You are here

SPCR's Updated 2012 Small CPU Heatsink Test Platform

SPCR's Updated 2012 Small CPU Heatsink Test Platform

June 5, 2012 by Lawrence Lee

From time to time we here at SPCR toy with the idea of updating older test
platforms with new hardware despite the extra time and effort required to retest
previously reviewed components. Generally, we avoid the practice unless things
break down or there are overwhelming reasons to enact a change, like if the
hardware is no longer representative of what is used by the majority of the
community.



Old AMD small heatsink test platform.

Our small CPU heatsink testbed is such a platform, consisting of an Athlon
II X4 630 installed on a 785G chipset motherboard. Though there was nothing
physically wrong with this setup and the power draw is still appropriate for
more compact coolers, AMD has been more or less been on the sidelines as the
mini-ITX motherboard form factor rises in popularity. The superb energy efficiency
of Intel's recent processors has resulted in several manufacturers churning
out Intel mini-ITX boards, first for LGA1156 and now for LGA1155, while only
a handful of AMD AM3/FM1 models are available today. With Intel being the overwhelmingly
dominant choice for small motherboards in small cases requiring small heatsinks,
shifting our small CPU heatsink test platform to Sandy or Ivy Bridge was an
easy decision. We are still happy with testing heatsinks in an open test environment
to rule out external factors like case layout and system airflow.



New Intel LGA1155 small heatsink test platform.

For the CPU, we opted for the Intel
Core i5-2400
, a quad core Sandy Bridge chip with a moderate TDP of 95W.
One advantage it has over the X4 630 is its higher throttling temperature; the
processor reduces in frequency at approximately 100°C, about 25°C higher
than the Athlon II on the old platform, allowing us to collect results for less
capable coolers at low fan speeds (even though it is not advisable to run it
like this for long durations). A dual core 65W Core
i3-2100
was considered but as its power draw is considerably lower,
we figured that better heatsinks would be bottlenecked, making it difficult
to determine superior performance.



Screenshot: Intel Desktop Utilities.

For the motherboard, we chose the Intel
DP67BG
which has temperature sensors for the memory and VRM area, often
lacking in most motherboards. This will give us an idea of how well each heatsink
cools the components surrounding the CPU socket. It's also equipped with multiple
controllable fan headers so we can finely adjust the fan speeds of temperamental
PWM fans that don't react well to voltage control.



A multimeter is used to measure the current passing through the EPS12V/AUX12V connector to determine the power pulled from the CPU.

While the Core i5-2400 is a 95W chip, at stock settings, the CPU and motherboard
together have an overall power consumption ~25W lower than our old Athlon II
X4 platform at full CPU load. Tapping the EPS12V/AUX12V +12V line using a shunt
resistor, we found the CPU was only pulling 60~65W (it also draws some power
from the main ATX12V plug as well). This in itself isn't significant, but conditions
inside a SFF case with limited airflow are more thermally demanding than out
in the open, so it was decided that the power draw should be increased through
overclocking/overvolting to compensate.



CPU screenshot: Core i5-2400 at 3.6 GHz, 1.300V.

After trying several different settings, we settled for a Turbo Boost multiplier
of 36x and a CPU voltage of 1.300V. Overclocked, the draw on the AUX12V line
varies between 77W and 83W depending how hot the VRMs get — this depends
on the cooling effect of the CPU cooler under test, of course — while the
entire system consumes 130~136W AC measured at the wall.



Fan controller.

Aside from the hardware changes, the rest of our test setup and methodology
remains mostly unchanged. Fan speeds are adjusted using a custom built external
voltage controller with a multimeter measuring the voltage to ensure precise
control. 3-pin/4-pin fans are connected to an adapter with a separated RPM sensor
header that is connected to the motherboard to report the fan speed, though
we also have a Tachometer
and Stroboscope
at our disposal.

To christen our new test platform, we re-tested four past favorites, the Noctua NH-L12, Scythe Big Shuriken, Scythe Samurai ZZ, and Scythe Kozuti.

