Sapphire's liquid metal cooling - fanless Radeon X850XT PE

Storm · Post by **Storm** » Wed May 18, 2005 1:14 pm

Early liquid metal cooling prototype cooled Radeon X850XT PE card to just 12 Celsius. The final design will be fanless.

Info and pictures @ theinquirer

Mats · Post by **Mats** » Wed May 18, 2005 2:03 pm

I wonder what this technology could do in other areas such as CPU cooling. Maybe the power draw of the CPU will become less important over time. Ok, don't missunderstand me, low power is always important. But what if we will see some HS that can cool hotter CPU's silently? That X850 draws like 70 W or what, so how much goes to the GPU only? 50-60 W? I hope Thermalright and Zalman gets this technology...

sgtpokey · Post by **sgtpokey** » Wed May 18, 2005 2:07 pm

Interesting,
But I'm wondering if the Inquirer has some things wrong... It's great about this "Liquid Metal" thing, but I'm not sure if it'll be able to get an 850XT-class card to run fanless.

Afterall, no matter how efficient the new heatpipes are, the heat STILL needs to be radiated away and if it was a fanless design, that would be a function of the exposed surface area (since they would lose the forced air circulation).

Seems to me they'd have the same issues that Zalman has: how to get enough surface area to radiate the heat away without benefit of forced air coolling.

PassiveMan · Post by **PassiveMan** » Wed May 18, 2005 2:16 pm

It's very interesting indeed. I always wondered about liquid metal from science experiments at school. Remember watching how mercury just seem to flow like T3.

Imagine the conduction of liquid metal?

Maybe the next phase in the future would be to have something like a Zalmans reservator but full of liquid metal instead.

StarfishChris · Post by **StarfishChris** » Wed May 18, 2005 2:26 pm

It may arrive sooner than you think

blaze · Post by **blaze** » Wed May 18, 2005 3:41 pm

Technology like this would be on CPU's already if its wasn't for large companies realizing their profits would be eaten into if they were to go ahead and release a computer with no moving parts (aside from the HD). Profiteering gluttons are well aware that heat will eventually become such a huge problem that they will eventually have to give in. Although I suspect most will still find a way to sneak a fan in there. Fans are good for the economy, retards buy whole new systems due to fans failing or fans buzzing due to bearings wearing down.

Mats · Post by **Mats** » Wed May 18, 2005 5:08 pm

sgtpokey wrote:Afterall, no matter how efficient the new heatpipes are, the heat STILL needs to be radiated away and if it was a fanless design, that would be a function of the exposed surface area (since they would lose the forced air circulation).

Remember that although this may be a silent solution, but it's not passive cooling (who cares?). Ok, no fans and no (regular) pumps, but still something that actively makes the liquid move. Look at StarFishChris' link, this kind of system may help breaking the downward spiral more noise>more enclosure>more heat>more fans>more noise. Use a more compact, well ventilated case instead with lower airflow and put the radiators on the outside and just make sure you got good EMI protection and good enclosure for the HD's only.

Still, I don't understand how they can get under ambient temp(assuming ambient temp>12 C). The EM pump moves the liquid, but there's no TEC function or such AFAIK.... strange.

rei · Post by **rei** » Wed May 18, 2005 5:24 pm

considering ditching my gigabyte silentpipe x800xl for a sapphire x800xl ultimate:

http://www.bjorn3d.com/read.php?cID=793

vf700 w/ warranty. i don't think it has VIVO though

might have to complement it with a sapphire rage theatre 550

Mats · Post by **Mats** » Wed May 18, 2005 5:55 pm

rei wrote:considering ditching my gigabyte silentpipe x800xl for a sapphire x800xl ultimate

WHY?
Edit: Ok, you don't really mean "ditching" that card...
Stupid review BTW. I don't think they bumped up the fan speed when overclocking, but still they complain about the VPU reaching 7 MHz lower than a standard card. That cooler can do better, just bump up the speed to the noise level of a 50 mm screamer and see what happens. Yeah I know how overclockers think....

StarfishChris · Post by **StarfishChris** » Wed May 18, 2005 6:21 pm

blaze wrote:Technology like this would be on CPU's already if its wasn't for large companies realizing their profits would be eaten into if they were to go ahead and release a computer with no moving parts (aside from the HD). Profiteering gluttons are well aware that heat will eventually become such a huge problem that they will eventually have to give in. Although I suspect most will still find a way to sneak a fan in there. Fans are good for the economy, retards buy whole new systems due to fans failing or fans buzzing due to bearings wearing down.

And you think the companies won't engineer these pumps to have shorter shelf lives?

rei · Post by **rei** » Wed May 18, 2005 7:38 pm

mats: having problems keeping it cool it seems and the vf700 seems to review good. i'd rather something with it bolted it on from the factory so it'd be under warranty since i'm particularly averse to removing the hsf from a $500 video card.

er, by ditching i meant selling

any other mfr-ers make their cards with vf700 already on? i know galaxy makes the 6800 non-gt and 6600 with it. not sure about any ones with radeon solutions.

i'm not going to take the silentpipe off my card! not brave enough.

looks like gecube has a 512mb silentpipe competitor:

http://www.gecube.com/mbu/productCategory.php?proID=106

tay · Post by **tay** » Wed May 18, 2005 7:59 pm

Mats wrote:Still, I don't understand how they can get under ambient temp(assuming ambient temp>12 C). The EM pump moves the liquid, but there's no TEC function or such AFAIK.... strange.

