Scythe Ninja 5 Review: Big Air On The Square

Big, monolithic air coolers are known for relying on multiple heatpipes and stacks upon stacks of individual cooling fins to provide vast surface area to effectively dissipate thermal loads. Of these, the newly released SCNJ-5000 Ninja 5 is the latest thermal assassin from the twin-blade cooling company, Scythe.  Sporting a pair of low-speed 120mm fans over a handsomely designed, interleaved, two-tone cooling fin stack, the Ninja 5 presents itself as an imposing heatpipe tower boasting near-silent operation, but still lands a little short in outright thermal performance for a cooler of its dimensions.


Offering support for nearly every modern CPU socket from both AMD and Intel, the Ninja 5 can sit atop just about every processor except Threadripper.  Shipping with a fairly-standard cooler mounting kit, the Ninja 5 goes a bit further with the inclusion of a syringe of thermal compound and Phillips head screwdriver.  Gone are the excuses for ‘not having everything you need’ to install this cooler.

Specifications

Height
6.25 inches / 158.8mm
Width5.125 inches / 130.18mm
Depth5.125 inches / 130.18mm (7.125 inches / 180.9mm with fans)
Base Height2.25 inches / 57.2mm
Assy. Offset0 inches (centered)
Cooling Fans2x 120 x 25mm
Connectors2x 4-pin PWM
Weight63.7 oz. / 1806g
Intel Sockets775, 115x, 1366, 2011x, 2066
AMD SocketsFM2(+), FM1, AM2(+), AM3(+), AM4
Warranty2 years
Price$60


The Scythe Ninja 5 features six, nickel-plated heatpipes that run vertically through the entirety of its 41 interleaved aluminum cooling fins, which can be seen overlapping through the center of the stack.  The base of the cooler integrates a broadly-grooved, mini-heatsink that integrates with the mounting arm of the cooler.  Each of the six heatpipes are symmetrically aligned and offset to allow linear airflow around them.

The nickel plated base of the cooler cocoons each of the six, 6 mm diameter heatpipes between the base block and the mounting top plate.  The center bar is permanently secured to the cooler base and holds a pair of spring mounted machine screws in position to secure to the hardware CPU socket crossbars. 

The strategic slotting of the cooler’s heatsink fin stack provides access for the long, narrow Phillip’s head screwdriver to reach each of the machine screws from above the cooler when securing the cooler to the mounting braces.

Twin Scythe Kaze Flex 120 mm fans adorn the front and rear of the cooling tower and are PWM capable.  Rubberized mounting elbows eliminate vibration noise while the fans are affixed using a pair of wire-spring clips.  The Kaze Flex fans are rated at 43 CFM of airflow and a max speed of around 800 rpm, however in our tests, we found them to peak at just a smidge over 900 rpm.

It becomes difficult to remember that our MSI X99S XPower AC motherboard is actually a full, E-ATX motherboard with the Ninja 5 (and fans) consuming nearly the entire width of the board above the CPU, from side to side. 

The cooler’s CPU socket bracing builds off either the included mounting backplate or, in our case, the motherboard’s integrated mounting threads. Then, using the handy magnetic-tipped screwdriver provided by Scythe, you can easily slip between the aluminum cooling fins to find the machine screws, secure the cooler and snug it down. 

MORE: Best CPU Cooling

MORE: How To Choose A CPU Cooler

MORE: All Cooling Content

Create a new thread in the Reviews comments forum about this subject
25 comments
Comment from the forums
    Your comment
  • stdragon
    Only 3.5 rating??!! You can only do so much with air cooling. How does it stack up with water cooling AIO units within the same price range?
  • panniiii
    Why no noctua comparison?
  • derekullo
    Would it be possible to include more modern cpu's in future cpu cooler charts?

    I get that the Intel i7-5930k is still being used so the comparisons against other coolers are still valid.

    Choosing the i7-9700k and/or i9-9900k and/or Threadripper 2950X as the processor would make the chart a lot more meaningful.

    Knowing that the Be Quiet Dark Rock Pro 4 keeps the cpu 11°C cooler than the Scythe Ninja 5 is a nice metric for ranking the cooling performance, but means absolutely nothing to me for current generation processors.

