In fact, there is. Zalman’s ZM80A-HP graphics chip cooler hasn’t been validated for use with NVIDIA’s upcoming GeForce FX, but the ZM80A-HP is a completely passive cooling solution for today’s top commercially available graphics cards. Without moving parts, the ZM80A-HP promises to be impervious to fan failure and completely silent. It looks pretty slick, too.
Is the ZM80A-HP a worthy graphics chip cooler for an enthusiast’s high-end graphics card? Join me as I examine the ZM80A-HP’s impact on system noise levels and its ability to dissipate the heat generated by stock and overclocked graphics cards from ATI and NVIDIA.
Passive cooling solutions are usually pretty simple affairs, but the ZM80A-HP is quite a bit more intricate than even most active cooling systems. What’s in the box?
First, the big stuff. Zalman’s ZM80A-HP uses two massive passive heat sinks to radiate heat away from the graphics chip. A U-shaped copper heat pipe connects the two heat sinks, helping to share the cooling load between the two.
The copper heat pipe gets sandwiched between the passive heat sinks. Zalman includes two different block sets to accommodate mounting hole patterns on GeForce4 and Radeon 9×00-based graphics cards, and users may be able to install the ZM80A-HP on future graphics cards, as well.
Both the mounting blocks and passive heat sinks have grooved channels to sandwich tightly the copper heat pipe, which is long enough to accommodate even NVIDIA’s behemoth GeForce4 Ti 4600.
Zalman includes all sorts of screws, nuts, and washers with the ZM80A-HP in addition to a tube of thermal compound. They even include extra thermal grease and a set of spare screws should a thick, shag carpet ingest one of the tiny screws. As if that weren’t enough, Zalman also throws in a screwdriver. Everything a user would need for installation is right in the box.
Because it ships with two sets of GPU blocks, the ZM80A-HP is compatible with the two different sets of heat sink hole patterns for GeForce4 and Radeon 9×00 cards. Unfortunately, Parhelia owners are out of luck; the ZM80A-HP isn’t compatible with Matrox’s unique board and chip layout.
We can get away with applying a razor-thin layer of thermal compound between a GeForce4 chip and the ZM80A-HP, but a little extra thermal compound must be used with Radeon 9500 and 9700 chips to ensure complete contact with the ZM80A-HP. ATI’s GPU shim is just a hair taller than the actual graphics chip. Incidentally, shim clearance is an issue with any cooler used with ATI’s latest Radeons; this problem isn’t unique to the ZM80A-HP.
Once a correct amount of thermal compound has been smeared over the GPU, installing the heat sink’s GPU blocks takes just minutes. Unlike many cooling solutions which use plastic push-pins, Zalman’s ZM80A-HP blocks are secured with a set of metal screws.
Zalman’s use of metal screws to secure the ZM80A-HP is a nice touch, especially since the locking nut makes it impossible to over-tighten the cooler. Zalman supplies some rubber nuts to ensure the screws and mounting brackets don’t make inadvertent contact with the graphics card, possibly shorting essential connections. At first, I marveled at the ZM80A-HP’s slick retention system, but the tiny metal nut used to secure the rear mounting block is really a pain to install. Trying to adequately tighten this diminutive nut is frustrating since fingers must be used; screwdrivers, wrenches, and even needle-nose pliers don’t work. In the end, I managed to secure the rear mounting block, but not without almost wearing my fingertips raw. Zalman would do well to key this tiny nut for use with a screwdriver or even pliers to make installation easier.
Once the mounting blocks are installed, one need only slide in the heat pipe and secure the massive heat sinks to the front and rear blocks. Zalman’s excellent installation instructions recommend spreading thermal compound between the GPU blocks, heat pipe, and passive heat sinks for more effective cooling. Zalman includes plenty of thermal compound to cover all necessary surfaces.
As you can see, the ZM80A-HP easily overwhelms a Radeon 9700 Pro. As large as the rear passive heat sink looks, it’s not actually making any direct contact with the back of the card:
Rather that sandwich the PCB between the cooling system’s GPU blocks, Zalman suspends the rear mounting block just millimeters above the PCB. The rear mounting block is held off the board’s surface by its mounting bracket and by the assembly’s copper heat pipe. Just in case, Zalman provides a clear sticker for the bottom of the mounting block to avoid any potential metal-on-metal contact that could short the graphics card.
