Gigabyte G-Power 2 Pro CPU cooler

Gigabyte’s top CPU cooler model has a different slant, much like their earlier Volar that we already reviewed. Five fat, long heatpipes, a 120mm fan, copper base, and a huge curved array of thin fins: Is it a formula for quiet cooling?

Dec 12, 2007 by Mike Chin

Gigabyte is unusual among Taiwanese motherboard companies in that they don’t actually do any OEM business at all, or so I was told by a key marketing person during the Computex computer trade show in Taipei last June. If this is true, it means that they rely solely on the retail market, which is naturally dominated by DIY enthusiasts, who are probably as fickle as any group of consumers. Perhaps as a response to this dependence, Gigabyte has diversified their range tremendously in the last few years, and it now rivals some of the biggest brands in the business.

Gigabyte’s thermal solutions range takes up a major subsection on their global website, and includes many cases, air and water cooling products for both VGA and CPU, notebook coolers, and power supplies. The G-Power 2 Pro is the second CPU cooler from Gigabyte to be introduced to our torture chamber.

Our G-Power 2 Pro sample came in a huge retail box with full color graphics that promised a cache of eco-unfriendly packaging materials. The promise was fulfilled.

Big box, full of plastics and….

…a big closed-cell foam throne for the heatsink.
All the mounting hardware was in the plastic bag on the right.

The G-Power 2 Pro is another odd-angled heatpipe-based design, like the Volar we examined a few weeks ago. It’s considerably bigger, however, and has many other differences. Judging by its position on the Gigabyte Air Cooling Solutions web page, the GP2P is their top model currently.

PHYSICAL DETAILS

The G-Power 2 Pro combines a polished (and nickel-plated) copper base, five thick heatpipes, and a big 9-bladed fan in a plastic shroud that blows into lots of fins. Even from the first photo, it’s clear that these are thin fins, not the aluminum extrusion used in the Gigabyte Volar heatsink reviewed previously. The fan has a 3-pin header instead of the 4-pin PWM header that’s showing up on many integrated fan coolers these days.

Lots of chrome on this big heatsink.

A variant of the 5V – 12V splitter with 3-pin fan outputs introduced by Zalman years ago.

The two-sided, full color, 8-language instruction sheet.

The photo below shows that the G-Power 2 Pro is angled, like the Volar. The fan does not blow straight down, not does it blow parallel to the motherboard, but rather, somewhere in between. The point is for the airflow to help cool motherboard components around the CPU and be effectively evacuated by the back panel case fan. Just from a visual assessment, there’s little doubt that these aims would be achieved as long as the fan is generating enough airflow in a typical case. The curve of the heatpipes as they rise up from the base gives the cooler a unique shape. (If you could imagine it without the chrome plastic fan shroud, it would resemble some kind of futuristic lounge chair… designed to inflict pain with cutting edges?) You also get a sense of just how huge this cooler is.

Shown here with user-friendly, locking mounting clip for AMD CPUs.

This view of the cooler laid face down on its fan may give a better sense of its size. The spacing of the fins is fairly tight, but the distance the air has to flow between the fins is fairly short, perhaps an inch, so it might not pose too much impedance for a slow fan. Because of the curvature, however, the spacing between the fins is slightly tighter on the fan side than on the visible side. Also because of the curvature, the fan does not sit flush against the fins. There is a gap of at least 2 cm at the center between the fan and the fins. We can conjecture that especially at high speed, the combination of this large gap and the tight spacing on that side of the bank of fins could cause some serious backpressure, which is not good for noise or cooling.

Note the even spacing of the heatpipes.

The image below shows the 775 socket mounting strips screwed on the base. There are two steel bars, each with two short threaded shafts on the ends. You’ll see how these work in the next section on installation.

Set up for socket 775. The base is shiny smooth and flat.

INSTALLATION

It’s impossible to install the G-Power 2 Pro without having full access to both sides of a socket 775 motherboard. In fact, it’s best if you have a helper, as the procedure is quite awkward. First, the board must be set on its edge as in the photo below. The heatsink’s four threaded shafts must be positiond so they go through the board’s mounting holes and are accessible from the PCB side of the board. Pressure must be appiled via the heatsink against the board to keep it in place. The bottom mounting plate must be correctly positioned on the other side, and then four capped nuts must be screwed on each of the four threaded shafts with a small phillips-head screwdriver. It’s actually even more laborious than it sounds because:

• The threaded shafts are very short.
• The four capped nuts are round and smooth; there’s nothing to grip.
• The screw slot on each capped nut is extremely small.

