Date: November 8, 2010
Author(s): Rob Williams
NVIDIA’s GeForce GTX 470 is one of the better GPU offerings on the market right now, but it’s also one of the hottest. It’s for that reason that after-market coolers exist, although few get quite as high-end as CoolIT’s Omni A.L.C. – and the ~$200 price tag proves it. Given that price, let’s take a look and see just what this thing is made of.
As our graphics cards continue to get faster and faster from one generation to the next, they’re also becoming more power-efficient thanks to the valiant efforts of AMD and NVIDIA. The problem, though, is that unlike our CPUs, which seem to reflect their increased power efficiency with lower temperatures, we don’t quite see the same thing with our graphics cards.
The reason is simple. Graphics cards today have a tremendous number of transistors packed in, and overall, the die sizes are generally much larger than that of a CPU. Thanks in part to this, and also the fact that GPU vendors are keen on packing in as much performance as possible, even today’s latest cards can run quite hot – upwards of 80°C or more at full load. Compare that to a typical ~60°C for even the highest-end of processors.
Although even at 80°C, the temperatures are still well within the specs that GPU vendors set, it’s natural that gamers would want to take some steps to see them go down. After all, if GPU temperatures go down, then the temperatures inside of the chassis will as well. To feed that desire, cooler companies have put out many different DIY products that prove to be far better solutions than reference designs.
In the case of CoolIT’s Omni A.L.C. (Advanced Liquid Cooling), this is a solution that’s not for the feint of heart. No, it’s not because it’s Ã¼ber-complex and difficult to install, but rather because it’s expensive. Oh so expensive. While NVIDIA’s GeForce GTX 470 retails for around $260, the Omni sells for about $200. Yes, that’s one high Cooler-to-Product price ratio.
Self-contained liquid coolers are not exactly new, but over the years their designs have been improved upon quite a bit. At the same time, products like these aren’t common for GPUs, but rather CPUs, so CoolIT is doing something quite unique here. Similar to other designs, the Omni includes a radiator/fan combo that hooks up to the back fan location of your chassis, and the tubing runs down to the block, which you must pre-install to your GPU.
With CPU coolers that use a similar design, such as Corsair’s H50/H70, the water pump is attached to the block, making installation a breeze. Because CoolIT locates the pump for the Omni on the radiator, it results in a water block that’s about as slim as could possibly be.
For the sake of not getting dust on the thermal pads prior to installation, I left the protective plastic tabs on for the sake of this photo. Because each one of these waterblocks (interposer plates) are designed for a specific GPU, the block is designed to perfectly sit atop your card’s PCB. Screw holes will align, and so will the thermal pads. Figuring out the orientation of how to install this block is made easy, since you can only rotate your card one way and have it line up with the copper GPU heatsink.
If you’ve seen self-contained liquid coolers before, this is a sight you are likely familiar with. In all liquid-cooled designs, heat is dissipated with the help of a radiator, which in this case is a single 120mm design. To help keep things even cooler, and get heat out of the problem area, a fan is also used to get the flow moving.
In this particular case, CoolIT also located the pump as part of this same unit, rather than locate it on the block itself. The reason this is done is so that the GPU keeps as slim as possible, preventing blockage from occurring if you want to fill up all of your PCI or PCI Express slots. This implementation is smart, but does carry a rather large caveat, which I’ll cover on the next page.
In this side view, you can likely begin to understand what that problem might be:
CoolIT’s Omni is quite a good-looking product, and has major potential to keep your cards about as cool as they could be without getting into more extreme measures. But, while some self-contained units are simple to install, others are not. Let’s see which club the Omni belongs to.
The first step to installing an after-market cooler to a graphics card is to first tear the original cooler off, and clean it up. Some coolers are easier to remove than others, and in the case of our particular model, an EVGA SuperClocked, the cooler was easy to pull off after the removal of what seemed like a hundred small screws.