Because the ZM80A-HP relies on passive cooling, it needs more surface area than your average GPU cooler to pull heat away from the graphics chip adequately. All that extra surface area makes the ZM80A-HP massive by today’s standards, but the cooler creates fewer clearance problems than one might expect.
Before I get into clearance issues, it’s important to note that the ZM80A-HP weighs nearly 400 grams, which is quite a bit heavier than the average graphics cooler. An AGP slot retention mechanism is all but essential to ensure that the cooler’s weight doesn’t help unseat the graphics card when a system is moved. With careful transportation, a securely-attached cooler and graphics card should survive multiple LAN party trips, but it’s a good idea to be extra gentle.
Though the ZM80A-HP adds a lot of height to the back of an AGP card, there’s still plenty of clearance for DIMM slots and large north bridge coolers. Even the tallest of motherboard capacitors are well clear of the rear passive heat sink.
Depending on the motherboard, the ZM80A-HP may not even rob users of a PCI slot. As you can see here, there’s plenty of room between the front passive heat sink and the AT7 MAX2’s closest PCI slot. The ZM80A-HP will, however, create clearance problems for the first PCI slot on many motherboards. Small form factor PCs like Shuttle’s XPC series are also incompatible with the ZM80A-HP without extensive case modifications.
All tests were run three times, and their results were averaged, using the following test systems.
|Processor||Intel Pentium 4 2.4B|
|Front-side bus||533MHz (4x133MHz)|
|Motherboard||Abit IT7 MAX2 Version 2.0|
|North bridge||Intel 82845PE (MCH)|
|South bridge||Intel 82801DB (ICH4)|
|Chipset driver||Intel 4.10.1012|
|Memory size||512MB (2 DIMMs)|
|Memory type||CAS 2 PC2700 DDR SDRAM|
|Graphics||Abit Siluro GeForce4 Ti 4200
VisionTek GeForce4 Ti 4600
|Tyan Tachyon G9700 Pro
Crucial Radeon 9700 Pro
|Graphics driver||NVIDIA Detonator 41.09||ATI CATALYST 3.0|
Maxtor 740X-6L 40GB 7200RPM ATA/133 hard drive
|Operating System||Windows XP Professional SP1|
Since the ZM80A-HP is compatible with GPUs from NVIDIA and ATI, I’ve rounded up a handful of cards from each for testing.
All noise testing was done with an Extech 407727 digital sound level meter that can measure sound levels above 40dB. Noise tests were conducted with the sound level meter positioned one inch (not directly in the path of air flow), one foot, and five feet from an open test system with only one low-RPM, 80 mm fan blowing over the processor. At a distance of 10 feet, noise levels were below 40 dB and thus immeasurable with our testing equipment.
We used the following versions of our test applications:
- Unreal Tournament 2003 full retail version
The test systems’ Windows desktop was set at 1024×768 in 32-bit color at a 75Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests. Most of the 3D gaming tests used the high detail image quality settings, with the exception that the resolution was set to 640×480 in 32-bit color.
All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
NVIDIA’s Dustbuster GeForce FX cooling solution has thrust noise levels into the spotlight, but there’s more to sound than just decibels. First, we’ll cover the decibels.
While the fan noise from Tyan’s Tachyon G9700 Pro and ATI’s stock Radeon 9700 Pro cooler get louder as we get closer to the GPU, the ZM80A-HP is incapable of making any noise at all. In fact, the ZM80A-HP-equipped system is silent enough at a distance of five feet to evade our noise testing equipment.
Our comparison coolers for GeForce4 Ti cards are a little more closely matched, but the ZM80A-HP is just as silent here as with our Radeon 9700 Pros.
As I mentioned earlier, there’s more to noise than just decibels. The decibel scale is logarithmic, so the two previous graphs don’t really communicate how loud the different cooling solutions really sound. A more appropriate way to represent noise levels is in terms of wattage per square meter. I’ve crunched the numbers and displayed our decibel results in watts per square meter below.
