Sign in with
Sign up | Sign in
Your question

VGA Heat-Pipe Cooler Roundup 2006

Tags:
  • Memory
  • Heat
  • Cooling
  • Round-Up
  • VGA
Last response: in Memory
Share
August 30, 2006 10:50:32 AM

A new generation of aftermarket heat-pipe-equipped VGA coolers has entered the stage. Their mission? To take the heat off the fastest, hottest GPUs on the planet.

More about : vga heat pipe cooler roundup 2006

August 30, 2006 1:49:23 PM

good testing, hot results, to be expected with a hot card. What was the case cooling config used?
August 30, 2006 1:50:51 PM

Pretty informative review. Nice to see a sound level test in there as well. Only thing I would have liked added would have been some max overclocks with the different solutions to see if the effect of lower temps and increased O/C. And to see if the solutions for memory allowed a better mem O/C. :) 
Related resources
August 30, 2006 2:16:32 PM

just one thing, test was only done on one(1) card last time I looked there were at least three or four,maybe more,after all I might only have access to one or two cardz but,THG should be capable of testing more,other than that it was a fairly good review,keeping in mind that frostytech,does nothing but cooling reviews.............................
August 30, 2006 2:26:25 PM

After reading through the results, I suddenly developed a new appreciation for the water cooling on my Toxic card. The water cooling is very quiet, (usually can't hear it at all) and keeps the card between 41c at idle and 55c under load.
August 30, 2006 2:40:50 PM

I was surprised by the poor performance of the VF900; I've seen other reviews that speak very highly of it. However, it is the most quiet, and relatively affordable.
August 30, 2006 2:44:21 PM

Quote:
A new generation of aftermarket heat-pipe-equipped VGA coolers has entered the stage. Their mission? To take the heat off the fastest, hottest GPUs on the planet.


It would be great if data from the 1950 could be added into these charts. Sure, it's a different card with lower current demands, but would be useful for comparison.
August 30, 2006 3:03:26 PM

Quote:
also if he thinks one stock cooler is bad try listening to 2 of them at startup. its like living next to an airport :p 


Indeed. Rockets to power, engines to thrust! I think it's a proper psyche up.
August 30, 2006 3:15:55 PM

Quote:
nice review although i wouldn't mind knowing what ambient temp that room he was in was at.

my room may be cool but right now my x1900xt is sitting at 47C. even on hot days it usually sits at 50-52C.

also if he thinks one stock cooler is bad try listening to 2 of them at startup. its like living next to an airport :p 


Woah, I'd hate to be you when your babies are at their max load. 8O

I would probably think that the room wasn't hot or AC'ed, probably at normal room temp. Its true tho' it all depends on ambient temp. In the NYC Hot&Humid days of the summer playing games can be somewhat annoying when VGA & CPU fans kick into high gear. I guess thats why the article mentioned a different cooler depending on your particular need.
August 30, 2006 3:26:58 PM

Quote:
I was surprised by the poor performance of the VF900; I've seen other reviews that speak very highly of it. However, it is the most quiet, and relatively affordable.


Yea, and not a lot of diffrence between the min and max speed as well. But it's the best performer in idle temperature though.
a b } Memory
August 30, 2006 4:33:01 PM

Was it in a case for the tests?

Would be interesting to see how much heat would build up in a 1hr+ test...

Might also be interesting to see the Thermalright V1 Ultra with the fan facing up?

How loud was the stock cooler with it's auto setting under load?

Damn that cards hot....and to think i have one sitting in the box(will be good in the winter)....just waiting for my e6600 :( 

Good review....
August 30, 2006 4:44:56 PM

The way I read it, it sounded like the biggest difference between the stock cooler and any of the aftermarket ones as far as case temp goes was 4C (presumably the delta between case and ambient for each cooler was found, and when compared against the stock, none was more than 4C higher than stock). Then again, didn't mention how long they ran the tests for...

If you wanted, I suppose you could find the thermal conductivity and then just figure out what these temps should be based on another card's GPU's wattage. If we had case temps we could even adjust it for the average temperature of our individual cases. It should be accurate to within a few degrees...
August 30, 2006 5:13:05 PM

silentpcreview gives a pretty bad review for Kuformula. I thinks that's probably they installed it wrong. Myself have pretty good result on that.

