I had a similar problem with my Swiftech ram sinks for my 8800gt. The solution (given by swiftech when i called for help) worked like a charm. Essentially, IPA does not act as a degreasing agent. The reason that the sinks don't stick is because they can't. There is a residue of thermal grease on the chips. The way to remove this grease is to use a cleaner that removes grease. Acetone works wonders. When i used the acetone on my chips the ram sinks stuck very well. If this still doesn't work and you don't want the permanence of thermal epoxy there is another solution, also given by swiftech. You get a bottle of krazy glue and put a dab on a piece of paper. Take the pointy end of a paper clip and dip it in the glue and then dab the tip on the ram sink. This leaves enough glue on the sink to make the sink stick, but doesn't interfere with the thermal conductivity of the tape.
Somewhat haphazard, but my ram sinks are still going strong.
Very interesting, thanks for the reply! I plan on building a system soon and will make sure to have both on hand when removing paste and reapplying better paste. I wonder if this holds true for all paste jobs (be it CPU, GPU, Northbridge,etc.). Might get a bit extra performance or at least be sure you have a clean starting point.
Just be careful with Acetone, it's not like IPA that you can spill anywhere and will just quickly evaporate. Acetone is a strong solvent for most plastics and synthetic fibres. You don't want to see that headline, "Acetone ate my Mother....board".
I am loving this cooler. With my mobo I can wrap it around the back of the card without interfering with the cpu or northbridge tower coolers. It means I lose no expansion slots. It also puts the cooler right next to my case exhaust fan, so I don't need to mount a fan on the cooler.
It's overkill, but I have it mounted on a 7900gt. Idles at around 33, loads at 39 (again, no fan on the cooler). The 7900gt isn't really a speed demon anymore, but I do get a decent peformer running very cool with no extra noise. It's very ideal for me.
I am amazed at the results, truly amazed. Unfortunately this design has a crippling flaw that your test (as you mentioned in the article so I'm not faulting Anand) will not show. The hot air is still INSIDE the case!
As the GPU is notorious for being the largest producer of heat in a case (barring some crazy OC'd quad-core), I don't want the product of that energy to be dissapated into my case, I want it going out the back!
During my last computer build (3 years ago) I was happy to see the Arctic Cooling VGA silencer (which I own) offer a way to keep temps lower, keep the noise quieter, but most importantly exhaust the heat directly out of the case.
With this cooler, however, you are at the mercy of your power supply/case fans. If they aren't great, the true benefit of this cooler (overclocking) will not be as good as it could be (noise will obviously be improved).
As an aside, since I've never owned the 8800GT, is the stock cooler exhausted out of the case or just inside like 99% of all stock graphics cards? From the pictures it looks like the latter and so I guess this cooler is better...but I would have loved to see a similar setup with some ducting to get the heat out of the case.
Those stick on heat sinks are a real problem, especially on the mosfets at the end of the board. There is just too little surface area to stick to. I had one heat sink drop of onto the mainboard during opperation (luckily no damage). I found that a better heatsink for the mosfets could be made by cutting of the end of the old stock heatsink and screwing it on using the original mounting holes. (Anand, If your doing a round up, I'm using a Zalman VF900-Cu )
BTW - did you end up using the GT's on board fan header or a seperate one?
come on man,
did you write this up before you had to take a dump?
more pictures - please, especially of what it looks like in a case in a slot - with a ruler showing how much it sticks out
sli? hello? does it work - will it fit?
It would have been nice if you coulda gots a
MSI NX8800GT 512M OC GeForce 8800GT and done a compare since it already has a quad copper heatpipe cooler with a image of a hot babe on it.
The problem with most of these stock coolers is not the stock cooler themselves, but the installation job. Most of the time, the manufacturer applies way too much thermal paste.
You should do a comparison of the same stock cooler properly installed with an ideal amount of high grade thermal paste.
BTW, when you removed the stock cooler, could you have resued the stock cooler?
I happened to try the HR-03 Plus (410 grams without fan), and I found it to be too heavy on the video card...video card was bending under the weight of it. It looked scary. This product is essentially similar to HR-03 GT.