TEST METHODOLOGY SUMMARY

Key Components in LGA1155 Heatsink Test Platform:

  • Intel Core i5-2400 Sandy Bridge core, LGA1155, 3.1 GHz, 45nm, 95W TDP, overclocked/volted to 3.6 GHz and 1.300V.
  • Intel DP67BG ATX motherboard.
    P67 chipset.
  • Kingston
    SSDNow V
    30GB 2.5" solid-state drive. Chosen for silence.
  • OCZ Platinum Extreme Low Voltage DDR3 memory. 2 x 2 GB, DDR3-1333 in dual channel.
  • Seasonic X-400 SS-400FL
    400W ATX power supply. Passively cooled.
  • Arctic Silver
    Lumière
    : Special fast-curing thermal interface material, designed
    specifically for test labs.
  • Noctua 140 mm fan (used when possible with heatsinks that fit 140x25mm
    fans)
  • Nexus 120 mm fan (used when possible with heatsinks that fit 120x25mm
    fans)
  • Nexus 92 mm fan (used when possible with heatsinks that fit 92x25mm
    fans)

The systems are silent under test conditions, except for the CPU cooling
fan(s).

Normally, our reference fans are used whenever possible, the measured details
of which are shown below (note these measurements were taken in open air, the
noise levels vary a bit when mounted on heatsinks, especially at higher RPM).

Reference Noctua 140mm fan

Anechoic chamber measurements
Voltage
Speed
12V
28~29 dBA
1250 RPM
9V
21 dBA
990 RPM
7V
15~16 dBA
770 RPM
6V
13 dBA
660 RPM


Reference Nexus 120mm fan

Anechoic chamber measurements
Voltage
Speed
12V
16 dBA
1100 RPM
9V
13 dBA
890 RPM
7V
12 dBA
720 RPM


Reference Nexus 92 mm fan

Anechoic chamber measurements
Voltage
Speed
12V
16 dBA
1470 RPM
9V
12 dBA
1150 RPM

Measurement and Analysis Tools

  • Extech 380803 AC power analyzer / data logger for measuring AC system
    power.
  • Custom-built, four-channel variable DC power supply, used to regulate
    the fan speed during the test.
  • PC-based spectrum analyzer:
    SpectraPlus with ACO Pacific mic and M-Audio digital
    audio interfaces.
  • Anechoic chamber
    with ambient level of 11 dBA or lower.
  • Various other tools for testing fans, as documented in our
    latest fan test roundup
    .
  • SpeedFan,
    used to monitor the on-chip thermal sensors. The sensors are not calibrated,
    so results are not universally applicable. The hottest core reading is used.
  • Prime95,
    used to stress the LGA1366 CPU heavily, generating more heat than most real applications.
    8 instances are used to ensure that all 4 cores (with Hyper-threading) are
    stressed.
  • CPU-Z,
    used to monitor the CPU speed to determine when overheating occurs.
  • Thermometers to measure the air temperature around the test platform
    and near the intake of the heatsink fan.

Noise measurements are made with the fans powered from the lab's variable DC
power supply while the rest of the system was off to ensure that system noise
did not skew the measurements.

Load testing was accomplished using Prime95 to stress the processor, and the
graph function in SpeedFan was used to ensure that the load temperature is stable
for at least ten minutes. The temperature recorded is the highest single core
reading. The stock fans were tested at various voltages to represent a good
cross-section of airflow and noise performance.

The ambient conditions during testing were 10~11 dBA and 21~23°C.

TEST RESULTS

Noctua NH-L12 (Street Price: US$60)



The Noctua NH-L12.

The Noctua NH-L12 is our previous small heatsink champion. Being essentially a compact version of the Noctua NH-C14, the six-heatpipe NH-L12 utilizes the same exemplary backplate mounting system and delivers some serious performance. Equipped with a both a 92 mm and 120 mm fan, it is the heaviest heatsink in our roundup at 680 grams and also the tallest at 93 mm. However, it only needed the single 120 mm model mounted underneath the fin-stack (for a total height of 66 mm) to earn top spot on our old small heatsink test platform.