Balancing act requiring two very agile hands; a third preferred to keep the motherboard stable on its edge.

It was a struggle to get it on.

Another approach to installation would have been to position the motherboard upside down, pressing against the heatsink, and somehow keeping it balanced while at least two screws were fastened. In any case, the amount of motherboard handling required during installation is highly undesirable. Such handling is a quick way to end up with a damaged and useless motherboard. We’ve experienced this before.

The installation would have been far less awkward if the securing screws or nuts were accessed from the top of the board rather than under. The design of the fins is such that there would have been plenty of room under them to allow large thumbscrews or similar to be used from the top side.

In contrast, the lever with locking clip using the stock retention bracket for AMD socket installation should be a 30-second job. We didn’t install it on an AMD board, but we’ve dealt with enough similar mounting levers to know.

Gigabyte G-Power 2 Pro ready for testing.

CAUTION: One final comment about installation is that the sheer size and backward slant of this cooler may cause interference with the back panel fan on some cases, depending on exactly where the CPU socket on your motherboard is located. There’s also a potential problem with the width; it hung over the top edge of our test board a bit; it may be enough in many cases to interfere with the power supply on that side.

TESTING

Testing was done according to our
unique heatsink testing methodology. A
summary of the components, tools, and procedures follows below.

Key Components in Heatsink Test Platform:

• Intel Pentium D 950
  Presler core. Rated for 130W TDP. Under our test load, it draws 78W, which includes the efficiency losses
  in the VRMs.
• ASUS P5LD2-VM
  motherboard. A microATX board with integrated graphics and plenty
  of room around the CPU socket.
• Hitachi
  Deskstar 7K80 80GB SATA hard drive.
• 1 GB stick of Corsair XMS2
  DDR2 memory.
• FSP
  Zen 300W fanless power supply.
• Arctic Silver
  Lumière: Special fast-curing thermal interface material,
  designed specifically for test labs.
• Our standard 120mm Nexus reference fan could not be used due to the integrated stock fan.

Test Tools

• Seasonic Power Angel
  for measuring AC power at the wall to ensure that the heat output
  remains consistent.
• Custom-built, four-channel variable-speed fan
  controller, used to regulate the fan speed during the test.
• Bruel & Kjaer (B&K) model 2203
  Sound Level Meter. Used to accurately measure noise down to
  20 dBA and below.
• Various other tools for testing fans, as documented in our
  standard fan testing methodology.

Software Tools

• SpeedFan
  4.33, used to monitor the on-chip thermal sensor. This sensor is not
  calibrated, so results are not universally applicable, but they should be
  comparable with the other tests we’ve done on this test bed. The current test
  system was put into service in January 2007.
• CPUBurn P6,
  used to stress the CPU heavily, generating more heat that most
  realistic loads. Two instances are used to ensure that both cores are
  stressed.
• Throttlewatch 2.01,
  used to monitor the throttling feature of the CPU to determine when
  overheating occurs.

Two iterations of CPUBurn were run simultaneously was used to stress both cores of the processor, and the
graph function in SpeedFan was used to make sure that the load temperature was
stable for at least ten minutes. All HSF are tested at four voltages: 5V,
7V, 9V, and 12V, representing a full cross-section of the fan’s airflow and
noise performance. The ambient conditions during testing were 18 dBA and 20°C.

TEST RESULTS

The cooling results were excellent. A temperature rise of just 13°C puts this cooler into world championship territory in an overclocker’s arena.

Most silent PC enthusiasts wouldn’t want to put up with the noise for more than a minute or two, however. The Gigabyte G-Power 2 Pro’s noise through most of the fan voltage settings was just terrible. The fan by itself was bad enough; it seemed to have the buzzy quality of a typical ball-bearing fan, and a lot of turbulence. There was an odd variability in the noise as well, as if the fan sped up and slowed down from time to time, perhaps due to the effect of eddy air currents causing back pressure in the structure.