I replaced the HR-03 plus with an Arctic Cooling S1 revision 2 (268 grams without fan), and I get 35 C idle and 62 C max load with an EVGA 8800GT SC (650 Mhz GPU Core clock)--no fans. The S1 is lighter and mounts more securely to the 8800GT.
I've had better success with Thermalright CPU Coolers. HR-01 Plus with a Nexxus 120 mm fan is a good combo. I tried this cooler with a Noctua fan, and found that the fan was quiet, but did not move air at all. Any resistance in front or behind the Noctua negated its air moving functionality. The Nexus fan is still pretty quiet and moves air even with obstructions. The design of the Nexxus fan required a little bit of creativity with the Thermalright wire fan clips.
Doesn't the Noctua have low static pressure? I remember reading that at SPCR. I've used the SFLEX 1200RPM w/ the HR-03 and HR-03 plus with good success. I'm wonder if the HR-03 plus will fit my GTS 512MB...
Page 5 is now fixed. Recent revisions to our posting tools still have a few bugs which is why the save and append didn't work as it should. My apologies.
The test system Matt used for this review is anything but typical. He is working on a full-blown high-end gaming system with a high-end quad CPU and triple SLI. The case and cooling were designed to support and cool this gaming monster. His testing was just a single 8800GT stock vs. 8800GT HR-03 vs. Sparkle fanless 8800GT however.
You will see more about the system when the gaming system review posts in a couple of weeks. Considering all the extra cooling on the gaming rig it seemed best to add precautions that this system is best case cooling for a passive card.
The 92mm fan measures below the noise floor of the testing room so it was pointless to report noise. The fanless is zero and the 92mm used is below the noise floor. We did not want to mislead those concerned about noise by reporting the room noise floor as the sound level for the 92mm fan. As always the fan you choose determines the noise level of the Thermalright HR-03.
Interesting article... could you please elaborate on this "test system", such as the case used?
Also, for whatever reason I can't view the conclusion page, I get redirected to an error page each time. Using Firefox, but it might be something they installed on the campus computers...
I want to see them test the ZEROtherm GX710, i use it on my Geforce 6600GT and it works fantastically. Overclocked to 540Mhz (from 450) under heavy sustained load it doesnt top 55 degrees, a huge improvement over the stock cooler (70+ load) and it is dead silent. come on Anand, test some graphics coolers too.
Thermalright makes a few different versions of this cooler that are compatible with different cards. You can get a compatibility list from Thermalright's website:
2 of my GT's in SLI made very little noise under Crysis. The fans never spun up past default. Perhaps the room was hot, case airflow lacking, or a bum card/bios?
That's a good point. I mean, not a conspiracy, but perhaps that's a way nVidia kept the prices down on the 8800 GT - by providing a cheapo stock cooler design. Adding a higher quality one in bulk could add $20 or more to the MSRP.
The direct competitor for this monster is Arctic Cooling Accelero S1
Which is significantly smaller and has less pipes, but will fit normal case and SLI compatible.
Will be nice to have comparison of this two coolers
What is ambient temperature for your tests ? my Asus 8800GT (first reference design) running at 70C idle in ~20C ambient.
i got an accellero s1 on a 3850, and let me tell you, its huge. larger than this by a good margin, i have a 120mm fan on it, mounted in the middle of the cooler with zip ties, i could probably fit 2 side by side without extending past the sides of the cooler.
its almost as long as the pcb, maybe a centimeter smaller. and almost twice as tall.
overclocked to 770 core, 1044 mem, it idles at 28c and loads at around 31 32, in crysis.
it was 23.99 at newegg. pretty unbelievable performance for the dollar.
That sounds great even though the 8800GT's much hotter. I like this cooler for the 8800GT, but I'd love to see it directly compared to the S1, which is much cheaper than most coolers...and yet everyone swears by it. Honestly, my first generation 8800GT is just annoyingly loud at any speed over 40%...and it still goes to 80 degrees or so during games, so I'd love to put a new cooler for it if it keeps it more quiet AND more cool at the same time.
i agree, a comparison w/ 2-3 top aftermarket coolers is what i was expecting. I hate it when a huge site like anandtech compares coolers to stock. you didnt even compare it to another aftermarket cooler but a cooler on another videocard. what the hell are u guys on a budget or something??!?! i might as well just check any other amateur site for the review since u really didnt do anything elaborate fitting the anandtech pedigree. c'mon we expect high calibre reviews from you guys, not these paltry small scale reviews i can see on any 2bit site.