Noctua NH-L12
Fan Voltage
Fan Speed
°C Rise above Ambient
CPU
VRM
RAM
Both Stock Fans
12V
1680 / 1420 RPM
28~29 dBA
31
14
15
9V
1320 / 1080 RPM
20 dBA
33
17
16
8V
1200 / 960 RPM
17 dBA
34
19
18
7V
1060 / 840 RPM
14~15 dBA
35
21
20
6V
920 / 720 RPM
12 dBA
36
23
21
Stock 92mm Fan (bottom mounted)
12V
1680 RPM
23 dBA
39
23
21
10V
1450 RPM
19~20 dBA
42
28
23
9V
1320 RPM
17 dBA
44
31
24
8V
1200 RPM
15 dBA
47
33
26
7V
1060 RPM
13 dBA
51
38
30
6V
920 RPM
11~12 dBA
57
43
32
Stock 120mm Fan (bottom mounted)
12V
1420 RPM
26 dBA
34
20
19
10V
1190 RPM
20~21 dBA
36
23
19
9V
1080 RPM
18 dBA
37
24
20
8V
960 RPM
15 dBA
38
26
21
7V
840 RPM
13 dBA
39
27
22
6V
720 RPM
11~12 dBA
42
32
24
Reference 120mm Fan (bottom mounted)
12V
1080 RPM
14~15 dBA
34
20
15
9V
880 RPM
12 dBA
37
25
23

The NH-L12 performed superbly once again, particularly in its dual fan configuration
93 mm tall configuration where only 5°C separated 12V and 6V operation.
The stock 92 mm model on its own doesn't do the heatsink justice, performing
substantially worse than bigger fan. As the fan speed and noise is reduced,
the difference grows ever larger, with the bigger fan achieving a 15°C advantage
at 6V. Even more astonishing is our single reference Nexus fan giving the dual
stock fan configuration a run for its money.

Scythe Big Shuriken (Street Price: US$35 for updated version)



The Scythe Big Shuriken.

Of all the coolers in our roundup, the Big Shuriken is arguably the most popular
as it offers decent performance despite its reasonable price and low height
and weight (58 mm, 410 grams). For its size, it is great, but installing it
on Intel motherboards is a serious pain. As its fin-stack sits quite low, there's
barely enough room for an index finger to slip in and depress the pushpins.
[Editor's Note: We can both recall many instances of cursing at nicked
finger while grappling the Big Shuriken... which might explain why the BS2 came
to be
.]

Note: the following results are for the original Big Shuriken which has
been discontinued. An improved replacement, the Big Shuriken 2 Rev.B will be
reviewed in the near future.

Scythe Big Shuriken
Fan Voltage
Fan Speed
°C Rise above Ambient
CPU
VRM
RAM
Stock 120mm Fan
12V
1770 RPM
27 dBA
39
23
15
9V
1380 RPM
19~20 dBA
43
28
19
8V
1110 RPM
14 dBA
46
30
21
7V
820 RPM
11~12 dBA
61
47
31
Reference 120mm Fan
12V
1080 RPM
14~15 dBA
41
25
14
9V
880 RPM
11~12 dBA
43
29
19

The Shuriken produced fairly good performance, at least until the fan speed
dropped past the 1000 RPM mark. Going from 8V to 7V, there was a 2~3 dB reduction
in noise but the CPU temperature shot up by 15°C and VRM cooling suffered
to the tune of 17°C. Our reference Nexus 120 mm fan fared much better, particularly
at the 11~12 dBA level. To be fair, the Nexus has a standard thickness of 25
mm while the stock fan is only 12 mm. The thinner fan cannot push as much pressure,
which has a big bearing on cooling effectiveness.

Scythe Samurai ZZ (Street Price: US$30)



The Scythe Samurai ZZ.

The 94 mm tall, 480 gram Scythe Samurai ZZ is basically a top-down version
of the Scythe Katana, bent into a "C" shaped to reduce its height.
A former champion on our old test platform, the Samurai ZZ is one of our favorites
because it fits mini-ITX LGA 1155/1156 motherboards without interfering with
the PCI Express slot. The only downside is its relatively tall height and the
use of pushpins for mounting.

