Best 'air king' for 'silence first'? Best heatsink for fan swap?

freeagent · September 22, 2020

I have to ask, just how much paste are you using? When people say the size of a pea, disregard what they have to say after that. That is a huge amount. A couple of grains of rice and that’s it, maybe not even that much. I can’t speak for AMD since I haven’t run with them since 939 days.

As for my system, it’s not silent, but it’s not loud either unless I want it to be heheh.

Hang in there man, the experience will come..

NewbieOne · September 22, 2020

On 9/21/2020 at 6:09 PM, Quadriplegic said:

Ninja 5 is the ultimate silent air cooler. Sadly, performance does suffer a bit.

On 9/21/2020 at 7:29 PM, freeagent said:

TY-147A and TY-147B both start off at 300 rpm, they came with each of my coolers. There maybe more but such a low rpm has never really intested me. I’m sure the Ninja 5 is ok, just not what I would be looking for. For reference my TY-143 starts off at 600rpm.

Well, let's say both Ninja 5 and Le Grand Macho RT — and possibly Mugen 5 PCGH and to some extent any Mugen or Fuma, especially Fuma 2, as well as a large part of Thermalright's line-up — are a much welcome gift and source of relief for people like me. They're all precious.

Speaking of which, I've just made an inventory of my options, and it looks like this (using USD, but the location is in EU):

NIB D15 — $105 shipped from Noctua ($91 shipped from Noctua outlet — refurb from consumer returns)

NIB D15S — $91 shipped from Noctua

NIB True Spirit Power — $66

NIB Le Grand Marcho RT — $86

NIB Fuma 2 — $71 (turning attention to this because it often comes out surprisingly silent, even beating Ninja 5)

NIB Ninja 5 — $73

used D14 naked heatsink plus clips — $45 (I already have 2x140 BQ SW3 HS or could buy 2xP14, which are probably better than A15)

used PH-TC14PE naked heatsink plus clips — also $45

exhibition Olymp 2 (w/both fans) — $57

exhibition Fuma 1 — $44 shipped

used Ninja 4 (1 fan) — $ 31 shipped

Noctua outlet/refurb U12A for $91 (could perhaps put the SW3s on that heatsink and use the 2xNF-A12x25 for a GPU fan swap)

Various Dark Rocks Pro and 'non-Pro' — 2, 3 and 4 — different, usually attractive prices especially for used units (I've seen a rare passive test showing the DRP to have a really strong heatsink, but the whole cooler often ends up noisier than Noctua under high loads), though I'd rather skip the legendary installation issues

Reason I'm including prices:

1. To avoid 'almost as good as D15 but much cheaper' for coolers that are no longer much cheaper than D15.

2. To consider a temporary solution while waiting for Noctua's 'D16' (D15 replacement) and 'U14A' (14cm version of U12A) and NF-A14x25 (NF-A12x25 successor).

3. To recycle my 2x140mmm BeQuiet Silent Wings 3 hi-speed, which are actually very good CPU fans. For example the idea behind Dark Rock Pro is probably more about these fans than the heatsink. And D14 has a better heatsink than D15. The same may be true about PH-TC14PE, even Olymp 2. The idea would be to stick with my Wings or, rather than the default Noctua/Phanteks/Alpenfohn/whatever, get 2xP14.

As for my target cooler, I don't care if it's $150 total for air cooler, as long as I can install and forget (within reason, obviously not +$50 for like -1 degree). However, something cheap could be a good placeholder for something that Noctua, Scythe or Thermalright or even Alpenfohn might release in 2021. However, it would still need to be a clear upgrade over my existing two heatsinks — TRU (non-extreme, hence not TRUE) and Gelid Phantom (looks like Fuma and has seven pipes but isn't as good as it looks but still can't really be all crap if it used to be hailed as the king of budget cooling in my neck of the woods).

2 hours ago, freeagent said:

I have to ask, just how much paste are you using? When people say the size of a pea, disregard what they have to say after that. That is a huge amount. A couple of grains of rice and that’s it, maybe not even that much. I can’t speak for AMD since I haven’t run with them since 939 days.

As for my system, it’s not silent, but it’s not loud either unless I want it to be heheh.

Hang in there man, the experience will come..

Krynoaut. Decided it was safer than other Grizzly pastes in my hands, but I still spilt it over the CPU slot on my Aorus Pro. I also have MX4 (but it may be very old) and Silver 5 (impulsive and regretted purchase of a large amount, but hey, it's a lot of paste in case someone wants to borrow some again ;)). Should also have some GC Extreme somewhere, but one tube is new and one is like 5 years old, can't tell which is which. For AS5 I used the rice-grain method, X for MX4, thin spread for GCE (hate that method with that paste), thin spread for Kryo (same feelings about it).