    Even if the cpu cooler being tested can't cool a modern processor at stock frequencies, establishing a safe/maximum temperature like 70°C and seeing how high a frequency each cooler can get would give great insight on which cooler to purchase.

    70°C was chosen due to it being just above the maximum temperatures we see here with the i7-5930k and with the expectation that these coolers would perform at a temperature above 70°C with more modern processors like the Threadripper 2950X at stock settings.

    Of course this temperature limit isn't set in stone if more testing data comes along that pushes for a higher or lower limit.

    I would imagine the easiest way to limit maximum frequency would be to set the limit within the power options of Windows, although bios could also be used for more granular control.

    For hypothetical example using a Threadripper 2950X as the cpu;

    Scythe Ninja 5 limited to 70°C achieves a maximum frequency of 4.0 Gigahertz on all threads.

    Be Quiet Dark Rock Pro 4 limited to 70°C achieves a maximum frequency of 4.2 Gigahertz on all threads.


    A similar setup could be done for liquid coolers with the difference being that due to there much higher cooling ability a lower temperature would need to be chosen, like 50°C.


    Another idea would be to have a 50°C limit for both air and liquid which would allow you to compare them directly against each other.

    Air-cooled frequencies would most likely be low and not realistic for the average person, but it would still be meaningful.


    With the temperature held as a constant this would make for interesting comparisons between models.

    It would allow you to see precisely how much the heat produced from the extra threads from the Threadripper 2970WX limits the frequency versus the Threadripper 2950X.

    Or another interesting comparison how much frequency is limited for the i9-9900k versus the i7-9700k with the inclusion of hyperthreading.
  • derekullo
    Even if the above is too much work simply testing each cooler on an i9-9900k with a pass/fail rating at a predetermined temperature would be useful.

    "This processor is able to cool an i9-9900k at stock frequencies at 70°C"

    "This processor is NOT able to cool an i9-9900k at stock frequencies at 70°C"
  • mischon123
    All cases are different. Mounting is not a science but fuddling. This test only applies to that case and this build. The errormargin is way to high. Stop.

    Why bother if you can get watercoolers for less money. Why bother with a case that does not blow upwards and fighting the chimney effect. Air to air heatexchange is long dead.
  • derekullo
    Anonymous said:
    All cases are different. Mounting is not a science but fuddling. This test only applies to that case and this build. The errormargin is way to high. Stop.

    Why bother if you can get watercoolers for less money. Why bother with a case that does not blow upwards and fighting the chimney effect. Air to air heatexchange is long dead.


    All cases are different except duplicate cases.
    Fractal Design - Define R6 White = Fractal Design - Define R6 White

    Mounting is indeed a science.
    https://www.struers.com/en/Knowledge/Mounting
    Not sure why you wanted to put them down.

    You are right in that a test with this case (assuming they don't go with open bench) would only apply to this case, but that is the benefit of creating a standard testbed where only one variable is changed at a time, which in this instance is hopefully just the cooling solution.

    https://en.wikipedia.org/wiki/Testbed

    I figured I would get air-cooler hate after posting this, but since the actual cooler in the article was air I had to at least ask about it too.

    I did suggest that we should also test liquid coolers with a similar methodology.

    The error margin is only high when you don't take all critical variables into account.

    For Tom's other tests they even go so far as use a thermal imaging camera and a specialized sound meter to record their data.

    If you look at the thermal imaging slides you can see that there are 2 sensors taking measurements; one taking a base reading 17-18°C and a maximum reading that is actually pointed at the component being measured.

    1°C between tests feels acceptable.

    Inside of a structure you can control most all variables that can affect heat transfer; external radiation (open windows letting the sun shine on your computer), air temperature, humidity and to an extent the air pressure. (Let me know if I'm missing one)

    Don't let the sun shine directly on your computer that you are testing ... close the window / move to a part of the room without a window.

    Humidity can be controlled with a humidifier/dehumidifier with the goal being to keep the humidity the same for all tests.
    I would imagine high humidity being beneficial to cooling due to its increased ability to hold heat along with the reduction of that dreadful static electricity.

