SPCR

Thank you Lawrence, for the good work and... the good news!

P.S.: With reference to your:

NOTE: "Our Noctua NF-P14 140 mm reference fan has been retired as it is no longer being sold"

there's the current P14s Redux, which is the same fan with different frame, or the P14r Redux, which is the same fan with a PWM IC.

Welcome back Ninja! We've missed you!

Oh look, finally a young ninja that makes his grandfather proud.

This one may even live a long life, for a ninja.

As a massive fan of the original, it warms my heart to see Scythe on top again with the new Ninja

Thanks!

Jordan wrote:As a massive fan of the original, it warms my heart to see Scythe on top again with the new Ninja

My thoughts exactly! And add to that 3 other Scythes in top 10.

The original Ninja still chugs along on a sligthly OCed I5 2400 in my main rig, coupled with a SlipStream. I got the cooler just before its next iteration, with the stupid pushpins, was to enter the market - early 2006 I think. At that time I didn't fully understand the importance of the mounting system, but somehow sensed the puny pushpins would not be as good as the sturdy metal brackets (which later proved to work great on the 1155 platform). Now, with the fan at 500-600 RPM the CPU idles at 33-35 C and never reaches 60 C (gaming stress).

A pity it took a good few years for Scythe to fully understand the importance of good contact & pressure.

SPCR, thanks for another quality review!

Great review. A major question rises tough. If I had to choose between these recent top-scoring models from Scythe (Mugen 4/Max - Kotetsu - Ninja), which one should I take and why?

Perhaps go for the one with best clearance over RAM modules if you have some fance sticks with tall heatsinks (useless, but it's not the point now). Or maybe go for the one with the best sounding fan? Ninja, Kotetsu and Mugen Max seem to have - at 11 db - very similar performance, with 40-43 C rise. 3 degrees may very well be attributable to sample variety and, for all practical purposes, it's a non-issue.

I'd personally go for Ninja as I don't give a damn about RAM clearance but I'd be happy with the small thermal bonus.

Or just get the one that pleases you most esthetically That'd be Max for me, in that case, a lovely cooler.

Great review. Of course I already bought a Mugen Max. Typical me - terrible timing.

I'm a bit puzzled why a 140mm fan was used as the reference, as the stock fan is 120mm?

Potenza wrote:If I had to choose between these recent top-scoring models from Scythe (Mugen 4/Max - Kotetsu - Ninja), which one should I take and why?

I'd take the Kotetsu because it's still the most efficient considering its lower price, size, weight, better stock fan acoustics, and is 2nd place only by 1dBA.

This is great news! However I just bought the Noctua NH-U14S

I still have a brand new original Ninja Plus in its box somewhere, I would definitly use it but support for newer sockets seems impossible.

The way I recall it, the orinigal ninja was marketed as a "fanless / passive" solution (well, as using the case airflow to function). Would the Ninja 4 be usable as such? I suppose with the low noise quality fans we have today this is not an issue anymore...

As for the support for 1155 (and the like) - the first Ninja can actually be used. See here - there's plenty of ways to make it work, all easy and cheap. As for the Plus version - you'd perhaps need some more aftermarket elements, making this a not so cost effective solution anymore.

The very concept of fanless is great and all, but (just like you said) for practical purposes a quality 120 mm fan @ 300-500 RPM is virtually and practically silent, yet provides plenty of direct airflow so why not use it? Today's mobos have great PWM control features and with PWM fans that can go as low as 300 RPM it would be a waste not to use such a set-up. That said, I bet Ninja 4 will perform perfectly OK with non-K quad mid-range Intels of today, provided there's some airflow in the case still. For K CPUs and perhaps higher clocked I7 - that might be tricky.

When does this get to retail?

Great review! Happy to see Scythe constantly improving their products.

Jay_S wrote:When does this get to retail?

everything ive seen said the last week of june... :/

Nice!

I still love my Mugen 2, although this i7-4770K is the first CPU I can hear the fan ramp up on. Too little contact area heats up much faster than anything before. :-/

nagi wrote:Too little contact area heats up much faster than anything before. :-/

That's just the internal TIM: delidding the CPU (in order to replace it) would lower the cores temp by 10-20°C (depending of the CPU sample), thus dialing down the fans speed.
On the other hand, delidding the CPU may also destroy the CPU itself, in case of execution errors.