939… I remember those days. Had a s478 Thermalright cooler that had the then-huge 400g of copper and 92mm fan. Had to rotate it some 30 degrees and use spacers (my pal and I cut up an old LAN cable to get some rubber), but it worked. Chances are the air entered higher than it exited, me being a noob at the time. But it worked. No case fans. Just CPU, GPU and PSU. Those were the days. I later had that gold-plated Aerocool thing — 750g of gilded copper plus two 92 fans with manual knobs for rpm control.

***

Edit:

Look what I found — Ed Hume's review of Fuma 1 shows just 1.5 dbA above ambient while providing a very good temperature, less than 3 degrees worse than two 2 fans producing a round 10 dbA more. This gimmick sounds like my thing. I once saw a similar experiment on the D15 with similar but less good results. So perhaps there's something unique about Fuma 1's heatsink or its Kaze Flex or whatever fans it uses, which might amplify the results of running three-fan, but I guess I could grab one of the industry's best naked heatsinks (D14, Silver Arrow, TC14PE) and strap 3xP14 to it.

… Or I could buy and use a third SW3 hispeed, since I already have two.

EDIT 2: Might not be the best idea to do this with Fuma 2, after all. Fuma 2 does tend to beat D15's noise-temp balance on less power-hungry processors, but looking at this review of it makes me think a D14-size heatsink with 3x140 fans could be a better idea.

NewbieOne · September 24, 2020

Hi, guys, another update. I've done more 'research' (isn't really worth its serious-sounding name) across the various interviews, and focused on these:

Tweaktown's review of Fuma 2
HWCooling.net's putting industrial Noctua F12 iPPV on Ninja and comparing with D15
Tom's Hardware's review of Ninja 5 focusing on powerful overclocked systems
Tweaktown's review of Ninja 5
(but also analysed a bunch more)

I focused on Ninja 5 and Le Grand Macho RT but included Fuma 2 because it sometimes manages to overtake them on performance, noise or both (narrow situations benefitting this or that cooler's construction) and included D15 because it's pretty much the reference cooler if not necessarily always the best choice, and even if it's not best, it's like always very close to the top.

So the results I got is that for something like 6700K (in the tester's system, yes, I know) the Ninja will totally own — at max fans it's going to be like shaving off 5 or 6 dBAs at the cost of running 2 or 3 degrees hotter. This is exactly the tradeoff I have in mind. The problem is that 9600K or especially some future 11700K in my system and my house vs the tester's CPU, PC and office will likely, while not being worlds apart, still cointain at least a handful of at least minor differences, shuffling the 'order of precedence' among the coolers quite a bit.

I can't claim these to be valid conclusions, more like somewhat (hopefully not to un-)educated guesses, but it seems that:

The Ninja rocks the boat for non-extreme but high OCs and loads.
At the less extreme end Fuma 2 will be quieter for the same temps, and LGMRT also like will be. As in, Ninja 5 can't reach the lowest of the low sound scores achieved by Fuma 2 and LGMRT.
At the super extreme end the Ninja won't exactly crap out, but you'll see it overtaken by D15 and Macho, or even DRP4.
The Macho kinda seems stronger overall, just losing by a notch to Ninja 5 in certain narrow situations.
Incidentally, Ninja's max rpm sound pressure, which is lower than on the other coolers, seems to be vaguely similar to the pump on Liquid Freezer II 280 (though at that point the latter's fans could perhaps be louder), so perhaps water shouldn't be off the table after all, especially being cheaper than LGMRT or D15 (although Ninja 5 is still cheaper by some margin), although I shudder to think of the flow-taxing and sound-amplifying effect of pushing through a rad, filer and case mesh or pulling through a rad, filter and case mesh.
Noctua's heatsink tends to win with the same fans, plus D15 cools mosfets/VRMs better, probably due to the oversized fan, which may be an important consideration in silent OC systems. Worth noting is that D14's heatsink often comes out on top of D15's with the same fans in tests. PHTC14PE also sometimes does, also also the original Silver Arrow. Such tests being pre-P14, I suppose P14 with its static pressure could make some difference blowing through thin stacks.
If the A15 on D15 is responsible for the better VRM cooling, then the LGMRT presumably can pull the same off with the TY fan.
Flimsy single-tower coolers with not too many heatpipes sometimes get surprisingly good results with proper ventilation. Doesn't really seem to be the case for my Ultra 120 with 2x140 Silents Wings 3 HS. Perhaps high mounting pressure and direct heatpipe touch make a difference.