Pffft thermodynamics shmermodynamics!!

Mats · Post by **Mats** » Thu May 19, 2005 5:38 am

rei: What about adding a fan in front of the card blowing backwards together with a duct? You probably won't need much airflow to keep it enough cool. Much cheaper to find a low noise fan than exchanging card...

frostedflakes · Post by **frostedflakes** » Thu May 19, 2005 6:03 am

The 12*C has to be a typo/misinterpretation. I'm thinking they might've meant 12*C above ambient, which would be a very good, but still reasonable with good cooling, temp for a VGA card.

rei · Post by **rei** » Thu May 19, 2005 6:27 am

I have a low-noise 120mm Nexus/Yate-Loon fan sitting at 45 degrees right at the tip of the card blowing air back. Right above the video card at the back of the case I have a 92mm Nexus exhausting.

At 6v they're inaudible but don't move much air. At 12v they are slightly more audible and move more air so that it cools the card down by 10 degrees slightly.

Temps w/ fans on low: 60-65 C idle in Windows 2D, 95-105 C load
Temps w/ fans on max: 50-60 C idle in Windows 2D, 80-90 C load

I also had a -third- fan, mounted on a Zalman FB123 fan bracket directly overtop of the video card/PCI slots, elevated on 12 Antec mobo standoffs:

But it was getting too noisy at that point. :/

I dunno, this doesn't seem normal, when other people's fanless silentpipes claim 40-50 C idle and 50-60 load with nothing more than 1 case fan.

45-55 idle with that 80mm fan on high, 65-80 C load

sgtpokey · Post by **sgtpokey** » Thu May 19, 2005 6:45 am

Remember that although this may be a silent solution, but it's not passive cooling (who cares?). Ok, no fans and no (regular) pumps, but still something that actively makes the liquid move. Look at StarFishChris' link

Not disagreeing with you, but StarFishChris's link is my point, i.e., the heatpipes transport the heat to a location AWAY from the CPU in starfish Chris's link.

What I'm talking about is the statement in the Inquirer article that a video card could be run fanlessly using Liquid Metal.

I'm wondering about how good of a solution that would be because in a standalone video card, there's not many places you can heatpipe far enough AWAY from your other components. The heat will still be relatively close to various components within your system, and in that sense you've got the same problem as a standard fanless heatpiped videocard. It would be better for the heatpipes to link to some radiator mechanism that can radiate some of the heat outside of the component area. This means either attach to the side of a case so that some of the radiation occurs on the external surface*, or at least heatpipe it further away from other components as in the StarfishChris link.

I guess I'm just wondering what benefit a more efficient heatpipe would be in this specfic case [fanless video card], since I am guessing that the limiting factor is not the transport efficiency of the heatpipes, but that the heat doesn't get transported far enough away from your other components.

* Note: There are two solutions that do what I think is better; both are fairly pricey HTPC cases from Atech-Fabrication, and one from mCubed (coming in June). Their solutions run the heatpipes to the the sides of the case, which are big finned radiators, ala the Hush systems.

Mats · Post by **Mats** » Thu May 19, 2005 6:48 am

rei: Well, power is nothing without control. What's the idea with having fans pointing in different directions with nothing controlling the air flow direction inside the case? Of course it gets noisy with all those fans!

Mats · Post by **Mats** » Thu May 19, 2005 6:53 am

sgtpokey wrote:
Remember that although this may be a silent solution, but it's not passive cooling (who cares?). Ok, no fans and no (regular) pumps, but still something that actively makes the liquid move. Look at StarFishChris' link
Not disagreeing with you, but StarFishChris's link is my point, i.e., the heatpipes transport the heat to a location AWAY from the CPU in starfish Chris's link.

What I'm talking about is the statement in the Inquirer article that a video card could be run fanlessly using Liquid Metal.

I'm wondering about how good of a solution that would be because in a standalone video card, there's not many places you can heatpipe far enough AWAY from your other components. The heat will still be relatively close to various components within your system, and in that sense you've got the same problem as a standard fanless heatpiped videocard. It would be better for the heatpipes to link to some radiator mechanism that can radiate some of the heat outside of the component area. This means either attach to the side of a case so that some of the radiation occurs on the external surface*, or at least heatpipe it further away from other components as in the StarfishChris link.

I guess I'm just wondering what benefit a more efficient heatpipe would be in this specfic case [fanless video card], since I am guessing that the limiting factor is not the transport efficiency of the heatpipes, but that the heat doesn't get transported far enough away from your other components.