I also have a slot fan under that vga card. IT helps blow the air out of case, very effective in keeping your case cool.
August 30, 2006 5:18:49 PM

Quote:
I also have a slot fan under that vga card. IT helps blow the air out of case, very effective in keeping your case cool.


I've heard people dis on slot fans, but have had good luck myself. I've run them above 1900XTs after flipping the fan shroud upside down. Thus the fan takes hot air off the back (top) of the graphics card and exhausts it. There is a surprising amount of heat coming off the back of the GPU - enough to lower my case temps 1-2C with the slot fan as described.
August 30, 2006 5:22:54 PM

Quote:
one thing i wouldn't mind seeing is how the new style cooler as seen on the x1950xt would do. it looks slightly different than the stock 1900 one.

does anyone know how it does. i'm interested for no other reason than curiosity.


I've seen reviews that say it is significantly quieter but I don't remember any temps discussion.
August 30, 2006 5:30:33 PM

Quote:
ye i heard due to it having less fins on the fan it is quieter but i think it also has a fully copper heatsink for the memory and has the fan blowing across the new memory.

obviuously when this review was done they probably couldn't get their hands on one. i am guessing this was the case as the cats used was 6.5's


I look forward to the release of Vista and DX-10. As the flocks make their mad scramble, I'll be looking for good sales on 1950s to upgrade my lamest current box. Imagine how inexpensive excellent PCs will be in a couple of years...
August 30, 2006 5:55:45 PM

Thanks for the comments guys. A couple answers:

- The temps were taken in a Gigabyte 'Aurora' case. Very nice case with good cooling

- Ambient temperature was 25 degrees Celcius

- Max temps were after 10 minutes on Atitool's stress tester. Atitools's stress tester is REALLY tough on the GPU, this is probably quite a bit hotter than you'd see it after hours of real gaming or 3dMark loops.
August 30, 2006 6:26:33 PM

I'm surprized that Silenx's big cooler didn't make it into the review.

Apparently it is very quiet - quieter than the others, so you can run it in high-speed mode without hardly any noise.

http://www.silenx.com/ixtremaproheatsinks.asp?sku=ixg-8...
it doesn't look special - and has less copper than the Zalman, but 18db at maximum speed is crazy low noise. 30cfm should do an adequate job and well, the exhaust fan on your case makes more noise. This should add a whopping 1-2db at full speed to the average user's setup.
August 30, 2006 6:30:41 PM

I don't know anything about the other coolers, but I have used the Zalman VF-900Cu on both a PNY 6800GT Ultra and no on my EVGA 7900GT KO and I can testify that the VF-900Cu does reduce the loaded temps on both gpu's. For example, the 6800GT Ultra was at 79C under load then dropped to 66C with the VF-900Cu. The 7900GT was 72C under load and now tops out at 66C under load. Idle temps have stayed the same on both GPU's but the VF-900Cu with its heat pipes really shines when you OC and put the GPU under load.
August 30, 2006 6:37:08 PM

Quote:
is it just me or is that one missing a key component needed that the article title might hint at


Only one?

No nvidia.

No water (I know - title is heat pipe...)

No max OC data on each.

I could go on.
August 30, 2006 6:54:17 PM

Quote:
Only one?

StrangeStranger was referring to the Silenx cooler, which doesn't appear to use a heat pipe...

Quote:
No nvidia.

I used the hottest card I had. Isn't the X1900 XTX hotter than the 7900 GTX? Plus, the review is for coolers, not video cards... the only use the video card for is to provide a 'control' so that the data we gather from the coolers is standardized.

Quote:
No water (I know - title is heat pipe...)

You answered your own complaint. :p 

Quote:
No max OC data on each.

Holy crap, that would have taken forever. I would have liked to have had the time to do that, but, a few degrees here or there isn't going to make a colossal difference in OC ability. More realistically, if I'd have had time I would have tested with the best and worst solution. Like I said, didn't have the time tho.

Quote:
I could go on.