    Finally we have air pressure.
    Assuming they conduct all cpu cooler tests at the same altitude (buildings tend to stay at the same altitude) and that the test isn't conducted at an extreme altitude we should be able to safely disregard this variable.
    Higher altitude would mean lower pressure meaning the air particles are farther apart which should reduce the cooling of both air and water coolers.
    Water coolers eventually have to transfer their heat into "something".
  • ubercake
    I'm reading these comments and something needs to be shared here.

    Most people jumping from stock CPU coolers to their first all-in-one (AIO) low-cost liquid-cooling (LCLC) solution think AIO liquid is the bomb when really the only thing they have to offer vs high-end air is they take up far less space around the motherboard making it generally easier to work inside the case or may even be necessary inside some cases due to space constraints. The AIO LCLCs also keep a really clean appearance to a build. And those pretty lights on some of the units...

    The downsides to AIO LCLCs are most don't cool as well as high-end air. They involve fans that create far more noise by requiring a higher RPM out of them under load. Most AIOs come with stock fans you have to replace to even tolerate the amount of noise they generate - of course ymmv based on what you consider tolerable. In addition, you do have the off-chance that you get a bad AIO liquid unit that will leak onto your other equipment.

    Further, one of the more expensive AIO units I had for an extremely brief period had associated software which controlled the pump in the unit by way of a USB connection to the unit through the motherboard. One day, after working for a while, I left my computer on and ran some errands. I returned after a few hours to find an error message from the unit's software and the pump had ceased running, while the fans were spinning at 100% (mind you the fans were in no way near my CPU). When I attempted to report the problem to the manufacturer, the company said they've found their software to have issues running with anything newer than Windows 7 and even suggested I downgrade. None of this information was available prior to my purchase. The manufacturer assured me the software would be compatible with the new OS within a couple of months. It was at that point I decided I would not put my trust in software running on top of the OS to keep an AIO pump running.

    My experience with quite a few AIO LCLCs (thick rad, thin rad, single fan up to triple fan) and my experience with the performance of high-end air (NH-D15 and even Hyper 212s) have convinced me to no longer seek out AIOs for my builds unless space over the CPU is an issue with the build. The temps are better with high-end air. There is less (fan) noise with high-end air. Your high-end air cooler is not going to leak onto your system.

    If the build facilitates it, I'll take a high-end air cooler over an AIO LCLC any day. This may change one day as I continue to try AIO LCLCs, but that day has yet to come. I haven't found an AIO yet that has impressed me.
  • stdragon
    Anonymous said:


    Most people jumping from stock CPU coolers to their first all-in-one (AIO) low-cost liquid-cooling (LCLC) solution think AIO liquid is the bomb when really the only thing they have to offer vs high-end air is they take up far less space around the motherboard making it generally easier to work inside the case or may even be necessary inside some cases due to space constraints. The AIO LCLCs also keep a really clean appearance to a build. And those pretty lights on some of the units...

    The downsides to AIO LCLCs are most don't cool as well as high-end air.


    Right, and they'll eventually become 'heat-soaked' if running sustained high CPU loads for an extended period of time. But that's the nice thing about water, its thermal mass. Otherwise, an AIO cooler will generally perform better for most brief heavy workloads.

    Also what's overlooked is the GPU. Idealy the GPU HSF will throw wasted heat outside the case. But many just blow them off the HSF and directly in the case. The later scenario just raises the internal case air temps which forces the air-cooled CPU HSF to work harder. With an AIO, it' doesn't' really matter all that much what kind of GPU you have installed because the wasted heat is going to dumped outside the case; where outside ambient temps are going to be lower.
  • ozicom
    Weight 63.7 oz. / 1806g
    I used laptops lighter than this CPU Fan :) What kind of a motherboard can handle this weight without cracking?
  • zfreak280
    This cooler has relatively poor performance because the radiator is too thick. The thicker the radiator, the worse this problem will get. This is a simple forced convection issue. The air towards the front of the radiator will be the coolest, and (to put it simply) will have the highest heat absorption rate. As the air travels through the radiator, it will get warmer and thus reduce is ability to absorb heat. Past a certain point, the air is nearly saturated for a given flow rate and it won't absorb heat. I would imagine the air in this design has reached peak heat saturation near the center, and the fin area at the center is very hot. Noctua and others get around this by having two, thinner radiators. It is disappointing that the engineers at a very reputable cooling fan designer wouldn't know better.
  • stdragon
    Anonymous said:
    Weight 63.7 oz. / 1806g
    I used laptops lighter than this CPU Fan :) What kind of a motherboard can handle this weight without cracking?


