You are here

NZXT Kraken G10 Graphics Adapter

NZXT Kraken G10 GPU Adapter
Product
NZXT Kraken G10

GPU Adapter
Manufacturer
NZXT
Street Price
US$27~35

Our recent NZXT Kraken X31,
X41
and X61 AIO CPU water
cooler reviews triggered discussion about an adapter that allows them to be
used with hot video cards. (See the forum threads here
and here.)
Such devices have been adopted by inveterate DIY gaming enthusiasts, who have
been machining brackets that allow the pump/heatblock of all-in-one CPU water
coolers to be attached to GPUs, mostly for better overclocking and cooling.

This section at Overclock.net
is dedicated to DIY entrepreneurs selling universal as well as custom fitted
GPU adapters for AIO CPU coolers. There is interest among other commercial heatsink
makers, as well. Silverstone showed a version at Computex in June this year,
but were not sure if they would go ahead with it. Arctic, a long established
aftermarket cooler brand, has been offering a complete AIO water cooler specifically
for GPUs since 2012; the name "Accelero Hybrid" refers to the existence
of a fan on the card itself, similar to the G10.

The question among the SPCR audiences is not whether such adaptations improve
GPU overclocking or cooling, but whether the additional headroom of a water
cooling system can be harnessed to lower noise. Reducing the noise of powerful
gaming PCs is an oft-recurring topic in the SPCR forums, so the existence of
the G10 CPU adapter bracket, and our positive review findings about the cooling
prowess of the Kraken X41 and X61 — really the first AIO coolers to match
or surpass the best air-only heatsinks by our requirements — were reasons
for some excitement.

The NZXT Kraken G10 has been available since late last year, but we became
aware of it only after our recent Kraken X31, X41 and X61 cooler reviews. The
G10 is said to be a copy of a bracket created and sold (for a while) by an Overclock.net
member called DWood. Anantech,
Legit
Reviews
and Puget
Systems
were among those who posted reviews of the G10 focusing mostly on
cooling performance, but none tackled the question of how it works for low noise
computing. Naturally, SPCR had to ask that question, and NZXT provided a sample
for us to try.



The box shows how the G10 looks mounted on a graphics card.



G10 box contents.

One obvious benefit of this bracket is the added fan, which provides cooling
airflow for VGA card components other than the GPU, such as the VRM and memory
chips, all of which tend to get pretty hot without some kind of cooling.

NZXT Kraken G10: Specifications

(from the product
web page
)
Model Number RL-KRG10-W1 (White)

RL-KRG10-B1 (Black)

RL-KRG10-R1 (Red)

RL-KRG10-U1 (Blue)
Compatibility Nvidia : GTX 980, 970, 780 Ti, 780, 770,
760, Titan, 680, 670, 660Ti, 660, 580, 570, 560Ti, 560, 560SE

AMD : R9 290X*, 290*, 280X**, 280**, 270X, 270 HD7970**, 7950**, 7870, 7850,
6970, 6950, 6870, 6850, 6790, 6770, 5870, 5850, 5830

The compatibility list is based on AMD Radeon and NVIDIA's reference board
layout, screw spacing and die height only. Please check the height restriction
before purchase.

* Some variation in die thickness and height may cause issues with screws
being too short to install on some AMD 290-based cards.

* Do NOT force the G10 to fit if your card is too thick, contact support
for a revised screw set.

** Some variations in AMD die height may require the use of a shim.
Compatibility NZXT : Kraken X61, Kraken X41, Kraken X31,
Kraken X60, Kraken X40

Corsair : H105, H110, H90, H75, H55 , H50 (CW-9060006-WW only)

Antec : KUHLER H2O 920V4, KUHLER H2O 620V4, KUHLER H2O 920, KUHLER H2O 620


Thermaltake : Water 3.0 Extreme, Water 3.0 Pro, Water 3.0 Performer, Water
2.0 Extreme, Water 2.0 Pro, Water 2.0 Performer

Zalman : LQ-320, LQ-315, LQ-310
Warranty 2 Years
Fan
Dimensions 92 (W) x 92 (H) x 25 (D) mm
Speed 1500 RPM ± 10%
Bearing Sleeve
Voltage/Current 12V DC / 0.18A
Connector 3-Pin

Examine the specifications, and you discover that the G10 is meant to be a
universal device. Not only is it meant to fit a wide variety of graphics cards,
it is compatible with AIO coolers from Corsair, Antec, Thermaltake, Zalman and
NZXT, of course. The commonality of all of these AIO water coolers is that they
are made by Asetek, and they all use the same bayonet-mount, short barrel-style
pump/heatblock.