Please do! :twisted:
August 30, 2006 7:01:33 PM

Quote:
the VF-900Cu with its heat pipes really shines when you OC and put the GPU under load.


That wasn't the case in our tests, but I suspect milage may vary depending on case airflow.

The one real regret I have is that if I'd have had more time and cases I would have tried benched all of these coolers in a different case - maybe one with a side fan - to see if the results varied.
August 30, 2006 7:16:16 PM

Quote:
Only one?

StrangeStranger was referring to the Silenx cooler, which doesn't appear to use a heat pipe...

That's fine, neither does stock.

Quote:
No nvidia.

Quote:
I used the hottest card I had. Isn't the X1900 XTX hotter than the 7900 GTX? Plus, the review is for coolers, not video cards... the only use the video card for is to provide a 'control' so that the data we gather from the coolers is standardized.


Well, believe me I'm not crying for an nvidia cooler review, just thinking of possible angles SS might have been taking.

Quote:
No water (I know - title is heat pipe...)

Quote:
You answered your own complaint. :p 


I wasn't complaining. The closest thing to a complaint I made was a request for 1950 numbers.

Quote:
No max OC data on each.

Quote:
[Holy crap, that would have taken forever. I would have liked to have had the time to do that, but, a few degrees here or there isn't going to make a colossal difference in OC ability. More realistically, if I'd have had time I would have tested with the best and worst solution. Like I said, didn't have the time tho.


Like I said, not a complaint, just an angle. I don't know what kind of OC difference there might have been. Plus we know how you like to stay busy.

Quote:
I could go on.

Quote:
Please do! :twisted:


Surely you have a bundle of DX-10 cards you're previewing... Yes, I was kidding. One thing I have piddled with is varying ambient temps then monitoring CPU and GPU temps. It's not a simple linear relationship and it does affect my CPU OC at the high end (30C ambient). So I'm thinking that inside exhaust actually can have an impact if you are already pushing the CPU or GPU OC hard. That's not a question or request, but I guess it would be interesting to see data (case temps, with and without door fan, inside vs. outside exhaust, etc.) I am data-centric and just like to see all relevent numbers.
August 30, 2006 7:54:02 PM

Quote:

Surely you have a bundle of DX-10 cards you're previewing...


I could tell you but then I'd have to kill you. :p 

Quote:
One thing I have piddled with is varying ambient temps then monitoring CPU and GPU temps. It's not a simple linear relationship and it does affect my CPU OC at the high end (30C ambient). So I'm thinking that inside exhaust actually can have an impact if you are already pushing the CPU or GPU OC hard. That's not a question or request, but I guess it would be interesting to see data (case temps, with and without door fan, inside vs. outside exhaust, etc.) I am data-centric and just like to see all relevent numbers.


Yeah, more data would have been nice. Once again I'll cowardly hide behind the time constraints. I did take the case temps with the stock cooler (that pushes air out of the case) compared tot he most efficient in-case solution (the Kuformula at max) which should have theoretically transferred the most heat into the case.. and got a small 4 degree difference, which didn't inspire me to dig further.
But yeah, would have been nicer to collect all that data and show it on a graph. Next time...
August 30, 2006 8:14:27 PM

I know you said that you didn't have the time, but testing the max OC of each cooler would have been a welcome addition to the article. Maybe next time you could take the top 2 and OC them and compare it to a max OC with the stock cooler.
a b } Memory
August 30, 2006 10:25:16 PM

Quote:
a few degrees here or there isn't going to make a colossal difference in OC ability.


That being the case, I would find it hard to spend the money on an upgraded cooler. The stock cooler, while noisy, cooled to within one degree of the best cooler with the fan at 100%. For some, the noise is worth it, and we can rest assure that by cranking the fan up, we already have one of the best coolers for this card. Hats off to ATI for including such a beast.
August 31, 2006 12:24:03 AM

Thanks! Someone finally included the Evercool in a comparison. I've been wondering how my heatsink compares to the expensive ones, and I'm glad it's in the ballpark.