There was also the resonating hum, buzz and honk of the plastic shroud vibrating in complex tones in sympathy with the fan. The resonant qualities of this hard plastic shroud are such that even though the SPL dropped to 29 dBA@1m (below our 30 dBA@1m "quiet threshold") at 9V, the sound quality still made it sound nasty. Holding the frame with our hands changed the sound, but it did not eliminate the complex vibrations and resonances. Even with the fan at 7V, when the measured SPL was down to a respectable 25 dBA@1m, the shroud continued vibrating and adding an audible hum to the overall noise signature. At the 4V setting — which is unrealistic because it’s difficult to get such a low voltage to a fan in a PC — the shroud resonance finally disappeared, but the fan rotation speed was just too low for the performance to hold up.

Comparing the Gigabyte G-Power 2 Pro against other large heatsinks is a bit difficult, because the noise of the Nexus 120 fan (used to test all these heatsinks) begins where the noise of the Gigabyte ends. Still, it’s instructive:

The cooling performance of the Gigabyte at 25 dBA noise level is bested by three of the four competitors at just 20 dBA. At the 20 dBA level, the G-Power 2 Pro can only pass the SI-128; all the others provide better performance. Below that noise level, it comes in last place.

NOISE RECORDINGS IN MP3 FORMAT

Perhaps we’re being hyper-sensitive about the noise of the Gigabyte G-Power 2 Pro. Listen and decide for yourself.

• Gigabyte G-Power 2 Pro: 5V-7V-9V-12V, 5s Ambient between
  levels: One Meter
  Note: For best results, adjust your volume control so
  that the ambient noise at the start of the recording is just barely audible, then leave the volume control alone while you listen to any comparatives.

Comparatives:

• Reference Nexus 120mm fan : 5V-7V-9V-12V, 5s Ambient between
  levels: One Meter,
  One Foot
• Scythe Infinity: 5V-7V-9V-12V, 5s Ambient between levels:
  One Meter, One
  Foot
• Zalman CNPS8700 LED: 5V-7V-9V-12V, 5s Ambient between levels:

  One Meter

• Scythe Mine w/ stock fan: 5V-7V-9V-12V, 5s Ambient between levels:
  One Meter, One
  Foot
• Thermaltake Big Typhoon: 5V-7V-9V-12V, 5s Ambient between levels:
  One Meter, One
  Foot

TIME TO STRIP… THE FAN & SHROUD

It seemed the only thing to do. There wasn’t much more to be done in the way of testing, and we’d at least learn more about the shroud and the fan.

We began by unscrewing the fan. It’s attached to the shroud with three screws that are just accessible between the fan blades. This did not release the fan because its power lead was attached to some other part.

Then came two screws at the bottom.

It was difficult to find the top screws for the shroud/frame, but they were found in the end, hidden under the logo tag glued over them at the top.

We discovered that the fan is an Everflow T121225SM, rated for 0.25A at 12V. The frame, as suspected, is made of a brittle-hard plastic, and the struts to which the fan was attached are quite easy to flex.

Here, finally, is the bare heatsink.

The first thing to do was check the specifications of the Everflow T121225SM fan. The closest match in the Everflow on-line catalog was the R121225_M. The M stands for medium speed, which they specify as 1800 RPM. We measured higher (2200 RPM) but this could well be within product variance. The S before the M stands for sleeve bearing — we were wrong when we wrote that it sounded like a typical ball bearing fan; that could have been an effect of the plastic vibration noise. Finally, the SPL is given as 34 dBA. The only other Medium speed (1800 RPM) 9-blade 120x25mm fan in Everflow’s catalog was the LED version of the same fan, with an SPL rating of 34.7 dBA. Note that the maximum noise in Gigabyte’s specs is given as 23 dBA.

The fan was re-attached to the frame, then listened to on its own without any impedance nearby. It was also measured again.

The numbers are plain enough: On its own without impedance, the fan is much quieter. What the numbers don’t tell is how much smoother it sounds. Much of the harsh, unpleasant, tonal qualities heard originally (with the fan mounted in the shroud on the heatsink) were just gone. Listened at very close distance, the fan still had some of the buzzy quality that’s so prominent in the G-Power 2 Pro, but it was low enough in level as to be inaudible from a meter away. Interestingly, all this was still with the fan attached to the shroud.

Our hypothesis, then, is that it’s not just the shroud vibrating in the G-Power 2 Pro, it’s the shroud and the metal structure of the heatsink itself. Dozens of thin aluminum fins, attached to the plastic shroud which easily passes on the vibration of the fan, all vibrating together — that is what’s responsible for the sound we hear in the G-Power 2 Pro.