The video card compatibility list also includes three caveats at the end. They're
easy to miss:

* Some variation in die thickness and height may cause issues with screws
being too short to install on some AMD 290-based cards.

* Do NOT force the G10 to fit if your card is too thick, contact support for
a revised screw set.

** Some variations in AMD die height may require the use of a shim.

So... These are cautionary statements. The G10 is clearly a product trying
to be a unversal matchmaker for two classes of gear: Video cards and AIO water
coolers not designed expressly to fit video cards. The user may have to be creative
to make things work just right. Given those cautions, I would hesitate to recommend
the G10 to anyone not at least somewhat experienced with physically handling
PC hardware.

PHYSICAL DETAILS & ASSEMBLY

The core of the G10 is a metal bracket that fits the AIO pump/heatblock
and an 80mm fan in place of the stock cooler on a graphics card. Three sets
of mounting holes are provided around the pump, to fit different graphics cards.
The frame is painted, and NZXT does not specify its composition. We checked
and confirmed that it is magnetic, so it is some type of steel. It is only 1.8mm
thick, and at its thinnest (around the mounting hole for the pump), just 2.2mm
wide. That's pretty flimsy for the mounting system which can be subjected to
fairly high pressure and tension. It is pretty easy to bend by hand. Compared
to the hefty, robust mechanisms used in the best CPU coolers, it's almost laughable.



The frame at its narrowest point around the pump is 1.8 x 2.2 mm. The
foam blocks on the corners are meant to keep the frame stable against
the video card. They come unattached so you can place them where they
work best on your video card.

The fan itself is an excellent idea, as the VRM heatsinks on many video cards
are integrated into the stock cooler; when the stock cooler is removed, additional
airflow is probably necessary to prevent the VRMs from overheating. The 92 x
25 mm fan is a seven-blade design with good geometry for low tonality, and spec'd
at 1500 RPM, a modest enough speed to hope for decently low noise.



A 92mm fan, 1500 RPM & 0.18A.



The pump/heatblock is clamped between the GPU and the G10; it actually
cannot be mounted on the G10 without a video card. The plastic tabs around
the pump take the pressure of this clamping action.



This is the back plate, with thin bolts mounted. Note the multiple holes
for different video cards. The center pad is fairly dense, which means
a lot of compression is needed before the nuts make firm contact with
the back of the video card. The highlighted end of the right front bolt
is not threaded. This glitch held up our review for a couple of weeks
until a replacement set of mounting hardware arrived. To be safe, NZXT
sent not only a spare mounting set, but another G10 sample.



The new mounting bolts are longer, as are the thumbscrews that attach
on the other side.

INSTALLATION

The most critical aspect of heatsink installation is firm, even
contact between the base and the die for efficient heat conduction. Ideally
it should also be a simple procedure with the user having to handle as few pieces
of hardware as possible. We chose the NZXT Kraken X41 to try on the G10.
This cooler was substantally better than the X31, due to its large radiator
and its 140mm fan. The HIS HD 5870 iCooler
V Turbo
we used for GPU cooler testing in the past was hauled out yet
again. We measured an estimated 215W DC power draw from this card at full power
load, which is in the ballpark for current high power single GPU cards.



The 215W
HIS HD 5870 iCooler
V Turbo
was hauled out yet again

The bare GPU on this card is somewhat protected by a ridged perimeter, which
looks like a raised frame, and limits the amount of angled or askew pressure
that can be applied to the die while the heatblock is being screwed down. Those
of us who recall the days of bare CPUs, before the advent of heatspreaders,
know that such askew pressure was the chief cause of many accidental CPU deaths.