Sometimes it’s easy to overlook some screws, though, so bear in mind that in addition to the screws on the back of the PCB, there is also one in the corner of the metal bracket, and also two more located right in the center of the bracket itself, in between the video ports.
Once the cooler is removed, you can clean the spent thermal paste off of the GPU core via a couple of different methods. Personally, I use some paper towel and wipe it all off (sometimes it might take more than one piece), and then I clean it up with some pure alcohol, making sure not to use too much.
The most difficult part of cleaning up a graphics card is the memory chips. On occasion I have used Q-Tips dipped in alcohol to get the job done, but if the chips are for the most part clean and free thick residue, which this card happened to be, I use my thumbs to clean everything up. It sounds like a strange method, but it works, and works well. Afterward, you could take the Q-Tip route if you wanted to make sure the chips are as clean as possible.
When finished, your GeForce GTX 470 should look like this:
I have to say that while NVIDIA’s recent GPUs tend to run rather hot, the core is much easier to clean thanks to its heatsinked design, akin to our CPUs.
With all of the protective plastic removed from the block, the card attaches to it as seen in the photo below, You don’t have to fiddle with too much, as the screw holes line up perfectly. A bit strangely, CoolIT doesn’t include screws to hold this block in place, but instead relies on you to use the same screws that were removed from your previous cooler. It’s an odd choice, to say the least.
As you can also see, the block itself is a true finger-print magnet. You’ll be set if anyone ever steals your GPU and it’s recovered later!
It was at this point that I realized just how much I goofed with the installation. Because the test PC uses Corsair’s H50 CPU cooler, mounting the Omni became a problem, since both use the same fan location in the back of the chassis for the radiator. After cursing a bit at the PC, I grumbled the H50 right out of there, and replaced it with a stock Intel heatsink. “Stock” should be used loosely though, as it’s a rather robust offering (it comes included with the Core i7-980X Extreme Edition).
Despite swapping out the CPU cooler, I wasn’t exactly home free. Because the Omni is rather bulky, which is not at all uncommon for enthusiast parts, it literally prevented me from installing the radiator properly. The reason is due exclusively to the pump being located there, and had it not been there, the install would have been a cinch.
You can see the less-than-stellar install above. Because I didn’t have a thinner CPU cooler easily-accessible to install, my only choice at this point was to unscrew the pump from the unit and push it to the side. Fortunately, even though the pump was no longer attached to the radiator, it was still oriented perfectly, and the chassis door could still be used.
I don’t entirely disagree with CoolIT’s decision to put the pump right onto the radiator, because in truth, it’s a common-sense design. I’ve never seen a super-thin pump before, and in order to retain the slim design of the block itself, the pump has little leniency.
At the same time, its location is a rather major problem for most builds. As I mentioned, the Intel heatsink we’re using is bulky, but even so, had we used something like Noctua’s NH-U12P, that pump would still be touching the heatsink. And for all intents and purposes, that’s a thin cooler. The issue can be worked around using the procedure I used, but that’s far from ideal, and may not be an option for a lot of people. Our test PC is using a rather hefty chassis here (Cooler Master HAF X), and had the door been less lenient, I simply wouldn’t have been able to use it.
To help see what CoolIT’s Omni is made of, we tested out our GeForce GTX 470 at both stock and overclocked speeds with both the original cooler and the Omni. Please note that the card we used, EVGA’s SuperClock, while sticking to a reference design, ups the cooling power with the help of a backplate that dissipates heat from the back portion of the PCB. This in essence means our GTX 470 model cools a little better to begin with than a standard version, so bear that in mind.
The reference clocks for the GTX 470 are 607MHz Core, 1215MHz Shader and 1674MHz Memory. Our overclock, which remained the same regardless of cooler, was 740MHz Core, 1480MHz Shader and 1725MHz Memory. In our experience with the Omni, having the much better cooling power didn’t help us achieve a higher overclock, and quite frankly, that’s not something we expected to happen. Nor is it something someone who purchases a $200 cooler cares about. At that price, it’d make more sense to purchase a second card, or a better one.