    At near 4 pounds, it's heavy. But modern gaming / overclocking MB will have a thick PCB and be well mounted anyways. It's only a problem if you plan on moving the computer to a LAN party or shipping it. In both instances, it would be well advised to remove it prior to transport.
  • ubercake
    Anonymous said:

    Anonymous said:

    ...


    Right, and they'll eventually become 'heat-soaked' if running sustained high CPU loads for an extended period of time. But that's the nice thing about water, its thermal mass. Otherwise, an AIO cooler will generally perform better for most brief heavy workloads.

    Also what's overlooked is the GPU. Idealy the GPU HSF will throw wasted heat outside the case. But many just blow them off the HSF and directly in the case. The later scenario just raises ambient air temps which forces the air-cooled CPU HSF to work harder. With an AIO, it' doesn't' really matter all that much what kind of GPU you have installed because the wasted heat is going to dumped outside the case; where ambient temps are going to be lower.


    I've had the issue with noise from the AIOs whether my GPU is blower-style or multi-fan style.
    I think there's a place for pressure in the conversation. Positive pressure in the case has kept noise down in my main air-cooled system by allowing much of the heat to passively "escape" as well as keeping my system virtually dust-free.
    With an air cooler over the CPU, you can use your front fans to bring in cool air, in turn, keeping the internal temperature cooler to start with, then you're adding heat from the internal components.
    In order to get positive pressure with the high-end AIO LCLC (since you need more air on the way into the case and most are using triple-length rads), you're pulling in the air from outside the front of the case, across the heated fins of the radiator and in doing so, increasing the temperature inside the case upon entry to the case.
    On the other hand, if you put the same triple-length rad up top and blow the air out (often causing negative pressure where there are not enough fans pulling in air through the front) you're taking internal heated air and blowing it across the radiator. Because of the negative pressure, heat is less likely to passively escape and will stew around in the "dead" pockets of the case. If you add more fans to the front of the case to counter the negative pressure, you add fan noise.
    Ideally, the rad and fans for an AIO liquid-cooled solution would be outside the case, but I don't know of any that are. I think this is why they don't perform any better (and often less so) than high-end air in an actual PC case. There's really a problem when it comes to pressure and accommodating a high-end AIO LCLC solution in anything other than an open test bench.
    There's really a lot to consider, but in summary, based on my experience with many AIOs in different types of cases, I really haven't found an AIO worth the money or risk.
  • stdragon
    Anonymous said:
    This cooler has relatively poor performance because the radiator is too thick. The thicker the radiator, the worse this problem will get. This is a simple forced convection issue. The air towards the front of the radiator will be the coolest, and (to put it simply) will have the highest heat absorption rate. As the air travels through the radiator, it will get warmer and thus reduce is ability to absorb heat. Past a certain point, the air is nearly saturated for a given flow rate and it won't absorb heat. I would imagine the air in this design has reached peak heat saturation near the center, and the fin area at the center is very hot. Noctua and others get around this by having two, thinner radiators. It is disappointing that the engineers at a very reputable cooling fan designer wouldn't know better.


    The CFM fan rating is important as it the amount of fins (surface area) on any HSF. But the problem with too much surface area is in resistance; that's where static pressure can help push the airflow through.

    I really wish vendors would publish the static pressure ratings. I find that many do not. Not sure if that's on purpose or that most people don't care.
  • shrapnel_indie
    It seems that the cooler would benefit from faster fans, from a cooling point of view. Yes, it would damage the sound metrics, but it should improve cooling, using static pressure fans of course. The old Cooler Master 620 had the same issue with stock fans... fans chosen more to keep it quieter than to actually be a cooling monster. (Sounds like this new one may suffer the same thing.)