Lining up the mounting holes to the bolts.

INSTALLATION CHALLENGES

My earlier comments about the relatively lightweight quality of the G10 bracket
were borne out during installation. The metal around the pump is thin and flexible
enough that it begins bending almost as soon as you start tightening the knurled
nut that fits atop the threaded bolt. It bent slightly on the other side as
well, but not nearly as much as there is much more metal, and thus less flex.
This was disconcerting, to say the least, and the feeling of lack of control
was exacerbated by the fact that there are no unambiguous mechanical cues on
when the screws are tight enough. Unlike the best bolt-through heatsink retention
mechanisms, the G10's 4-bolt mounting does not have "preset" maximum
tighteness points or preloaded captive springs that screw down to a certain
point and no further. I suspect these bolts are threaded far enough that it
may be possible to overtighten to the point where the die could crack, especially
if the four nuts are not evenly tightened. The foam pad on the backplate adds
to the risk factor as well, because it can compress unevenly, and allow askew
pressure on the GPU.



The G10 bracket is highlighted to show the bend on the left side.

I proceeded very slowly, eyeballing the mating between the pump and GPU often
to ensure they remained as parallel as possible while turning the nuts systematically
and incrementally. There is no way to know for sure whether the pressure is
even across the GPU surface. All you can do is hope it is even, tight enough,
but not too tight. Worst case scenario: If you overtighten, you might damage
the GPU; if you don't tighten enough, the contact might not be good enough and
your GPU could burn as soon as it powers up.

To be fair, other tech sites have reviewed the G10, and none have reported
a GPU mishap. Still, I am not convinced that such mishaps cannot happen, and
I believe the mounting system could be made far more foolproof. The desire for
a more robust and foolproof mounting system is not unreasonable, considering
that the G10 could be used with video cards that cost many hundreds of dollars.



Kraken X41 mounted on the HIS HD 5870 card via the G10, lying atop the
side panel of a Fractal Design Define R3.

A Fractal Design Define R3 stripped of 3.5" drive cages to make
more room has been our graphics card and cooler test platform case for a couple
of years. It is not ideal for housing large watercooling radiators. While the
radiator of the X41 is not that big, it's quite wide, and therein lies the problem.
The Define R3 is equipped with 120mm vents and fans; they are too small to comfortably
fit the X41 radiator.

A choice had to be made: Find a way to make the X41 radiator fit in the R3
for this test, or upgrade to a new larger case. The latter option would make
all of our previous test data not directly comparable to future test data; the
former would be a PITA and be a somewhat compromised installation for the G10/X41.
It was a tossup, and in the end, the R3 was retained for this last test.

DEAD GRAPHICS CARD

Before the test could proceed, however, another obstacle loomed: The HIS HD
5870 iCooler V Turbo video card refused to produce a picture on our monitors.
I tried two different monitors, with DVI, HDMI and old VGA connectors. I also
tried installing the card on a different system. I tried, finally, with the
original stock cooler. The card remained resolutely dead.

When the GPU was examined, there was no apparent evidence of physical damage;
the die was not chipped. It's not clear why the HD 5870 card died. It has been
handled around the lab for four years; it may have lasted longer than we could
reasonably expect, given the amount of handling reference test gear is subject
to.

So back to the drawing board.

A known card was needed, one already tested thoroughly by SPCR, with thermal,
acoustic and power qualities well documented. Ideally it would be a high power
card, like the 215W HIS. The closest thing still on hand was the ASUS
GeForce GTX 680 DirectCU II OC
, a top-end card going back two generations
with a measured maximum power of 203W.

Considering the bent frame of the G10, I decided to start fresh with a new,
second sample. During installation to the GTX 680, this sample bent in exactly
the same wasy as the first. The bending was no fluke. Installation was complete
successfully, but this was known only when the monitor flickered on with the
boot up info. Whew!



G10 going on the ASUS GTX 680.



Kraken G10/X41-equipped ASUS GTX 680 installed in the test system. The
CPU cooler had to be rotated in order for the X41 radiator and fan to
fit on the top vents. It was an extremely tight fit due to limited space
between the top panel and the top edge of the motherboard. The RAM sticks
on the board prevented the radiator from being centered on the vents,
so it ended up being secured only with two screws. As explained earlier
in the text, this was a compromise installation.