With watts per square meter, the numbers are a little messy, but we see a truer representation of each cooling system’s “loudness.” Zalman’s ZM80A-HP is still easily the quietest of the lot.
Without any fans or moving parts, it’s no surprise the ZM80A-HP is quieter than active GPU cooling solutions, but how effective is Zalman’s heat sink at cooling the graphics chip? Because overclocking potential is very much dependent on the characteristics of each individual chip, I overclocked Tyan’s Tachyon G9700 Pro and Abit’s Siluro GeForce4 Ti 4200 with their standard coolers and with the ZM80A-HP to measure the passive heat sink’s cooling prowess.
In testing, I was able to overclock our Tachyon G9700 Pro to core and memory clock speeds of 380 and 620MHz, respecitvely, with the card’s standard cooler. The card hit speeds of 370 and 640MHz with the ZM80A-HP. With our sample, the overclocking penalty incurred moving from the Tachyon G9700 Pro’s high-performance stock cooling solution to the ZM80A-HP was only 10MHz, which is a testament to the passive heat sink’s ability to keep things cool. Interestingly enough, our ZM80A-HP-equipped Tachyon G9700 Pro’s memory actually overclocked a little higher with the ZM80A-HP than with the card’s stock cooler. This is likely because the Tachyon’s stock cooler directly connects the GPU and memory chips, radiating heat from the former to the latter.
With our Siluro GeForce4 Ti 4200, I was able to get core and memory clock speeds up to 320 and 620MHz with the card’s stock cooling and 315 and 615MHz with the ZM80A-HP. Unlike with the Tachyon G9700 Pro, the GeForce4 Ti 4200’s stock cooler doesn’t make any direct contact with the card’s memory chips, but since the ZM80A-HP hovers directly over the memory chips, radiating heat may have slightly hindered our memory overclocking. Still, using the ZM80A-HP only forced a 5MHz drop in overclocked core and memory clock speeds; that’s not much of a penalty at all.
So the ZM80A-HP isn’t quite as good for overclockers as active cooling solutions, but how much of an impact does the ZM80A-HP’s slim overclocking disadvantage have on actual game performance?
Not much. While graphics card overclocking can have a definite impact on frame rates in Unreal Tournament 2003, especially with our GeForce4 Ti 4200, the performance impact of the ZM80A-HP’s slight overclocking penalty is negligible.
I have to admit I’m a sucker for passive cooling, not so much because I find ordinary graphics chip coolers to be unbearably loud, but because I’m all for eliminating noise where possible. I’m also all for innovative graphics card cooling solutions, especially those compatible with multiple graphics cards from different card manufacturers using different graphics chips. So, from a purely academic perspective, the ZM80A-HP is great. The cooler is dead silent, free of failure-prone mechanical parts, and compatible with numerous graphics cards. The only thing that annoys me is that tiny nut that can’t be tightened with a screwdriver.
But just how silent can a passive graphics chip cooler make a system?
Well, a closer examination of the ZM80A-HP’s impact on system noise levels reveals a dramatic reduction in sound intensity an inch away from the graphics card, but as we move farther away, even active graphics cooling becomes nearly as silent. That doesn’t mean that there’s no value to passive graphics cooling, but only those with near-silent systems will likely notice much of a difference in real-world noise levels.
I expected the ZM80A-HP to be silent, but what surprised me was how little it affected my ability to overclock my GeForce4 Ti 4200 and Radeon 9700 Pro graphics cards. Overclocking success is generally very dependent on the characteristics of individual graphics chips, but it’s still impressive that the ZM80A-HP allowed overclocking to within just a few MHz of what was possible with active cooling solutions, at least with the cards I used. At just over $30, the ZM80A-HP isn’t cheap, but its price isn’t outrageous. There aren’t many passive graphics chip cooling solutions available today, and there are even fewer appropriate for high-end cards and overclocking. The fact that there’s potential for the cooler to last through multiple graphics card upgrades makes it even more attractive. In the end, the ZM80A-HP is the ultimate graphics cooler for a high-end silent gaming rig.