My explanation just created more confusion. I apologize. We also has a Sparkle fanless 8880GT available in another system we were testing. We included those results in the graph to show how the fanless Thermalright performed compared to the Sparkle fanless.
I will add some explantion in the article to better explain this comparison.
The stock Sparkle cooler we tested uses the stock nVidia cooling solution. Almost all the base 8800GT cards out there use this stock cooling design. We have corrected the labels in the graph so it is clear we are comparing apples and apples.
thanks for the clarifications wesley, we'll be looking forward to that gpu roundup then! i didnt realize this was anandtechs first gpu cooler review, bit of a surprise since they've been being used by enthusiasts since the first vid cards came out.:p
A GPU cooler roundup is a good idea, but that was not the goal of this article. This is our first GPU cooling review, and we wanted to see the impact of the aftermaket Thermalright cooler compared to stock cooling. Cooling performance was compared at idle and load, and frankly the performance improvement was better than we expected. The results certainly make the idea of a GPU cooler roundup in the future a good one.
As you will see in the review we did compare the Thermalright to both the stock nVidia cooler and the Sparkle fanless 8800GT cooler. The results for all three are in the performance chart on page 4.
I have revised the article a bit to make it clearer that the article compares the Thermalright HR-03, the stock nVidia cooler, and the Sparkle fanless 8800GT.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
43 Comments
Back to Article
axiomatic1 - Wednesday, March 5, 2008 - link
I had a similar problem with my Swiftech ram sinks for my 8800gt. The solution (given by swiftech when i called for help) worked like a charm. Essentially, IPA does not act as a degreasing agent. The reason that the sinks don't stick is because they can't. There is a residue of thermal grease on the chips. The way to remove this grease is to use a cleaner that removes grease. Acetone works wonders. When i used the acetone on my chips the ram sinks stuck very well. If this still doesn't work and you don't want the permanence of thermal epoxy there is another solution, also given by swiftech. You get a bottle of krazy glue and put a dab on a piece of paper. Take the pointy end of a paper clip and dip it in the glue and then dab the tip on the ram sink. This leaves enough glue on the sink to make the sink stick, but doesn't interfere with the thermal conductivity of the tape.Somewhat haphazard, but my ram sinks are still going strong.
7Enigma - Thursday, March 6, 2008 - link
Very interesting, thanks for the reply! I plan on building a system soon and will make sure to have both on hand when removing paste and reapplying better paste. I wonder if this holds true for all paste jobs (be it CPU, GPU, Northbridge,etc.). Might get a bit extra performance or at least be sure you have a clean starting point.Thorsson - Saturday, March 8, 2008 - link
Just be careful with Acetone, it's not like IPA that you can spill anywhere and will just quickly evaporate. Acetone is a strong solvent for most plastics and synthetic fibres. You don't want to see that headline, "Acetone ate my Mother....board".swing848 - Wednesday, August 4, 2010 - link
Although this thread is ancient, someone may find this useful.Another very good, and easily found, cleaner for computer components is ArctiClean 1 followed by ArctiClean 2.
ArctiClean 1 & 2 can be found as a package at newegg.com
Philscbx - Wednesday, June 2, 2010 - link
I'd like to recommend a safer degreasing agent used for twenty years.It's 'electronics contact cleaner' which is not acetone, and wont melt plastic.
We use acetone to deliberately melt plastics on aircraft interior repairs.
Also used for epoxy.
Acetone is not really a good degreaser where brake/contact cleaner is, and less damaging.
Smear axle grease sample on scrap anything, from aluminum foil - etc, and actually do the test with both solvents.
Brake/contact cleaner is highly used for final building of drag race engines.
It removes all trace of grease and oils holding fine grinding residue.
The difference using 'electronics labeled' contact cleaner is insurance it won't damage.