I removed the plastic cover and replaced the fan with a nearly silent Arctic Cooling 80mm to lower noise and it ended up lowering temps even more. I'm very happy with the results.
August 31, 2006 2:08:07 AM

What a difference a day makes.

Back in April Tom's said the KuFormula VF1 Plus was a "Must have" and now it's no better than tied for 3rd in a five cooler roundup.

I guess this means I'll have to frequently look in on Tom's to stay current. :wink:
a b } Memory
August 31, 2006 5:55:57 AM

Quote:
I also have a slot fan under that vga card. IT helps blow the air out of case, very effective in keeping your case cool.


I've heard people dis on slot fans, but have had good luck myself. I've run them above 1900XTs after flipping the fan shroud upside down. Thus the fan takes hot air off the back (top) of the graphics card and exhausts it. There is a surprising amount of heat coming off the back of the GPU - enough to lower my case temps 1-2C with the slot fan as described.

used to have one run at 5 volts over my 9800pro..... boy did that take allot of heat away....
Also under my GF4ti 4200...

Slot fans are good....as long as you keep them clean....

Again....great review....more like this in the future....
August 31, 2006 7:07:27 AM

While the Silenx doesn't have a heat-pipe, I maintain that it really doesn't NEED one to provide decent cooling. The only reason you need a heat-pipe is if you are moving the heat to another location. Like the other side of a card or away far enough to run a big fan(CPUs come to mind).

It would be interesting to see how it compares to the Zalman, as it appears to be a clone of the slightly smaller than what was tested Zalman VF700-Cu.

I know this is far superior to any single-slot stock cooler(X1800, GF6800, etc). It's tons quieter and more efficient that the stock one on my X850XT was. I barely hit 45C at full load, and idle is pretty much under the scale/won't register on the Catalyst utility. For older, less efficiently cooled cards, aftermarket solutions are a godsend.

I'd love to see what slapping a big Zalman on a GF6800 would do. I
August 31, 2006 2:31:36 PM

Quote:
What a difference a day makes.

Back in April Tom's said the KuFormula VF1 Plus was a "Must have" and now it's no better than tied for 3rd in a five cooler roundup.

I guess this means I'll have to frequently look in on Tom's to stay current. :wink:



Did you look at the numbers though? At the medium fan setting, the KuFormula had by FAR the best noise-to-performance ratio in the whole test. At the high setting, it offered the best performance, and the noise was still acceptable.

The stock solution cooled well at 100%, but it was so loud I'd never use it in a million years at that setting.

It lost points for the fan bracket being a size that didn't fit stock, but I'll tell you, the Kuformula is the cooler that I kept in the system after the testing was done... if you're OK with some hardware modification, it's a great cooler.
August 31, 2006 5:34:08 PM

Quote:
I know this is far superior to any single-slot stock cooler(X1800, GF6800, etc). It's tons quieter and more efficient that the stock one on my X850XT was. I barely hit 45C at full load, and idle is pretty much under the scale/won't register on the Catalyst utility. For older, less efficiently cooled cards, aftermarket solutions are a godsend.

I'd love to see what slapping a big Zalman on a GF6800 would do. I


I also have an X850XT (Sapphire) and have an Arctic Cooling NV Silencer 5 and it works really well (also vents out the back too). I do have a Zalman CPU HSF though, the 7700cu (its massive!) and it works well too. The original cooler on my 850 worked alright, but it was ridiculously loud. The Arctic lowered my temps, allowing much more overclocking and is much less loud.
August 31, 2006 8:01:47 PM

Quote:
What a difference a day makes.

Back in April Tom's said the KuFormula VF1 Plus was a "Must have" and now it's no better than tied for 3rd in a five cooler roundup.

I guess this means I'll have to frequently look in on Tom's to stay current. :wink:



Did you look at the numbers though? At the medium fan setting, the KuFormula had by FAR the best noise-to-performance ratio in the whole test. At the high setting, it offered the best performance, and the noise was still acceptable.

The stock solution cooled well at 100%, but it was so loud I'd never use it in a million years at that setting.

It lost points for the fan bracket being a size that didn't fit stock, but I'll tell you, the Kuformula is the cooler that I kept in the system after the testing was done... if you're OK with some hardware modification, it's a great cooler.