Only two screws held up the radiator, but wedging a bit of foam on the
other side prevented vibrations. The vent is smaller than the radiator;
the dividing strip in the middle imposes further airflow impedance.

TEST METHODOLOGY

Testing the G10 is considerably more complex than testing other GPU heatsinks.

  • First of all, it's not a heatsink, but a device which enables any number
    of heatsinks to be used on video cards; 27 is the number of officially compatible
    AIO water coolers.
  • Secondly, when a water cooling system is tested, for acoustics, both the
    pump and the fan(s) on the radiator come into play.

Despite the fact that this review is of the Kraken G10, the end results are
intrinsically dependent the actual AIO water cooler that is employed, in this
case, the NZXT Kraken X41. It should be obvious, but to be perfectly clear,
if you use a different AIO CPU water cooler, you will get different results.

Test Platform

Measurement and Analysis Tools

Test Procedure

Our test procedure is an in-system test, designed to determine whether the
cooler is adequate for use in a low-noise system. By adequately cooled,
we mean cooled well enough that no misbehavior related to thermal overload is
exhibited. Thermal misbehavior in a graphics card can show up in a variety of
ways, including:

  • Sudden system shutdown, reboot without warning, or loss of display signal
  • Jaggies and other visual artifacts on the screen.
  • Motion slowing and/or screen freezing.

Any of these misbehaviors are annoying at best and dangerous at worst —
dangerous to the health and lifespan of the graphics card, and sometimes to
the system OS.

Our main test consists of FurMark stability test running in conjunction with
Prime95 to stress both the graphics card and processor simultaneously. This
combination produces more CPU/GPU stress than a typical gaming session. As our
test system has very limited airflow, our results are not indicative of a real-world
situation, but rather a worse-case scenario. If the heatsink in question can
cool the card and its components adequately in this environment it means there
will be some degree of thermal headroom when deployed in a more conventional
situation.

The cooler's fan(s) (if applicable) are connected to a custom external fan
controller and tested at various speeds to represent a good cross-section of
its airflow and noise performance. If the heatsink is passively cooled, we use
different predefined system fan speeds to determine the effect of system airflow
on cooling performance.

System Acoustics Baseline

There are three fans in our VGA test platform. Two Antec
TrueQuiet 120
fans are for case airflow, one for intake and the other for
exhaust. The Kingwin Lazer
Platinum 1000W power supply is set to ECO mode
, and its fan does not spin
up till ~500W load is reached. The PSU fan has never spun up in any video card
testing we have done so far. With the case fans at their LOW setting, the residual
noise level is so low as to be silent in any normal environment.

GPU Test System Acoustics
Setting
System Fan Speed
High
1130 RPM
26 dBA
Med
820 RPM
18 dBA
Low
580 RPM
12~13 dBA
Note: mic is positioned at a distance of one meter
from the left panel at a 45 degree angle.





Another view of the final test setup. The fan on the radiator actually
became the second fan in a push-pull dual fan setup for the CPU cooler.

The ambient conditions during testing were 10~11 dBA and 22°C.

TEST RESULTS

The presence of a pump forces changes in our testing methodology. Tests were
conducted by manually varying speeds of both the pump and the fans. Some levels
were not tested if the fan was obviously going to drown out the pump (or vice
versa) by a big margin. Total AC power drawn by the system was ~280W.

Test Results: NZXT Kraken G10/X41 on ASUS GTX 680

Maximum Simultaneous GPU & CPU Load
Pump

(RPM)
X41 Fan
G10 Fan
CPU Fan
Case Fans
GPU

°C
CPU

°C
PCH

°C
2900
Max
Max
Max
High
47
36
39
35 dBA
2500
920rpm
500rpm
560rpm
Low
62
36
45
18 dBA
2200
740rpm
500rpm
560rpm
Low
67
39
46
17 dBA
Ambient: 10~11 dBA, 22°C

These are the best full load temperature results ever obtained in our video
test system. To give you some point of reference, here's how the ASUS GTX680
card compares with its stock cooler/fans.