Cheers.
joeythecat - Wednesday, March 5, 2008 - link
I guess my first question would be does this cooling solution fit in a Antec P182 case?MamiyaOtaru - Tuesday, March 4, 2008 - link
I am loving this cooler. With my mobo I can wrap it around the back of the card without interfering with the cpu or northbridge tower coolers. It means I lose no expansion slots. It also puts the cooler right next to my case exhaust fan, so I don't need to mount a fan on the cooler.It's overkill, but I have it mounted on a 7900gt. Idles at around 33, loads at 39 (again, no fan on the cooler). The 7900gt isn't really a speed demon anymore, but I do get a decent peformer running very cool with no extra noise. It's very ideal for me.
7Enigma - Tuesday, March 4, 2008 - link
I am amazed at the results, truly amazed. Unfortunately this design has a crippling flaw that your test (as you mentioned in the article so I'm not faulting Anand) will not show. The hot air is still INSIDE the case!As the GPU is notorious for being the largest producer of heat in a case (barring some crazy OC'd quad-core), I don't want the product of that energy to be dissapated into my case, I want it going out the back!
During my last computer build (3 years ago) I was happy to see the Arctic Cooling VGA silencer (which I own) offer a way to keep temps lower, keep the noise quieter, but most importantly exhaust the heat directly out of the case.
With this cooler, however, you are at the mercy of your power supply/case fans. If they aren't great, the true benefit of this cooler (overclocking) will not be as good as it could be (noise will obviously be improved).
As an aside, since I've never owned the 8800GT, is the stock cooler exhausted out of the case or just inside like 99% of all stock graphics cards? From the pictures it looks like the latter and so I guess this cooler is better...but I would have loved to see a similar setup with some ducting to get the heat out of the case.
Sooticus - Monday, March 3, 2008 - link
Those stick on heat sinks are a real problem, especially on the mosfets at the end of the board. There is just too little surface area to stick to. I had one heat sink drop of onto the mainboard during opperation (luckily no damage). I found that a better heatsink for the mosfets could be made by cutting of the end of the old stock heatsink and screwing it on using the original mounting holes. (Anand, If your doing a round up, I'm using a Zalman VF900-Cu )BTW - did you end up using the GT's on board fan header or a seperate one?
KLC - Monday, March 3, 2008 - link
Great article, I'd love to actually see the product but all I see is text and blank spaces where pics should be. What is going on?dare2savefreedom - Monday, March 3, 2008 - link
come on man,did you write this up before you had to take a dump?
more pictures - please, especially of what it looks like in a case in a slot - with a ruler showing how much it sticks out
sli? hello? does it work - will it fit?
It would have been nice if you coulda gots a
MSI NX8800GT 512M OC GeForce 8800GT and done a compare since it already has a quad copper heatpipe cooler with a image of a hot babe on it.
dare2savefreedom - Monday, March 3, 2008 - link
The hot babe image makes it go faster too - even faster than using the color red like the slow little red guys do.tigersty1e - Monday, March 3, 2008 - link
The problem with most of these stock coolers is not the stock cooler themselves, but the installation job. Most of the time, the manufacturer applies way too much thermal paste.You should do a comparison of the same stock cooler properly installed with an ideal amount of high grade thermal paste.
BTW, when you removed the stock cooler, could you have resued the stock cooler?
derk - Monday, March 3, 2008 - link
I happened to try the HR-03 Plus (410 grams without fan), and I found it to be too heavy on the video card...video card was bending under the weight of it. It looked scary. This product is essentially similar to HR-03 GT.I replaced the HR-03 plus with an Arctic Cooling S1 revision 2 (268 grams without fan), and I get 35 C idle and 62 C max load with an EVGA 8800GT SC (650 Mhz GPU Core clock)--no fans. The S1 is lighter and mounts more securely to the 8800GT.
I've had better success with Thermalright CPU Coolers. HR-01 Plus with a Nexxus 120 mm fan is a good combo. I tried this cooler with a Noctua fan, and found that the fan was quiet, but did not move air at all. Any resistance in front or behind the Noctua negated its air moving functionality. The Nexus fan is still pretty quiet and moves air even with obstructions. The design of the Nexxus fan required a little bit of creativity with the Thermalright wire fan clips.