I too would be interested to know how well it performs fanless. I've used all my CPU heatsinks fanless for a few years now, and it's always hard to work out which heatsink is going to work best, because there is no reliable data.

Mr Evil wrote:I too would be interested to know how well it performs fanless. I've used all my CPU heatsinks fanless for a few years now, and it's always hard to work out which heatsink is going to work best, because there is no reliable data.

Given that fanless heatsinks need (more) case airflow, fanless is pointless.

quest_for_silence wrote:

nagi wrote:Too little contact area heats up much faster than anything before. :-/

That's just the internal TIM: delidding the CPU (in order to replace it) would lower the cores temp by 10-20°C (depending of the CPU sample), thus dialing down the fans speed.
On the other hand, delidding the CPU may also destroy the CPU itself, in case of execution errors.

You are both right. It's a twofold problem:
- smaller die have poorer C/W.
- once you reduce the die to a certain size while retaining the wattage, Solder based TIM interface will cause physical stress fractures on the die with the repeated hot/cold cycles. That's why Intel went to a paste style TIM which has a poorer C/W than the solder style, and hence poorer temps. The incredibly sad part is that they chose to save a penny and use a particularly crappy C/W paste TIM.

quest_for_silence wrote:Given that fanless heatsinks need (more) case airflow, fanless is pointless.

I strongly disagree. Given that a case fan is required anyway, why not use it to cool the CPU as well as everything else?


Note that in both of those systems, the CPU is cooled adequately at full load with the fans at 0% PWM (about 300-400rpm).

Mr Evil wrote:why not use it to cool the CPU as well as everything else?

Because I did, more than once (very recently with the Zalman FX70: there are some note about that addressed to a forumer interested in cooling that way a Core i5 here in the forum), since about maybe ten years ago, and it actually never worked as expected (I mean better/quieter than a fanned heatsink, and very often worse than it).

The only way which works good is either using extremely low power parts (like 5-7W BayTrail CPUs, I have a fanless HP with one of those quad), or heatsinked enclosures with low/mid-powered hw.

CA_Steve wrote: - smaller die have poorer C/W.
- once you reduce the die to a certain size while retaining the wattage, Solder based TIM interface will cause physical stress fractures on the die with the repeated hot/cold cycles. That's why Intel went to a paste style TIM which has a poorer C/W than the solder style, and hence poorer temps.

Then why does Intel still solder on their high end CPUs like the Haswell-E if there's a risk?

Example of temp difference between Haswell CPUs: http://www.techpowerup.com/reviews/Inte ... son/6.html

yakuman wrote:Then why does Intel still solder on their high end CPUs like the Haswell-E if there's a risk?

Because they are larger die. The breakpoint seems to be <~200 sq. mm. Take a look at the die sizes (a third of the way down the page) vs the parts you know to have solder TIM and those you know to have paste TIM. Sandy bridge (216mm^2) was solder, Ivy (160mm^2) was paste, Ivy-E (257mm^2) solder, and so on.

Thanks. I didn't know.

Cistron wrote:Great review. Of course I already bought a Mugen Max. Typical me - terrible timing.

I'm a bit puzzled why a 140mm fan was used as the reference, as the stock fan is 120mm?

+1

I wanted to see numbers for a 120mm fan at 800 RPM and for fanless and anything in between you feel like testing.

I too am curious how it compares to other "fanless" advertised CPU coolers like the thermalright macho and that one heatsink from silverstone that I forget the name of. Any chance this data can be collected and added to the review?

Thanks for the review nonetheless. Sounds like a resurgence of the original Ninja. It was my favorite CPU cooler back in the day.

Do we really still need such huge heat sinks when the Broadwell line maxes out at a TDP of 65W for the i7-5775? Can't much more modest sized coolers do the job in a virtually silent setup at that power level?

Or do serious overclockers still need something like the Nina 4?

Intel never brought out an i5 or i7 unlocked clock equivalent for Broadwell. Really, I'm hard pressed to see why anyone would buy the desktop Broadwell. However, Skylake goes up to 95W.

Do you need something the size of the Ninja 4 for a 95W CPU? Not really - look at the Kotetsu review. If you want to OC, however, the power use can certainly get up into (and past) the 125W range of the review CPU.