Scythe Samurai ZZ
Fan Voltage
Fan Speed
°C Rise above Ambient
CPU
VRM
RAM
Stock 92mm Fan
12V
2480 RPM
34~35 dBA
37
25
23
9V
1800 RPM
25 dBA
42
31
27
8V
1440 RPM
18 dBA
45
38
30
7.7V
1300 RPM
15 dBA
46
39
31
7.6V
1100 RPM
12~13 dBA
52
47
32
Reference 92mm Fan
12V
1470 RPM
14 dBA
44
36
28
9V
1150 RPM
11~12 dBA
51
43
32

The Samurai ZZ's PWM fan speed drops dramatically below 8V, so we had to test it at the unconventional 7.7V and 7.6V levels to get good data points at lower noise levels. As the heatsink is on the large side, it doesn't suffer from severe performance degradation when the fan speed is lowered to close to inaudible levels. While the CPU was cooled fairly well, VRM/DIMM cooling was a bit poor.

Scythe Kozuti (Street Price: US$35)



The Scythe Kozuti.

The Scythe Kozuti is easily outclassed by the larger coolers in this roundup, but for a petite heatsink standing only 40 mm tall and weighing 250 grams, it's an impressive little performer. Its slim 80 mm fan is placed directly over the base with three heatpipes and a stack of fins sitting on top. It's one of the few heatsinks on the market that can actually cool a mainstream desktop processor inside a truly low profile case while generating a reasonable noise level.

Scythe Kozuti
Fan Voltage
Fan Speed
°C Rise above Ambient
CPU
VRM
RAM
Stock 80mm Fan
12V
3290 RPM
34 dBA
50
25
24
9V
2500 RPM
23~24 dBA
53
31
27
8V
2140 RPM
18 dBA
57
36
29
7.5V
1850 RPM
14~15 dBA
62
40
32
7V
1580 RPM
12 dBA
65
45
33

While the Scythe Kozuti is the runt of the litter, it still managed to cool the CPU well enough to prevent it from throttling. CPU temperatures were uncomfortable at quiet levels (below 2200 RPM / 20 [email protected]) but it survived, no small feat for such a diminutive heatsink. The results were more or less linear with the CPU temperature increasing 3~5°C for every 300 RPM reduction in fan speed.

Heatsink Comparison Tables

°C rise Comparison (CPU Temperature)
20
19
18
17
16
15
14
13
12
Noctua NH-L12

(both fans)
33
-
-
34
-
35
-
36
Noctua NH-L12

(ref. 120mm fan)
-
-
-
-
-
34
-
37
Noctua NH-L12

(120mm fan)
-
-
37
-
-
38
-
39
42
Scythe Big Shuriken

(ref. 120mm fan)
-
-
-
-
-
41
-
43
Scythe Samurai ZZ

(ref. 92mm fan)
-
-
-
-
-
-
44
-
51
Scythe Samurai ZZ
-
-
45
-
-
46
-
52
Noctua NH-L12

(92mm fan)
42
-
44
-
47
-
51
57
Scythe Big Shuriken
43
-
-
-
-
46
-
61
Scythe Kozuti
-
-
57
-
-
62
-
65

In our new performance chart, the Noctua NH-L12 comes out on top once again.
The two stock fan configuration was neck-and-neck with just one of our reference
Nexus 120 mm fans, while Noctua's 120 mm fan by itself was 4~5°C worse.
The 120 mm Nexus also enabled the Scythe Big Shuriken to claim fourth place
despite its much smaller size compared to the Noctua. There's a big gap at the
12 [email protected] level before the Samurai ZZ with reference Nexus 92 mm fan appears,
followed closely by the same heatsink utilizing its stock fan. The NH-L12 with
the stock 92 mm fan isn't far behind, clearly not the same heatsink when saddled
with a smaller fan. The stock Big Shuriken sits in second to last place, but
with a fan half the standard thickness, it's a respectable result. Finally,
the Scythe Kozuti brings up the rear as one would expect though it's hardly
a disappointment for a heatsink literally half the size of the competition.

