Scythe Ninja 4 Heatsink Review

Scythe was once omnipresent in the enthusiast marketplace, but have since been in and out of the scene. We saw the Kaze Master Flat II and the Kama Panel 3.1 in 2014. Now we have the Ninja 4. The Ninja heatsinks are beautiful, symmetrical heatsinks, and the Ninja 4 is all that. You would love to have this heatsink decorating your computer from its looks alone, but will it cool? Is it easy to mount? The Scythe Ninja 4 is a six-heatpipe heatsink that splits its twelve shoots into four finstacks. It’s too symmetrical to cool a CPU, right? Well, Scythe didn’t think so. They have revised this design over and over. Have they finally got it right?

Specifications & Features

(Couresty of Scythe)

Hm. Sleeve bearing. I don’t trust Sleeve bearing fans on heatsinks, even sleeve bearing fans as nice as this one.

Features

The Ninja 4 is the latest model in the Ninja CPU Cooler Series, developed for uncompromising performance. It features the new T-M.A.P.S technology, an optimized alignment of heatpipes, and the back-plate based Hyper Precision Mounting System (H.P.M.S) for firm mounting and easy installation procedure. These improvements and a special, adjustable Glide Stream 120mm PWM fan result in an increased cooling performance while reducing the weight compared to his predecessor. Also the design of the heat-sink allows fan mounting on all four sides. This enables the optimal integration of the Ninja 4 in the air flow of the pc-case and reduces turbulence and the emergence of hotspots.

Three-dimensional Multiple Airflow Pass-through Structure

By introducing our new T-M.A.P.S technology for the Ninja 4, we were able to increase the cooling capacity compared to previous models based on the older M.A.P.S lamellar structure. The clever arrangement of the cooling fins allows the usage of an additional pair of fins without suffering negative effects to weight or height of the heatsink and optimizes the cooling power in the lower speed range of the supplied fan.

Glide Stream 120 mm PWM Fan

The enclosed 120mm fan comes from the Glide- Stream PWM series. This new model offers the ability to adjust the maximum fan speed in 3 settings via a slide switch, allowing the user to choose between extremely silent operation or maximum cooling power.

Flexible Fan Mounting

Inspired by its predecessors, the Ninja 4 offers the option to install the fan on any of the four sides of the heatsink. This allows for very effective integration into air flow of the case. To further increase the performance, a second fan* can be installed by using the included additional pair of fan clips.

* Extra fans must be purchased separately.

User-friendly Installation System

The proven HPMS (Hyper Precision Mounting System) mounting system is also to be found in Ninja 4 again. The mounting of the cooler is much easier. It requires fewer installation steps and time without compromising reliability.

Packaging

Scythe packs their heatsinks in boxes hardly larger than the heatsinks themselves. They don’t take up much room that way. You could imagine these things hanging on a wall display. As a matter of fact, given that Scythe got their start in the Akihabara Electric Town, it seems pretty likely.

Pop open the carton, and you are greeted with a small box holding the small parts. Under that is a panel of cardboard stabilizing the heatsink, with the included fan off to the side.

Here is a face-on picture of the top of the heatsink. Unfortunately, while you can put two fans up-and-down or side-to-side (we’ll be doing that) you can’t attach four fans; the clips would cross. I wanted to kludge a four-fan mount, but for various reasons (see below) I did not. No reason you cannot do that, though. For now, we will admire the 4-way symmetry of this heatsink.

Closer Look

Let us look at the unique 3-speed PWM Glide Stream Fan. It looks like a Slip stream with grooves. That’s not a bad thing. The Mugen II came with a PWM Slip Stream, and it cooled well. The only thing wrong with this fan is that it’s a sleeve bearing fan, which tend to wear out.

Notice that the switch positions are clearly delineated. Other fans have switches, but none others have positions so clearly marked. This one is not just a three-position switch, though. At each of the positions the fan can vary its speed according to the PWM signal. You don’t have to reach into the case to set the fan to a different speed, depending on what you are doing. The PWM does that. What the switch setting changes is the maximum speed the fan can reach at each position. This is like getting three variable speed PWM fans, each with a different maximum speed.

An oblique side view tells us a number of things. First, the fins are not too thin; they are solid and should carry some heat. Second, the fins are far enough apart that this should be a decent passive heatsink. Not for overclocked systems, mind you, but maybe OK for non-overclocked systems – especially two-core systems. Last, Scythe tells you in the instructions to blow your air through the heatpipes, from one side to the other. This may have something to do with the aluminum nub fins growing out of the base.