Yes, I read the entire article, text and numbers. My comment about the VF1's 3rd place finish was based on the article's Conclusion where the Turbo2 and Accelero X2 were given 4's and the VF1 was given 3 1/2. This was Tom's rating not mine. I can see the VF1's performance. Perhaps the 1 1/2 point deduction because the VF1 is a "pain to install" was too severe.
August 31, 2006 8:47:45 PM

I noticed a mistake in graphs where it is written Thermaltake V1 instead of Thermalright V1 :wink:
September 1, 2006 4:31:52 PM

Thanks, we actually caught that yesterday. Still waiting for the fix to be posted.
September 1, 2006 6:18:21 PM

I´ve got the Accelero X1 cooler on a 7800GT SLi mode (X1 and X2 design are the same). Sure the cooler does the job and is silent too, but there is a catch: Because of the design and depending on the placement of the graphic card, the hot air will be blowed down either on the motherboard itself, or to the back of the motherboard.
The result is that temp on my graphic card dropped with 8 degrees C, but the temp on my motherboard raised with 5 degrees C. Keep that in mind before a buy.
I ended up buying a NV silencer 5 (rev. 3) for my #2 graphic card (both are from arctic cooling) cause this one blows the hot air out of the cabinet.
September 4, 2006 7:59:02 PM

Hi there,

I'm kind of having a problem with my X1900XTX using the Zalman VF900CU
In idle mode it cools the card down to about 46°C but after running the first 3dMark 06 demo (the long proxycon version), riva tuner shows an increase to over 100°C. The stock cooler kept it at around 92°C I think - the VF900CU is running at max speed btw.
Serious Sam 2 pushes it to around 80°C and Call of Duty 2 to around 90°C
Did I do something wrong?
With the Stock cooler, there was some kind of metal... thingie screwed on for whatever purpose... Is that supposed to stay there? I didn't screw it back on with the VF900CU.

How is it possible to test the temperatures with ATI Tool anyway? The card only switches to the higher clock settings when full screen 3d is used...
September 5, 2006 5:31:39 PM

If you run 3d simulation thing that comes with ATI Tool it will run your card at full speed. It shows the temp while doing this so you should be able to see your max temps.
September 5, 2006 7:25:10 PM

Quote:
How is it possible to test the temperatures with ATI Tool anyway? The card only switches to the higher clock settings when full screen 3d is used...


It'll switch to higher settiongs when a windowed 3d mode is used as well.
September 5, 2006 7:54:17 PM

No it doesn't... according to the Riva Tuner diagrams it stays at 500/600mhz and only heats up until around 74°C after 10mins of ATI Tool artifact scanning. But when minimizing or closing a full screen game, I can clearly see the change between 2d and 3d mode. Is this a problem with the new 6.8 driver or with Riva Tuner.

BTW: before installing the ATI Control Center, Riva Tuner also didn't show any clock increases in fullscreen although the 3dMark 06 result was the same as far as I can remember...
September 5, 2006 8:40:49 PM

You might like to edit the link. (New York may have been named twice, but .com wasn't.)
Synergy6
September 5, 2006 11:09:38 PM

Quote:
No it doesn't... according to the Riva Tuner diagrams it stays at 500/600mhz and only heats up until around 74°C after 10mins of ATI Tool artifact scanning. But when minimizing or closing a full screen game, I can clearly see the change between 2d and 3d mode. Is this a problem with the new 6.8 driver or with Riva Tuner.

BTW: before installing the ATI Control Center, Riva Tuner also didn't show any clock increases in fullscreen although the 3dMark 06 result was the same as far as I can remember...


Hmm thats odd. I have always done my overclock stability testing using ATI Tool 3d view in windowed mode. It usually only takes about 10 minutes or so to get to the max temp, or it will starting artifacting / reboot my pc. I only have a lowly x850xt though which doesn't get nearly that hot though.
September 6, 2006 9:16:58 AM

Hiho,

all in all it's a nice test but I miss loads of information about the testing-conditions, specially about how you measured the loudness of the coolers. You should definitely add these information aswell as the ambient-temperatures to the test. Unless knowing how they were meassured the mentioned values are more or less worthless.

