed to mention when running the stock fan shorted is that while it may perform well you dont want to be in the same room as it due to the excessive high pitched noise @5000RPM. Thanks for being gentle my first time out guys. :D
 
Mazgazine1 said:
Nice review, very informative about the installations. (Something I really look at, cuz damn I hate it when I need to switch out the stock bracket).

but, I would like to see if the thermal pastes that come with some of the HSFs (probably just the Zalman) would perform..

I mean, if I'm spending $65, I probably don't have the extra cash for arctic silver 5 or something. I'd like to know how the Zalman thermal grease does.
Click to expand...

I imagine a thermal interface material roundup is in our future.
 
I had an XP-90C and then switched to the Zalman 9500AM2.

The Zalman definitely cools better but I guess it didn't help with the fact that I had my friend apply the white goop. Now Ive got AS5 in action with the Zalman.

Here are some pics I have of the Zalman. Better cooler than the XP-90C IMO. However as the review had put it, the Thermalright gives you the option of having whichever fan you want.

XP-90C in action with Panaflow fan (recommended)
cpu1.jpg


Zalman CNPS 9500 AM2 (cooling SKT939 A64 3700+ right now idle @ 39C)
IMG_0156.JPG


Bottom of Zalman CNPS 9500 AM2 (this came stock!)
IMG_0104.jpg


Once again great review!
 
As you can see in my sig, I'm not using the HSF that came with my processor. However, as wattage goes down what's included in the box is a great solution. The difference is only 3 degrees idle and 5 under load (compared to the best), is that worth $60+?
 
I am using a stock AMD HSF that came with my friends opteron, looks just like the four pipe shown here in the review, it seems to work better then my hyper6 from CM, and I was wondering what the heatpipe orintation is supposed to be, I did not recive the install instructions from my friend and I cant find any info on AMDs site, also a review of these HSFs in the vertical position that most people would use just to show if theres any differance with the heatpipes
 
Bbq said:
Do all the AM2's have the heatpipe cooler, or only certain higher-end processors?
Click to expand...
Back on Socket 939, only the CPUs above the X2 3800+ got the 4-heatpipe cooler, as far as I know.

Good to see Thermalright came out with their new AM2 bracket.
 
while I too once believed testing on an open testbed would give you the best reference results, I changed my mind after doing extensive test in/outside a case.

Heatsinks which performed excellent outside a case (like the XP-90) didn't do all to well in a closed environment.

Now I know it's impossible to test inside all cases with different layouts, but a case with one rear 120mm is not too uncommon; and that's what I choose to do my tests in. I've tested and compared over 100 heatsink up till now (list: http://www.madshrimps.be/gotoartik.php?articID=389 ) with many group comparisons, the tower coolers definitely take the lead in low airflow cases, while the "blow onto the motherboard" designs like the XP-90C can be extremelly good with a high performance fan and lots of case airflow.

However, with great airflow comes great noise and the enthusiast market is moving toward silent computing, more and more, with larger CPU fans, more plug and play Water Cooling kits, the end result is that NOISE generation has become a needed ingredient for any cooling device test/review, otherwise you are comparing apples to oranges.

XP-90C with 92mm Tornado fan will surely outperform Scythe Ninja with 700rpm 120mm fan, but I can assure you that not a lot of people will opt for the "coolest" HSF solution once they "hear" it.


my 2 cents.
 
I can't believe I'm posting this.. I'm such a nerd...
dR.Jester, those are some gorgeous pictures!

If you think about it for maybe two seconds it isn't difficult to realize that Zalman really does have the best design. Less metal and more surface area is what you want.
 
Kyle, I think your agrument is valid, guess I missed it in the article so don't mind me - non-zeroed graphs generally throw me off since other sites are pretty careless with them. :)
 
DryFire said:
Using your current setup you may end up with meaningless results given the uncertainty present processor's on die diode and the motherboard's onboard temperature monitoring. You'd have to do some modification to your current setup; I've seen on diode accuracy given as +/- 5 C to +/- 10 C. The current setup probably lacks the resolution needed as well.

Assuming the +/- 10 C is true, then the graphs for this write up basically tell you load temps are higher then idle ones, as all the other values are within +/- 10C. I'd prefer to see points with error bars then a bar graph.

Perhaps the error is systematic and as such all the values are shifted in one direction, but nothing was done to reduce the effect of random error (only 1 trial per heatsink), or if there were repeated trials it was not stated.
Click to expand...

The +/- is from actual temperature not how much the thermisistor will vary. We are comparing the heatsinks directly against each other using the same measuring instrument for all tests. While you cannot compare these temperatures to what someone else is getting with a different board/chip you can compare them to each other.
I did neglect to mention in the article that each heatsink was re-mounted 3 times to make sure the mount was good.
 
Isn't this point where we tear up the review and argue the merits of real world vs. synthetic benchmarking? :rolleyes:

Just kidding... :D Good read guy!
 
You must log in or register to reply here.
Back
Top