HTH
Jodiuh - Monday, March 3, 2008 - link
Doesn't the Noctua have low static pressure? I remember reading that at SPCR. I've used the SFLEX 1200RPM w/ the HR-03 and HR-03 plus with good success. I'm wonder if the HR-03 plus will fit my GTS 512MB...NARC4457 - Monday, March 3, 2008 - link
No, I've got the same error for page 5.And I'm definitely interested in a GPU HSF roundup as well!
Wesley Fink - Monday, March 3, 2008 - link
Page 5 is now fixed. Recent revisions to our posting tools still have a few bugs which is why the save and append didn't work as it should. My apologies.The test system Matt used for this review is anything but typical. He is working on a full-blown high-end gaming system with a high-end quad CPU and triple SLI. The case and cooling were designed to support and cool this gaming monster. His testing was just a single 8800GT stock vs. 8800GT HR-03 vs. Sparkle fanless 8800GT however.
You will see more about the system when the gaming system review posts in a couple of weeks. Considering all the extra cooling on the gaming rig it seemed best to add precautions that this system is best case cooling for a passive card.
The 92mm fan measures below the noise floor of the testing room so it was pointless to report noise. The fanless is zero and the 92mm used is below the noise floor. We did not want to mislead those concerned about noise by reporting the room noise floor as the sound level for the 92mm fan. As always the fan you choose determines the noise level of the Thermalright HR-03.
alkalinetaupehat - Monday, March 3, 2008 - link
I got to the conclusion in IE and then in Firefox after C/P the link from IE...Kougar - Monday, March 3, 2008 - link
Interesting article... could you please elaborate on this "test system", such as the case used?Also, for whatever reason I can't view the conclusion page, I get redirected to an error page each time. Using Firefox, but it might be something they installed on the campus computers...
rudolphna - Monday, March 3, 2008 - link
I want to see them test the ZEROtherm GX710, i use it on my Geforce 6600GT and it works fantastically. Overclocked to 540Mhz (from 450) under heavy sustained load it doesnt top 55 degrees, a huge improvement over the stock cooler (70+ load) and it is dead silent. come on Anand, test some graphics coolers too.rudolphna - Monday, March 3, 2008 - link
http://www.newegg.com/Product/Product.aspx?Item=N8...">http://www.newegg.com/Product/Product.aspx?Item=N8...adam92682 - Monday, March 3, 2008 - link
can this be used for the 8800 gts as well?KikassAssassin - Monday, March 3, 2008 - link
Thermalright makes a few different versions of this cooler that are compatible with different cards. You can get a compatibility list from Thermalright's website:http://www.thermalright.com/new_a_page/product_pag...">http://www.thermalright.com/new_a_page/product_pag...
It looks like the HR-03 GT will work with the G92 GTS, but the G80 GTS would use the HR-03 Plus.
cubeless - Monday, March 3, 2008 - link
it's the first time i've had to change the hsf on a brand new card... over 25c difference with a zalman...at least the big, noisy hsf on my x1800 kept the gpu cool whilst sounding like a hovercraft...
the hsf on the 8800gt is a big, noisy, expensive looking piece of junk... it's got to be a conspiracy to sell aftermarket coolers...
Jodiuh - Monday, March 3, 2008 - link
2 of my GT's in SLI made very little noise under Crysis. The fans never spun up past default. Perhaps the room was hot, case airflow lacking, or a bum card/bios?crimson117 - Monday, March 3, 2008 - link
That's a good point. I mean, not a conspiracy, but perhaps that's a way nVidia kept the prices down on the 8800 GT - by providing a cheapo stock cooler design. Adding a higher quality one in bulk could add $20 or more to the MSRP.Zefram0911 - Monday, March 3, 2008 - link
No sound readings?crimson117 - Monday, March 3, 2008 - link
The sound reading for the fanless one is 0 DBA.The sound reading with the fan turned on is very close to the fan's own DBA reading, and depends on which fan you choose for yourself.
gonks - Monday, March 3, 2008 - link
Would a 120mm fan fit in this beast?nowayout99 - Monday, March 3, 2008 - link
It can if you use Thermalright's 120mm fan clips. (the package only includes the 92mm ones).akak1997 - Monday, March 3, 2008 - link
HiIs it possible to post some pictures of the card with the heatsink (with and without fan) when it is installed on the motherboard and case?
thanks a bunch
n0nsense - Monday, March 3, 2008 - link
The direct competitor for this monster is Arctic Cooling Accelero S1Which is significantly smaller and has less pipes, but will fit normal case and SLI compatible.