NZXT Kraken G10/X41 vs Stock Cooler/Fans on ASUS
GeForce GTX 680 DirectCU II OC

Maximum Simultaneous GPU & CPU Load
Cooler

fan speed
GPU

°C
CPU

°C
PCH

°C
G10/X41 max
47
36
39
35 dBA
G10/X41 low
62
36
45
18 dBA
G10/X41 lowest
67
39
46
17 dBA
Stock*

max
73
59
55
28 dBA
Stock*

low
83
60
59
17~18 dBA
Ambient: 10~11 dBA, 22°C

*
Ambient temp during Stock GTX 680 testing was 24°C, so stock
cooler temps were adjusted down 2°C.

For SPCR audiences, the relevant comparison is at 17~18 dBA system SPL. Depending
on whether you use the 17 or 18 dBA level of the G10/X41 combo, the GPU temperature
lead is either 21°C or 16°C. The CPU temp difference favors the G10/X41
by a similar margin, and the PCH is also cooler by 13~14°C.

At maximum fan/pump speeds, the Kraken G10/X41 combo is considerably louder
(+7 dBA) but the win margins are even bigger: 27°C for GPU, 23°C
for CPU and 16°C for PCH.

It is an utter demolition by cooling standards. And this in a setup where the
X41 radiator/fan was partially impeded.

Aside from the GPU temperature difference, the impact of the G10/X41 on the
temperatures of the CPU and the PCH — which are not even on the VGA card
— are striking. With a standard VGA cooler, the heat of the GPU increases
the temperature of any nearby components, which include the CPU and PCH. Even
with its fans blowing some of the heat out the slots in the back, the stock
cooler of the ASUS GeForce GTX 680 DirectCU II OC cannot prevent the GPU from
heating up the interior of the case. With the G10/X41 setup, the GPU heat is
pumped away to the radiator, positioned above the motherboard at the outer periphery
of the case, and blown out and away by the radiator fan. Hence, much of the
video card's estimated 200W heat is dissipated outside the case. Also, with
the CPU tower cooler rotated, the fan on the X41 radiator became part of a dual-fan,
push-pull setup, further helping to lower CPU temperature.



The stock cooler of the ASUS GeForce GTX 680 DirectCU II OC does vent
out the back but with nowhere near the efficiency of the G10/X41.

How does the G10/X41 compare against the best air-only aftermarket GPU coolers
like the monstrous 3-fan Acceleros from Arctic Cooling? Unfortunately this question
can't be fully answered at this point. We did review the Arctic
Cooling Accelero Xtreme Plus
some years ago, but the test was conducted
on the higher power HIS Radeon HD 5870 iCooler V Turbo graphics card, the one
found to be dead earlier in the course of this review. The Accelero Xtreme Plus
was examined in an earlier test case, which was somewhat less open than the
Define R3 used here. Examining the
results from that test
, it would probably be a pretty close contest. My
guess is that the AC Accelero (today's version) would nose out the G10/X41 for
noise while the latter would provide slightly better cooling all around. The
main point of contention in the acoustics is the noise of the pump.

ACOUSTICS ANALYSIS

The single most contentious aspect of water cooling for SPCR enthusiasts is
the buzzy, chattery quality of the water pumps. In our experience with AIO water
cooler review samples over several years at SPCR, pumps are always more tonal
and objectionable than good quality 120~140mm axial DC fans. With most air coolers,
if you don't like the noise it makes, you can change the fan. With a water pump,
especially in these AIO closed loop coolers, all you can do is slow it down.
As with a fan, slowing down the pump reduces its cooling ability, but as long
as the water keeps flowing, some cooling is achieved. However, at least subjectively,
the pump noise doesn't seem to change as much as a fan when its speed is changed.
The basic chattery quality always seems to remain.

Shown below is the acoustic spectrum of the X41, measured out in the open in
our review of AIO cooler. The overall SP was low but not without some tonal
elements that were audible, and perhaps objectionable for some people, at close
distance.



The spikes are audible tonal elements.