I agree:

Mats wrote:Use a more compact, well ventilated case instead with lower airflow and put the radiators on the outside and just make sure you got good EMI protection and good enclosure for the HD's only.

rei · Post by **rei** » Thu May 19, 2005 7:12 am

While my original reply had -something- to do with the new fanless Sapphire (referring to the Zalmanified Sapphire as an alternative) it's gotten off topic to my self-absorbed Silentpipe troubles. So if you could help me out, please do so in the other thread:

http://forums.silentpcreview.com/viewtopic.php?t=19067

thanks

(gee, should i tape up the sides vents or something?)

~El~Jefe~ · Post by **~El~Jefe~** » Thu May 19, 2005 2:56 pm

well, one thing must happen, the case will heat up, a fanned psu will still ramp up. only way to do this is to still have fan in the system.

I like this idea. i just hope it doesnt crack open and instantaneously fry every piece of equipment in the case as well cause a severe toxic shock/death to anyone near the case..

a load hot, splattering mercury is a bad thing.

StarfishChris · Post by **StarfishChris** » Thu May 19, 2005 3:14 pm

What makes you so certain they'll contain mercury?

alglove · Post by **alglove** » Thu May 19, 2005 4:08 pm

It would not be mercury... most probably some gallium/indium alloy.

Mats · Post by **Mats** » Thu May 19, 2005 4:43 pm

~El~Jefe~ wrote:well, one thing must happen, the case will heat up, a fanned psu will still ramp up. only way to do this is to still have fan in the system.

I like this idea. i just hope it doesnt crack open and instantaneously fry every piece of equipment in the case as well cause a severe toxic shock/death to anyone near the case..

a load hot, splattering mercury is a bad thing.

I still don't think a traditional case would be the best thing for this (just like it isn't for fanless PSU's), especially if you use the same technology on the CPU. If there's no noise there's no reason for a sealed case with only fans doing the ventilation job. Imagine using this in an Antec P180 but with the top of the case being even more open. There's nothing to enclose except EMR and keep the dust outside, the latter is also quite easy to achieve beause you don't need much airflow.

BenW · Post by **BenW** » Fri May 20, 2005 11:00 am

Don't know if anyone else mentioned this but how can they cool a card to 12C if the ambient temperature is more than that? I cant see there being a freezer in the cooling system, bet that measurment was taken in a room at 1C or something silly like that

halcyon · Post by **halcyon** » Fri May 20, 2005 11:56 am

the 12C is probably a bollocks figure. As far as I've understood it, it's just a more efficient heat transfer mechanism than direct solid metal heatsinks or water cooling.

Hence, it can't cool down to a lesser degree than the surrounding temp + it's own efficiency, regardless of how big a radiator it has.

So, it still has to have some way of dissipating with the heat, thus it has two fans blowing onto some fins.

ddrueding1 · Post by **ddrueding1** » Fri May 20, 2005 12:17 pm

Some people are getting it and some people clearly aren't.

*This isn't magic.*

The heat is still being generated and still has to be dissapated somewhere. Since we are talking about liquid metal, don't expect to be able to plumb this yourself to an outdoor ratiator. Therefore all this technology does is move the heat around the card. This is the exact same problem as with existing heatpipe-based cards: The heat is still on the card and must be dissipated via massive heatsinks and some airflow.

So what is this doing for me again?

mathias · Post by **mathias** » Fri May 20, 2005 12:33 pm

So the question is how could this be different from heatpipes?

Maybe the piping could be bendable. Not openable llike watercooling, but it could be pre-built for connecting to something outside. And because it wouldn't be hard and in a fixed position, that should get rid of the problems that people see with the Thermaltake Schooner.

On second thought, why couldn't this be done with heatpipes? I don't see why the centre of heatpipes couldn't be nonmetal. That part wouldn't let heat out, but so what?

Mats · Post by **Mats** » Fri May 20, 2005 12:39 pm

Just forget that 12 degrees for now, I still think they mean 12 degrees above ambient. Anyway, I prefer Sapphire's own words -

Sapphire wrote:10 degrees cooler then Artic Cooling solution

This is what they say about noise, I guess it got a fan or two in there after all -

Sapphire wrote:25% more silent then standard ATI solution

frostedflakes · Post by **frostedflakes** » Fri May 20, 2005 3:33 pm

Liquid metal cooling differs from traditional heatpipes because it is an active method of cooling. The substance inside the pipes is being actively pumped through the loop. Whereas heatpipes still rely completely on natural processes (convection, condensation, etc.) to do all the work.

As others have mentioned, this is not some magic cooling. The same amount of heat still has to be dissipated. However, what this can do is spread out heat much more effectively than heatpipe technology. Keep in mind that with a heatsink, most of the heat is being dissipated directly over the core, and not much spreads out to the fins. Heatpipes help to spread this heat more evenly over the entire surface area of the heatsink. Liquid metal should be able to do this even more effectively.

Tobias · Post by **Tobias** » Sat May 21, 2005 1:07 am

The bigger the difference between the surface to be cooled and the cooling air, the more effetive the cooling. What the liquid metal is doing is to raise the temperature on the surface to be cooled (just like Arctic silver allows more heat to be passed from the core to the heatsink) and thus make the coolingsolution of choice more efficient.