Will be nice to have comparison of this two coolers
What is ambient temperature for your tests ? my Asus 8800GT (first reference design) running at 70C idle in ~20C ambient.
snarfbot - Monday, March 3, 2008 - link
i got an accellero s1 on a 3850, and let me tell you, its huge. larger than this by a good margin, i have a 120mm fan on it, mounted in the middle of the cooler with zip ties, i could probably fit 2 side by side without extending past the sides of the cooler.its almost as long as the pcb, maybe a centimeter smaller. and almost twice as tall.
overclocked to 770 core, 1044 mem, it idles at 28c and loads at around 31 32, in crysis.
it was 23.99 at newegg. pretty unbelievable performance for the dollar.
iSOBigD - Tuesday, March 4, 2008 - link
That sounds great even though the 8800GT's much hotter. I like this cooler for the 8800GT, but I'd love to see it directly compared to the S1, which is much cheaper than most coolers...and yet everyone swears by it. Honestly, my first generation 8800GT is just annoyingly loud at any speed over 40%...and it still goes to 80 degrees or so during games, so I'd love to put a new cooler for it if it keeps it more quiet AND more cool at the same time.iSOBigD - Tuesday, March 4, 2008 - link
put a new cooler ON it*poohbear - Monday, March 3, 2008 - link
i agree, a comparison w/ 2-3 top aftermarket coolers is what i was expecting. I hate it when a huge site like anandtech compares coolers to stock. you didnt even compare it to another aftermarket cooler but a cooler on another videocard. what the hell are u guys on a budget or something??!?! i might as well just check any other amateur site for the review since u really didnt do anything elaborate fitting the anandtech pedigree. c'mon we expect high calibre reviews from you guys, not these paltry small scale reviews i can see on any 2bit site.Wesley Fink - Monday, March 3, 2008 - link
My explanation just created more confusion. I apologize. We also has a Sparkle fanless 8880GT available in another system we were testing. We included those results in the graph to show how the fanless Thermalright performed compared to the Sparkle fanless.I will add some explantion in the article to better explain this comparison.
Wesley Fink - Monday, March 3, 2008 - link
The stock Sparkle cooler we tested uses the stock nVidia cooling solution. Almost all the base 8800GT cards out there use this stock cooling design. We have corrected the labels in the graph so it is clear we are comparing apples and apples.poohbear - Tuesday, March 4, 2008 - link
thanks for the clarifications wesley, we'll be looking forward to that gpu roundup then! i didnt realize this was anandtechs first gpu cooler review, bit of a surprise since they've been being used by enthusiasts since the first vid cards came out.:pWesley Fink - Monday, March 3, 2008 - link
A GPU cooler roundup is a good idea, but that was not the goal of this article. This is our first GPU cooling review, and we wanted to see the impact of the aftermaket Thermalright cooler compared to stock cooling. Cooling performance was compared at idle and load, and frankly the performance improvement was better than we expected. The results certainly make the idea of a GPU cooler roundup in the future a good one.Wesley Fink - Monday, March 3, 2008 - link
As you will see in the review we did compare the Thermalright to both the stock nVidia cooler and the Sparkle fanless 8800GT cooler. The results for all three are in the performance chart on page 4.I have revised the article a bit to make it clearer that the article compares the Thermalright HR-03, the stock nVidia cooler, and the Sparkle fanless 8800GT.
fendell - Monday, March 3, 2008 - link
It's in the name I guess, ThermalRIGHT? ;)I do wonder tho, why didnt you overclock the card and include Temps at various speeds as well?
soydeedo - Monday, March 3, 2008 - link
"I do wonder tho, why didnt you overclock the card and include Temps at various speeds as well?"I was wondering the same. When someone is looking for info on such a heavy duty cooler, that should be the obvious test.