    I understand that with software controlled pump speeds on liquid cooling, the idea is to cut down on noise, but my experience has been, at least with what I've owned, they aren't really all that noisy... skip the faulty software, and let it run full speed... minimal issues,

    It's a trade-off, and most understand this. Also, how well you hear the fans (or a pump with liquid cooling) depends on ambient noise as much as how much noise the unit produces.
  • ubercake
    Anonymous said:
    ... (or a pump with liquid cooling) ...

    The AIO includes both a pump noise (usually close to inaudible) and at least a fan (usually running at high RPM and loud). Any high-end AIO LCLC worth anything will include 2 or 3 fans and those that come with it are usually pretty loud.
  • TJ Hooker
    Anonymous said:
    Weight 63.7 oz. / 1806g
    I used laptops lighter than this CPU Fan :) What kind of a motherboard can handle this weight without cracking?

    According to Scythe's specs, it's 1190g. Which would make way more sense.
  • shrapnel_indie
    Anonymous said:
    Anonymous said:
    ... (or a pump with liquid cooling) ...

    The AIO includes both a pump noise (usually close to inaudible) and at least a fan (usually running at high RPM and loud). Any high-end AIO LCLC worth anything will include 2 or 3 fans and those that come with it are usually pretty loud.


    I'll admit to that. My Cryorig A40, you really can't hear the pump, even within a foot of it and the case open. The fans on start up, rev up enough they are loud, but not so much under CPU load. I suppose if I pushed my CPU long and hard enough, they'd get loud though. Most cases though, they're just not much different than other background noise. I'm not much for a 1x120mm or 1x140mm class radiator unless space is a critical and deciding factor.
  • TJ Hooker
    Anonymous said:
    Only 3.5 rating??!! You can only do so much with air cooling. How does it stack up with water cooling AIO units within the same price range?

    There are other air coolers that perform better and cost less, as shown in this very review. Why do you feel it deserves a better rating?
  • stdragon
    Anonymous said:
    Anonymous said:
    Only 3.5 rating??!! You can only do so much with air cooling. How does it stack up with water cooling AIO units within the same price range?

    There are other air coolers that perform better and cost less, as shown in this very review. Why do you feel it deserves a better rating?


    I'm not sure it does, i'm just trying to understand why it's a 3.5. I mean, that's beefy HSF that should perform well relative to other HSF units on the market. OTOH, compared within the same class, sure, 3.5 seems fine. But then there's the whole price/performance rating. Would an AIO not be worth considering as an alternative too if near that price point too? Just asking.

    BTW, I was looking at building an AMD Threadripper rig for someone. I was seriously considering the DARK ROCK PRO TR4, but not sold in the US. Can't find it anywhere, and only available to import from the EU? WTF??! Major bummer.
  • rubix_1011
    Price isn't the issue - it is the need for better fans or better flow-through airflow. I know the Ninja 5 is marketed as being big and silent, but looking at comparison coolers I referenced, it was still producing slightly higher noise levels while running slower fans, although at those levels, it is hard to audibly tell the difference, which is why the meter is used.

    From the perspective of why we're using the old, i7 5930k - we're testing for upper threshold limits of these coolers, which is different than testing 'current CPU' results. The i7 5930k is a 140 watt TDP part while the i9 9900k is a 95 watt part. There are 2 separate schools of thinking when it comes to benchmarking coolers:

    1) Test cooling products to determine highest limit loads we can produce; aka, a CPU cooler's most difficult challenge

    2) Using modern CPUs to benchmark which might not load the cooler as high, but provides a 'today's product' look at the cooler

    For us, we have chosen #1 in attempt to isolate small details and design points between each cooler because how we see it, if all coolers always perform well and rate really highly, then we're not using a difficult enough benchmark to separate products being tested. We are also not doing our due diligence to the overclocking and builder community if we are not providing feedback on any limits or shortcomings found by using hardware that generates higher thermal loads.

    In short, we're looking to provide comparisons based on highest TDP needing to be addressed by a cooling solution, and not the model of CPU specifically.

    I'll have the Dark Rock Pro TR4 coming soon - I'm working on the new 2990WX test setup for all Threadripper coolers, including re-tests of some for updated comparison.