A SIMPLE MOD & A HOME IN SPCR’S GRAPHICS CARD TEST RIG

Now that the Gigabyte G-Power 2 Pro was stripped to its substantial bones, was there a way to mount a standard 120mm fan on it? Not without some modification. But why not? We knew we didn’t want to use this heatsink in the lab in stock form; perhaps a better fan without the shroud could make it usable in one of our silent test rigs. The graphics card test rig’s CPU cooling could use improvement…

The top and bottom mounting tabs for the shroud are attached by screws. They were removed. Then some leftover stiff steel wire clips for fans from Scythe and Thermalright heatsinks were cut and bent so that a fan could be clamped against the heatsink. It wasn’t difficult to do, and in the end, a fan was securely mounted, as shown in the photo below.

Stripped and modded with a better fan.

Which fan did we choose? It’s a Scythe SlipStream 500, one of a huge batch of 120mm fans awaiting testing in the lab. It’s called 500 because it spins at 500 RPM, but that’s all the speed we needed. It was mounted back on the test platform with a bit more cursing, and we obtained the following CPU stress test result:

This test leaves no doubt in our minds that without the shroud, and with a high quality, quiet fan, the G-Power 2 Pro would perform on par with the best giant CPU coolers we recommend. But then it would be a different product altogether. For one, it would be cheaper without the plastics, and two, you’d have free reign about which fans to use.

The modded G-Power 2 Pro sample did end up in our graphics card platform, which has undergone other substantial changes, enough so that we are writing up a new testing methodology / platform article about it. CPU cooling has definitely improved with the G-Power 2 Pro. The other option we were considering was our beat-up, original Ninja, but the modded Gigabyte is working out about the same as we expected the Ninja to. More on that in our new graphics card testing methodology article soon.

CONCLUSIONS

The Gigabyte G-Power 2 Pro is a difficult product. On the one hand, it had the potential to be a great cooler. The basic design concept is great: Nicely finished copper base, five long, thicker-than-usual heatpipes running through a big array of thin aluminum fins angled so that the air from the fan flows both onto the VRMs and towards the exhaust fan in a typical case. The problems are in the execution, specifically, the plastic shroud that secures the fan to the heatsink proper, and the very awkward installation system for socket 775.

The fan by itself turned out to be not that bad, even when still attached to the plastic shroud. But the combination of fan, shroud and heatsink produces a tonal noise that quite annoying at most fan speeds. When the fan speed is slowed enough to make the noise palatable (to us), the cooling performance is only mediocre. It’s true that there are lots of less aurally sensitive people who will not find the noise a bother, especially when they’re shredding their enemies with any number of weapons at >60 FPS. For them, the GP2P might be a reasonable option. But then, the installation procedure will bother just about anyone, though, and there will be questions about whether this big awkward heatsink will actually fit in their system, even if it’s in a big gaming case, as techPowerUp! discovered.

Especially at the high anticipated retail price of US$69, it’s difficult to recommend the Gigabyte G-Power 2 Pro. There are many better options at better prices for quiet, high performance CPU cooling.

As with the Gigabyte Volar (see the Editor’s Postscript at the bottom of the linked page), we can’t help but question the claimed acoustics. Where in the world did the "23 / 16 dBA" spec (at 12V and 5V) for the Gigabyte G-Power 2 Pro come from? Actually, that’s not quite fair; the 16 dBA claim is believable, as our ambient noise is too high for measurement accuracy down to such low levels. We know that at 5V, the noise was below 18~19 dBA@1m, so OK, it’s possible that at 5V, the thing measured 16 dBA in Gigabyte’s very nice anechoic chamber. But 23 dBA? Sorry, that’s just totally wrong, twice in a row now. Once more, and we’ll have to believe that overstating of their cooling products’ quiet characteristics is standard Gigabyte policy. It’s not unusual, of course, as the industry is full of marketing exaggerations, especially about acoustics these days, but it is disappointing when we know what kind of sophisticated acoustic testing facilities and equipment Gigabyte has at their disposal.

Much thanks to Gigabyte for the G-Power 2 Pro sample.

* * *

Articles of Related Interest

* * *

Comment on this article in our Forums