Again, looking down at the finstack. It kind of looks like a square four-leaf clover, doesn’t it? Maybe you didn’t notice before, but this is a single, quadra-lobate finstack.

Here we have a picture of two fans mounted on the Ninja 4 in push and pull. They are both matching Glide Stream fans. Look at those fan clips. Just grab the handle and yank to get them in position. To get them off, simply pull away from the finstack. Easy. Why make things hard? Finally, notice the direction of the fans. They are set as Scythe instructed, through the heatpipes. Combining the effect of air on the heatpipes where they are hottest and air on the nub fins, it must make a difference to cooling.

Here are two closeups of the fins gripping the heatpipes.

The first picture shows a bottom view of the Ninja 4 with the label still on the contact surface. The second picture shows the bare contact surface. You can see the basement beams in the reflections on the contact surface. I want to point out that this is a post-mount shot of the contact surface. The pics of the pre-mount surface were unsatisfactory. But the surface held up very well.

Accessories & Mounting Hardware

A boring but obligatory picture of your instructions, but the next shot is more interesting. Let’s see. We have a side bracket, a baseplate, another side bracket and a tension bar – all in black. Then we have a – What’s that? Is that a screwdriver I see? Hm. Perhaps emulating a certain high end competitor, but supplying a proper handle rather than an L-shaped piece of metal. I’m keeping that. Useful. Oh, and under that is a tube of TIM.

Now we have a picture of some silver-colored small parts. First off, a fan clip. It looks too small to fit on the fan. That’s not an optical illusion. The clip has to stretch to fit on the fan. The stretching causes it to apply pressure, and hold itself on the fan. That means you can let go of the clip, so you won’t need three hands to mount the fan on the finstack. When the time comes, just grab the handles and pull, and the fan goes on. Did I tell you it was easy? Then, we have a Glide Stream with attached fan clips, another fan clip, and we’re ready for the second row. Wrench (small. To keep you from overtightening something) and fan cable. For the third row, there are two washers, two tensioning screws (where are the springs?), four combo screw-nuts for LGA 2011, two more washers and a double-sided sticky for LGA 775. In the fourth row, we have four short screws and four screw-nuts for spacers. That’s a lot of small parts!

Installation

The spacer screw-nuts are just long enough to reach through and engage the threads in the baseplate. If this baseplate looks a little off to you, that’s because it is. The screw positions of LGA 775, 1125x and 1366 are close enough together that only by offsetting them could Scythe get all the screws into threaded holes. As you can see, the screw part was just long enough to reach. It could have been longer. Or the spacers could have had pilot shafts. I have included a photo of a thumbscrew with a pilot shaft so you will know what I am talking about.

Here is what the spacers look like when they are screwed in. The screw parts go through the motherboard and are threaded into the backplate, as previously shown. The nut parts are ready for – what are they ready for?

They are ready for the flat screws. Don’t put on the LGA 2011 spacers, even though they fit. Notice the threaded holes, They are recessed so the tensioning screws will be able to find they. Here the sunken threads form a little pit to catch your screw. It’s all covered in the Instructions, thankfully.

Now you mount the heatsink. Just plop it down on top of the TIM, then position the tension bar over the grooves in the heatsink. Practice this bit a few times with the sticker still on the heatsink until you can get it right. I don’t like loose tension bars, but it is crosswise to the airflow. It doesn’t butt up against a heatsink, so you have room to maneuver your hands. Plus, there are little pits to catch your screw tips. Still, you can run into problems getting the screw in. So practice.

Notice that the side bracket can bow up to provide pressure where there is no spring. The instructions tell you not to “overtighten” the tensioning screws (now you see why that wrench is so short?). But really, how tight is too tight? I put it to my Scythe contact, and he gave the engineers my question. They replied saying that no matter how tight I got it, I could not apply too much pressure to the CPU in its socket. OK, but I still prefer springs and positive stops. As it was, I tightened those screws until they were tight, but did not feel “overtight,” whatever that might be. As you can see, I carefully left two threads showing. In my opinion, this is a serious source of variance. But this is not the first time Scythe has used this mount. They get no complaints, so it must be an OK mount for you.

Those flat screws? Mandatory. Spacers interfere with the heatpipes. You could still make an error if you don’t read the directions, so read them.