In our VGA test setup, with the top vents of the case completely open for the
radiator, there was no appreciable or measurable blocking of the most audible
noise source, the chatter of the pump. The measured SPL was quite low, at 17
[email protected]



The cyclical nature of the pump clicking cannot be capture in a 2-dimensional
plot; better to listen to the MP3 audio recording.

It's best to listen to the MP3 recording to get an idea of what the pump sounds
like when all the fans in the system are at nearly inaudible levels.

Our audio recordings are made with a high
resolution, lab quality, digital recording system
inside SPCR's
own 11 dBA ambient anechoic chamber
, then converted to LAME 128kbps
encoded MP3s. We've listened long and hard to ensure there is no audible degradation
from the original WAV files to these MP3s. They represent a quick snapshot of
what we heard during the review.

This recording is intended to give you an idea of how the product sounds in
actual use — one meter is a reasonable typical distance between a computer
or computer component and your ears. The recording starts with a 7 second segment
of room ambient (ie, silence), then the product with settings at various levels.
For the most realistic results, set the volume so that the starting ambient
level is just barely audible, then don't change the volume setting again.

  • NZXT
    Kraken G10/X41
    on ASUS GTX 680 in VGA Test System


    - The first 7 seconds is the ambient of the chamber, ~10 dBA

    - The second 10 seconds is the sound of 2 case fans at min speed, CPU fan
    at 500rpm, G10 fan at 700rpm, X41 fan at 500rpm, and the X41 pump at ~2200rpm,
    ~17 [email protected]

    - The last 10 seconds is the sound with the pump turned off, and the fans
    ramped up slightly in speed to the same 17 [email protected] SPL level.

It is easy to hear the pump, but keep in mind that you've probably got the
volume level set higher than the noise would be in real life, and there is no
other sound in the recording space. The chatter/clatter is plainly audible to
me from a foot above the case. From a meter away, I can hear some noise, but
the chatter recedes to a low level, and sounds something like a slow fan with
rough sounding bearings. From 1.5m away, it's not intrusive at all. Under my
desk, less than a meter away, the pump noise could be bothersome. It depends
how absorbed I was in a game, which would be the main reason to have one of
these devices on a video card.

FAN & PUMP SPEED CONTROL

Untouched in all of the previous discussion is the question of how the fans
and pump in a G10 + AIO water cooler system would be controlled. During the
testing, everything was set manually, for the sake of complete control. This
is not how the typical end user would want to control the cooling of their video
card. Stock video card coolers have fans that are thermally controlled by a
chip on the card, with some user control of the fan curve via provided utilities.
All stock VGA cooler fans ramp up to a plainly audible speed when pushed hard;
many are downright noisy. The 28 [email protected] maximum SPL of the ASUS GeForce GTX
680 DirectCU II OC is one of the lowest for stock VGA coolers; it is still a
level SPCR would not consider acceptable for any significant length of time.

There is no easy way for the G10 + AIO cooler fan(s) and pump to use the fan
controller in the video cards, so the options are motherboard fan header controls,
and/or whatever software is provided with the AIO water cooler. There is probably
no way to use the GPU thermal sensor as a control reference for G10 + AIO water
cooler pump/fan control. Most current motherboards have fairly sophisticated
fan control utilities in the BIOS, and even more sophisticated utilities for
Windows. It's still possible that some canny combination of motherboard fan
head controls and the fan/pump control utilities provided by the AIO cooler
maker will work to provide dynamic fan and pump management that is both very
quiet and very cool. Whatever options you choose, it can get complex, and you'd
be well-advised to think through the cooling/noise for the whole system, and
experiment to make sure things work as you intend.