The PC used for testing is the same one all of our GPU testing is conducted on, and for the sake of saving space on this page, we’re not going to post a specs table or delve into our methodology too deeply. If you want to learn more about how we do things, you can refer to our methodology page from a recent review. In a listos version, our PC includes a Core i7-975 overclocked to 4.0GHz, 12GB of memory and a Cooler Master HAF X chassis.
To stress the GTX 470, we use OCCT 3.1, as it’s absurdly brutal on any GPU in terms of power draw and temperatures. Our test PC is highly optimized for airflow, and our Windows OS tweaked to keep results as reliable as possible. After booting up the PC, we start GPU-Z immediately to record temperatures, and after five minutes of remaining idle, we employ OCCT to do what it does best. The “Load” wattage is the maximum value recorded.
Well. It’s clear that the reference GTX 470 cooler, even when used in conjunction with the backplate, is just no match for the Omni. The differences here are enormous. Because we couldn’t access the voltage level on the card, the temperature differences between stock and overclocked settings were almost nil, and in the end, the Omni dropped more than 30°C off of our load temperatures, and 6°C off of the idle. Impressive showing, but to be expected.
It’s clear that where temperatures are concerned, CoolIT’s Omni excels. While overclocking wasn’t improved, the temperature drops were massive, and even after a half-hour long OCCT stress, our GTX didn’t move an inch past 64°C, even with our huge overclock in place. The best part is that through all this, the sound from the cooler was almost non-existent.
I don’t have a sound meter here, unfortunately, but I can assure you, it just doesn’t get much quieter than this. To get a proper measure, I turned off anything in the room that had noise, and disabled all of the system fans in our test PC, aside from the Omni’s. During this testing, it was the hard drive I could hear easily, not the Omni (another reason to toss an SSD in that machine, perhaps).
Omni’s fan is rated at 1100~2500RPM, but in all our testing, the speed didn’t seem to change at all. This was measured simply by keeping my hand near the fan exhaust while the PC was idle and going into the stress-test mode. Simply put, the Omni is silent, merely equating to a barely audible hum during use.
Noise and performance aside, I’m a little baffled in understanding just who the Omni is designed for. At its ~$200 price-point, the cooler costs almost as much as the graphics card itself, making it an ultra-niche product. If noise and temperatures are all that matter to you, it’s a great option, but still a a pricey one.
Compare the Omni to Corsair’s H50 CPU cooler, which retails for $70. Both feature a similar design, but where the Omni differs most is that its block is much, much larger. Is that alone what almost triples the price? Any way you look at it, the Omni is expensive, almost unreasonably so.
Making matters worse, the installation process isn’t going to be silky smooth for everyone. In the installation manual for the Omni, the test PC shown in the photos uses a small stock Intel CPU cooler, and I can begin to understand why. As soon as an enthusiast cooler is introduced, there’s just no room for the radiator to be installed as it is. We had to unhinge the pump from our unit, and that’s clearly not desirable.
So I ask again, just who is this cooler for? It’s not for enthusiasts, because thanks to the cost of the Omni, it’s more sensible to purchase a larger GPU to begin with, or a second GPU to enable SLI mode. It’s not for overclockers, because of the same reasons. The people it’s for are those who want the best temperatures possible, and also the lowest noise. For that reason, it’d be a great HTPC or portable gaming PC cooler, providing it’ll fit in with the pump being strapped to the radiator.
In the end, CoolIT’s Omni cooler is impressive, but its price will keep it out of the hands of most people.
Have a comment you wish to make on this article? Recommendations? Criticism? Feel free to head over to our related thread and put your words to our virtual paper! There is no requirement to register in order to respond to these threads, but it sure doesn’t hurt!
Copyright © 2005-2018 Techgage Networks Inc. - All Rights Reserved.