°C rise Comparison (VRM Temperature)
20
19
18
17
16
15
14
13
12
Noctua NH-L12

(both fans)
17
-
-
19
-
21
-
23
Noctua NH-L12

(ref. 120mm fan)
-
-
-
-
-
20
-
25
Scythe Big Shuriken

(ref. 120mm fan)
-
-
-
-
-
25
-
29
Noctua NH-L12

(120mm fan)
-
-
24
-
-
26
-
27
32
Noctua NH-L12

(92mm fan)
28
-
31
-
33
-
38
43
Scythe Samurai ZZ

(ref. 92mm fan)
-
-
-
-
-
-
36
-
43
Scythe Big Shuriken
28
-
-
-
-
30
-
47
Scythe Kozuti
-
-
36
-
-
40
-
45
Scythe Samurai ZZ
-
-
38
-
-
39
-
47

When we look at VRM cooling, the results are shuffled a bit with the Samurai
ZZ being the biggest loser; paired with its stock fan, the Samurai was even
edged out by the tiny Kozuti. The distance of the fan from the board seems to
be the most important factor.

FINAL THOUGHTS

The high energy efficiency of Sandy/Ivy Bridge processors makes them a perfect
fit for smaller systems and motherboard manufacturers have responded by releasing
a slew of mini-ITX models. The hardware of our updated small heatsink test platform
is now much closer to what is currently being used by DIY builders, making our
results more relevant. Also, the Intel
Core i5-2400
's higher throttling temperature gives us a wider range
of data, making it easier to differentiate the performance of competing coolers,
while the Intel
DP67BG
's additional temperature sensors reveals each heatsink's ability
to cool the area around the CPU socket.

With the high-end cooler market more or less saturated with high performance products, we hope to see more attention being paid to modestly sized heatsinks. It's a more interesting area in a sense, as the thermal and physical limitations of smaller cases makes it difficult for poorly designed heatsinks to perform well. Manufacturers of larger models often compensate for this with mass, more fins, more heatpipes, and bigger fans, but it's not so easy when a size limit is imposed. The four heatsinks we re-tested today vary significantly in weight, size, and design, at least more so than the countless tower heatsinks utilized by enthusiasts.

The Noctua NH-L12 is the
only heatsink in our roundup that can be considered premium. Except for its
size, it is a classic Noctua in every way, shape, and form, carrying a high
price tag but offering exceptional performance. It retains its previous champion
status, but with both fans mounted, it is fairly large. Thankfully if you can
fit its 120 mm fan under the fin stack, it can still achieve superb performance
without a second fan on top, cutting its height down to a modest 66 mm.

The Scythe
Big Shuriken
isn't the best performer, but it does provide an unbeatable
level of cooling for any heatsink up to 58 mm high. It doesn't have the best
mounting system, either, but this has been rectified in the latest version,
the Shuriken 2 Rev.B. Even if the new version only matches the performance of
the old, you'd be hard pressed to find something better at this size and price.

The Scythe Samurai ZZ fails to impress given its 94 mm height. It's not a bad cooler by any stretch of the imagination, but it's certainly outclassed by the NH-L12. However, if you're planning on a LGA1155 mini-ITX build with a discrete graphics card, the Samurai might be your best bet. The CPU socket on mini-ITX Sandy Bridge boards are so close to the PCI Express slot that most heatsinks interfere with it. The Samurai doesn't.

The Scythe Kozuti seems awfully expensive for such a tiny piece of metal and plastic, but it's a formidable little beast that is truly low profile. This 40 mm tall heatsink successfully tackled our overclocked quad core Sandy Bridge processor, though the CPU temperature did get quite toasty at lower fan speeds. It should be able to comfortably handle a 65W CPU without producing too much noise, even in an ultra slim case.

Our thanks to Noctua and Scythe for the CPU coolers used in this article.

* * *

Articles of Related Interest


Noctua
NH-L12 Low Profile Cooler

Scythe
Kozuti Low Profile CPU Heatsink

Gelid Slim
Silence & Prolimatech Samuel 17 Low Profile CPU Coolers


Fan
Roundup #6: Scythe, Noiseblocker, Antec, Nexus, Thermalright

AMD Phenom II Stock Coolers
Intel LGA1366 Stock Cooler: Good Enough?
Gelid Silent Spirit &
Scythe Samurai ZZ CPU Coolers

* * *

Discuss
this article in the SPCR forums.

Google

www SPCR