Another important point is why you didn't use the provided heat-conductive paste? Some manufacturers provide very well heat-conductive paste with their coolers while others don't give away any at all. Imho that's a verry important point.

Greets eldoreo.
September 6, 2006 3:28:25 PM

Quote:
Unless knowing how they were meassured the mentioned values are more or less worthless.


Uhm, not really. The only reason knowing how they were measued *would* be useful is if you were getting the exact same setup. If not, the conditions don't really matter, all that matters is that they were all measured in the same way, and so are comparable.
Synergy6
September 6, 2006 4:06:38 PM

Quote:
No it doesn't... according to the Riva Tuner diagrams it stays at 500/600mhz and only heats up until around 74°C after 10mins of ATI Tool artifact scanning. But when minimizing or closing a full screen game, I can clearly see the change between 2d and 3d mode. Is this a problem with the new 6.8 driver or with Riva Tuner.


Didn't use Riva Tuner, used Atitool. Seemed to work for me...
September 6, 2006 4:12:31 PM

Quote:
Hiho,
all in all it's a nice test but I miss loads of information about the testing-conditions, specially about how you measured the loudness of the coolers.


Loudness was measured with a microphone and software Db meter.
Wasn't the most accurate system, but it did give results that agreed with with my experience.


Quote:
Unless knowing how they were meassured the mentioned values are more or less worthless.


I disagree with you there. The comparison is cooler-to-cooler, so as long as the test system remains identical - as a 'control' - then the results are very valid for comparing cooling performance between coolers.

Certainly you might get different results on your rig, but cooler-to-cooler comparison on your own rig should remain very similar if not identical.


Quote:
Another important point is why you didn't use the provided heat-conductive paste? Some manufacturers provide very well heat-conductive paste with their coolers while others don't give away any at all. Imho that's a verry important point.


Good point, I had considered that.
In the end I decided to benchmark the coolers, not the thermal paste the manufacturers included... once again, part of the 'control' to ensure that the cooler-to-cooler results were valid.
September 6, 2006 7:10:25 PM

Ah now the 3d change works with the new ATI Tool 0.25 beta 15. But the 3d view doesn't need much time to heat up the X1900XTX to over 100°C. I don't know how much more it will heat up because I'm aborting the tests there.
Any suggestions?
September 6, 2006 8:59:51 PM

I would avoid overclocking it that much if it is getting to 100°C that quickly. I know that the cards specs state that it can handle that kind of heat, but i'm sure its lowering the lifespan of the card significantly at that point.
September 7, 2006 7:30:57 AM

Well - you used a Software Db meter? Should I start laughing or crying right now? That's that precise as making stars observatory throuhg a digicam. Years ago there came a messure-software with my gyrotwister to detect the rpm, think that's hardly the same. ;) 

That's the thing why I meant the test is worthless unless you don't give precise information about how you made it. Maybe you've mounted fans somewhere which influence the results heavy and you don't even know about it. It's just to get a higher trancparence and to be able to value the results.

At the last point I can slightly agree with you, but I still think it would be better to test those who came along with preapplied or delivered thermal compound with this one and others which weren't delivered with a normal one like a silmore.

Another nice addition would have been to see what benefit you get from the lower temperature e.g. through proving it with an oc-test.

Non the less keep up the work

greets eldoreo
September 7, 2006 4:31:58 PM

Quote:
Well - you used a Software Db meter? Should I start laughing or crying right now?


Heheh. Well, I don't think it's all that bad.

I could have done what some reviewers do, and simply give my impressions about relative sound...

But the mic produced a quantifiable number, and like I said, it totally synced up with my experience... and I did describe the sound experience with he loudest and quietest cards as well, to give the reader an idea of what it's like.

Is it perfect? Of course not. But just because it's not the most accurate measurement in the world doesn't make it useless or invalid. It's a very helpful benchmark if you want an idea of how loud the coolers are compared to each other, even if it isn't accurate to the nearest 0.01 dB.
      • 1 / 2
      • 2
      • Newest
!