RAM clearance. With the low voltage RAM available these days, you should be able to avoid buying RAM with tall heatsinks. As a matter of fact, you can get ultra-low profile RAM, and you should get those. However, some of you will have taller RAM. Below, the Ninja 4 sits directly on the socket of the practice motherboard. Because the screws bottom out, we know the Ninja 4 is sitting 2-3 mm lower than it will sit on a CPU. The RAM here is 40 mm tall, from the bottom of its connectors to the top of its tallest teeth. The bottom of the fin stack is 26 mm taller than the top of the RAM holder. So the Ninja 4 allows for RAM of moderate height.

Testing Setup

Scythe provided the Ninja 4, along with the extra Glide Stream fan. The other comparators were provided by their respective OEM’s, except for a Prolimatech Megahalems, which was purchased retail at the end of 2009, with a Prolimatech Genesis and a Noctua NH-D14 SE2011, which were both purchased retail at the end of 2013. Each heatsink was mounted on the night before testing. This gave the TIM most of a day to do any migrating it was going to do.

Linpack runs in surges. When the temperature is graphed, you see ragged plateaus. In looking for cooling solutions, you want to know how well a heatsink cools those plateaus. So the temps under 70 °C (the valleys) are ignored in analyzing to core temps.

Each test run was 30 minutes in duration. The last 20 minutes of each run was measured, and the core temperatures were logged and analyzed in Open Office spreadsheets. An Intel chip reports its temps in one degree increments, so for best accuracy these reports should be averaged in aggregate. Here the core temps were measured once a second, resulting in 1200-line spreadsheets. Three test runs were averaged.

The digital thermometer measuring air temp reported its measurements in increments of 0.1 °C. The ambient temperature was measured every five seconds, resulting in 240-line spreadsheets. The mean ambient temp was subtracted from the mean core temps, resulting in a net temp for each run. The net temps were then averaged.

The wattage did not appear to be predictable, but it normally ranged plus or minus 2%. The main value it has is a check on the system, to make sure the CPU is heating up properly. The CPU package normally ran around 140 watts during Linpack surges, so it represented a cooling challenge for a heatsink. Heating less than 130 watts was ignored, because that was when the CPU was resting between Linpack surges.

The sound pressure level was measured 1 meter from the heatsink, with the motherboard set vertically, the way it would in your case. The ambient noise for this testing was 31 dBA. So the net SPL is the sound pressure level measured at 1 meter, less 31 dBA.

After the NH-C14S review, the heatsink sat for weeks. This turned out to be a mistake, for the bottom of the heatsink, the contact surface, was convex. It produced great cooling – too great. Letting the heatsink sit so long dented the IHS or something, because heatsinks that previously cooled the CPU no longer did so. Now the old CPU throttled from max temps. So what you see here is a new CPU. Advances in the year since the i7 4790K was introduced means that we had to run the CPU slightly faster to get it hot enough to really test these heatsinks. These results, then, cannot be compared with earlier results.

Testing Results

I have lined up the seven heatsinks, with the best cooling at the top. I’ll let you take a moment to ponder the results. I’ll resume the text below the chart.

Well, we hardly expected the Ninja 4 to thump the two D15’s, and by such a large margin. We’ll come back to that but first look at the Ninja 4 differences between one and two fans, I arranged them in couplets so you can see the point I would like to make: using two fans adds mostly noise at high speeds. But with low-speed fans, having two fans makes more of a difference the slower you go. With the Ninja 4’s long air path, you can see how a pull fan would really make a difference here. Overall, the winning combination is to use two fans set on low. If you don’t trust sleeve bearing fans (I don’t), get two of the new Scythe Slip Stream DB PWM fans. The DB stands for Double Ball bearings. Or buy two 800 RPM Scythe Slip Stream DB fans and let them run at their fixed speed. You won’t hear them.

The reason the top line shows an average of four runs is that I didn’t believe the first one could be accurate; it was. Then, I decided to hurry up and get this cooling beast off my CPU — the last time I had a heatsink that cooled this well, it ruined my old chip for further testing. So the runs with the other fan combinations were cut short and I got the CPU off there after two days on the motherboard.