One viable option with an AIO as capable as the Kraken X41 is to set up a fixed
speed fan/pump system that provides safe temperatures under high load... and
simply turn the pump off when not gaming or doing any high GPU-demand activities.
I tried this with the Kraken X41 on our ~200W GPU. When the pump was switched
off while the system was at 17 [email protected], with just the fans on, the noise level
dropped to a very smooth 14~15 [email protected], quiet enough to be perfectly acceptable
even to me. With the pump kept off for about an hour, GPU temperature remained
mostly under 50°C with modest Windows activities like web surfing and HD
video play. Yes, even with the pump turned off, there is cooling, via water
conduction, and from the 92mm fan on the G10 bracket. With modern GPUs that
draw little power until 3D processing is demanded, this seems a perfectly viable
option. The audio amplitude reduction with the pump off is small but quite perceptible
because of the sound quality difference (as you will have heard in the MP3 recording
on the previous page). In any case, most gamers hardly notice a little more
noise during intense gaming, especially over the sound effects.

FINAL THOUGHTS

Getting back to the actual product under review, the NZXT Kraken G10 does provide
excellent cooling with surprisingly low noise (but not without some effort)
when combined with a high performance AIO CPU cooler. Despite reservations about
the quality of the fittings, mismanufactured parts, and the sturdiness of the
bracket itself, the G10 worked as intended. Perhaps at just $30 and with such
broad compatibility, we cannot expect flawless precision.

Keep in mind that our test involved only one 200W video card, and although
the latest nVidia GTX 980 and 970 GPUs are rated for just 165W TDP, single GPU
cards approaching 300W peak still abound. We cannot comment on how well a G10
+ AIO water cooler would work with such hot GPUs, nor the effect on cards where
the VRMs become laid bare when the stock cooler is removed. Puget
Systems'
take on the G10 is well worth reading for those who are seeking
to cool such GPU beasts.

The G10 is not a good VGA cooling solution for any system in a case that doesn't
have at least decent support for water cooling. You need space to mount a 120mm
radiator & fan, which is the minimum size of radiator used in AIO water
coolers. They combine to be at least 2" thick, which is a lot to mount
at a vent. The most practical position is usually the top panel vent, which
unfortunately allows a direct noise path to the seated user. Putting it at the
front panel, ideally, you'd want the fan blowing out to get the heat out of
the case. This means the back panel fan might be better as an intake. If there
is much in the way of grills and dust filters in front, the impedance will hurt
both cooling and acoustics. Having the front vent be wide open also would hurt
acoustics. A pack panel position for the radiator and fan would be ideal, thermally
and acoustically, but it's the most difficult position, because both the CPU
and the VGA card impinge into that space. The length of the water hoses could
also come into play, and those hoses could also impinge on the CPU cooler.

A relevant question is whether the product makes sense from a cost point of
view. The lowest priced AIO coolers start at around $50~60 before taxes or shipping.
Add the $30 for the G10, and we're at $80~90. Such a setup would probably be
better than a similarly priced air-only cooler for VGA cooling as well as overall
case cooling (because of the water cooling system's ability to move the heat
to the perimeter of the case where it can be efficiently evacuated). But it
is more complex, and because of the pump, always likely to be a bit more audible
even if the fans can be turned way down.

From the silent computing point of view, AIO water coolers are double-edge
solutions for CPU cooling. They promise higher ultimate cooling capability,
but at the cost of the pump noise, which always seems worse in quality than
a good fan. This assessment changes for VGA cooling because stock coolers are
far noisier at high load than the G10 + a decent AIO cooler — at least
if it performs as well as the X41. But monster aftermarket VGA coolers may offer
better acoustics and perhaps nearly as good cooling. One last thought is that
because the heatspreader (or heatsink base) makes direct contact with the GPU
die rather than a heatspreader as on a CPU, the relatively mediocre performance
we've documented with most AIO CPU water coolers may be irrelevant for VGA cooling.
To put it another way, one of the AIO coolers that didn't fare so well in our
CPU cooling test might shine on a GPU due to the direct die contact.

Our thanks to NZXT
for the Kraken G10 Graphics Adapter samples.

* * *

SPCR Articles of Related Interest:

Recommended Heatsinks
Enermax Liqtech 120X AIO Liquid CPU Cooler
Sub-$20 CPU Coolers: A Reader's Roundup
LEPA LV12 Direct-Touch Heatsink
NZXT Kraken X31 & X41 Liquid CPU Coolers
Noctua NH-D15: Update to an Icon

* * *

Discuss
this article in the SPCR forums.

Google

www SPCR