So I looked at the curvature of the contact surface. First I looked at the flatness aligned with the airflow. Looks reasonably flat. Then I looked cross-wise to the airflow. There is a convexity there, though admittedly this picture does not show it well. But the convexity is gentle, no more than the Noctua heatsink bases. I have to conclude that part of the success of this heatsink comes from applying more than a 60 lb. pressure, but only part. The rest of the explanation must come from a combination of the following elements: the length of the air path; the thickness of the fins; the fan air blowing across the heatpipes where they are at their hottest; those nub fins sprouting from the base. Whatever, the combination of features produced a whale of a heatsink, a cooling monster.

I was afraid that the Ninja 4 cooled perhaps too well. I feared that if I left it on too long, it would bend another CPU. I tested the cooling afterward on a known heatsink. It cools as well as it ever did. So the Ninja 4 did nothing to the CPU. Whew.

One more thing: if the bottom cooler ran 2 °C hotter, the spikes inherent in Linpack with AVX2 would cause the CPU to throttle. There are heatsinks that don’t cool as well as what you are seeing. This, then, is the major league.

Conclusion

I am cautious with my CPU since I have to use it over and over, and I don’t get them gratis from Intel. So it was a relief to know that the Ninja 4 did not harm my CPU’s Integrated Heat Spreader. You probably don’t test heatsinks, so you will likely pick one, install it, and forget it. After the Ninja 4 shows up with only a gentle convexity, there is no reason to hold back: this is one excellent heatsink.

The craftsmanship was impressive. The Ninja 4 was a joy to handle. The thickness of the fins felt just right. Not too thin, not too fat. Sturdy without being clunky. For mild overclocks on a machine that never works hard, you could probably use this heatsink passively.

As for the fan, it does well: no clicking, three clearly marked speeds. The only thing wrong is that it is a sleeve bearing fan. I don’t trust those, but this fan will last until you can do better. Soon I’ll be bringing you a review of Scythe’s new line of Slip Stream fans, now with double ball bearings. They call them Slip Stream DB fans, and they are just the thing for the Ninja 4. However, as it stands, having three speeds on the stock fan is like getting three fans in one.

I don’t know about you, but I cut out the rear grill before I use a case. On steel cases, that back grill is just an obstruction to airflow. If you’re going to mount two fans on the Ninja 4 – and you should – you will need every bit of room on this long heatsink. You can use that back fan – the pull fan – to extract the air of your case through the Bernoulli Principle. You won’t need a case exhaust fan.

Based on our testing, a pair of 800 RPM Slip Stream DB fans will cool your system, and you will never hear your heatsink. You could also go for a pair of Slip Stream DB PWM fans. At 1300 max RPM, they will spin up only when you need them. Otherwise, silence.

One could be very happy with the Ninja 4’s cooling. It’s great. I am less happy with the mount. Let’s start with the load bar. It comes off the heatsink. Because the tension screws are not held captive, it is a relatively simple and frustration-free exercise to put the load bar on the heatsink, and the tension screws on the load bar. Then you notice the instructions telling you not to over-tighten the screws. How tight is that? It would be better to have positive stops, IMO. That way, you could not over-tighten those screws. Back to the load bar: a little screw that holds it in place on the heatsink would be nice.

The bolt ends of the spacers were barely long enough to catch the threads of the backplate; the threaded part of the shaft should be a few threads longer and there should be a pilot shaft to help guide you. I am quibbling here, but this is a good enough heatsink that a few quibbles are in order. The Ninja 4 is close to perfect.

The fan clips hold themselves onto the fans, so that you can snap the fans on the finstack with no hassles. Nice.

When I put everything back in the carton I noticed something: There is enough room in the small parts box to hold the second fan, and there is still enough room for all the parts. Also nice.

Now, I will share with you the nicest part. The MSRP of the Ninja 3 Rev B is $52. Now, we should expect the MSRP of the next heatsink in this series to be a little more, right? Well, if you’re standing up, sit down. According to Scythe, the MSRP of this soon-to-be-released heatsink is “around $45 – $47.”

Bottom line: the Ninja 4 is an all-around nice heatsink to have.

Pros

• Beautiful
• Cools very well
• Sturdy
• PWM fan has three maximum speeds, making it three variable-speed fans in one
• The packing can be recycled
• The PWM fan does not click
• MSRP makes this a bargain

Cons (Quibbles)

• RAM limited to about 40 mm in height
• The threads on the spacers are barely long enough to engage the backplate
• The tensioning screws do not have a positive stop
• The fan has a sleeve bearing

Click the stamp for an explanation of what this means.

– Ed Hume (ehume)