[SOLVED] Front fans for Meshify C ?

[hornetmax]

Hi all,

I've got a Meshify C with a 2070 super and a 9900K, no overclock. Cooling-wise, I've a Thermalright Macho RevB, 2x Corsair AF140 front (intake) and 2x Fractal 120mm fans out (rear and top-rear).

Since the 9900K (had a 9600K before), CPU (and VRMs) temps go a bit high during heavy loads (e.g h265 encoding): VRMs hitting 98C, CPU cores overing at 88C, hitting (very occasioanlly) +95C and causing (again, very occasionally) thermal throttling. It's not a huge problem, but I'd be more comfortable with lower temps on both CPU and VRMs.

While under full load, I removed the front and top filter of the Meshify C (metal grid + foam): that gained me 6C on the CPU and 15C on the VRMs (!!).

Now I'm wondering: what shoud I do to have more flow from the two front fans ? Options I see:

1. Take the foam away from the front filter (keep only the metal grid).
2. Take the foam away from the front filter (keep only the metal grid) but put some good filtes on the front fans: things like Silverstone F143 or Demci Flex.
3. Swap the two front Corsair AF140 in the front for something else ? Higher static pressure to "overcome" the foam ? Which fan ?

Option 1: just a matter of judging fs the dust is manageable without the foam.
Option 2: I'm unsure how much better this would be compared to the foam. Any idea ?
Option 3: I'm OK with the fans being a bit loud if spinning fast, but at low loads I'd like them to be quiet (i.e. similar to the AF140 I have).

Any advice is welcome !

 

[hornetmax]

Just saying, I did 3x 3h00 h265 encoding test (plus a bunch of quick 10-20min Prime95 tests to check stability and top temps).

 

[Karadjgne]

But you understand what pulling those top vents is doing? That's unfiltered fresh air almost direct to the cpu cooler. It's fine to pull the top-rear, especially if you put a fan there as exhaust. Improvements to air Flow are what's going to be the better balance of temps now that you have fans that'll actually move it.

I've run p95 for over 8hrs on an i7 at 4.9GHz in a fractal design R5 with the door closed and didn't top 72°C, and thats a considerably worse airflow case than the meshify can be.

It's just a matter of getting the flow right for your equipment.

 

[hornetmax]

But you understand what pulling those top vents is doing? That's unfiltered fresh air almost direct to the cpu cooler. It's fine to pull the top-rear, especially if you put a fan there as exhaust. Improvements to air Flow are what's going to be the better balance of temps now that you have fans that'll actually move it.

Well, even with my old fans removing the front filter (or the front and the top one) made the temps go down very significantly.
The new fans don't seem to bring anything. I'm not complaining, just reporting back. No big deal of course.

I've run p95 for over 8hrs on an i7 at 4.9GHz in a fractal design R5 with the door closed and didn't top 72°C, and thats a considerably worse airflow case than the meshify can be.

Well that's manual OC, I'm running stock. And it's an i7 not a 9900K. And I don't know if you run prime95 with small FFT and AVX ... becasue without AVX, of course temps are way lower. And then, all the CPU are kind of different so yeah, hard to compare.

It's just a matter of getting the flow right for your equipment.

 

[Karadjgne]

It was an i7-3770K, so a 1GHz OC, p95 small fft 26.6, so no AVX. Pushing @ 180w out of a 77w cpu.
Gaming never went over 55°C. 280mm AIO.

So yeah, hard to compare exactly, but the wattage is close enough to a 9900k @ stock, the aio was front mount and only 2x exhaust on top. That gave me optimal flow, chimney style, nothing going sideways or circular, air in low, and straight up and out the top. Adding a rear exhaust actually increased temps 2°C because it lessened the straight up pull and changed the airflow pattern. Leaving it open was just as bad for the gpu as there was less pull from above as that rear hole became an intake feeding the top-rear fan.

Having the intakes at 1100rpm in push got the same temp as 700rpm in pull, but pull dropped the gpu temps 6°C because of the direct air feed not being hindered by the rad.

Bunch of nickle and dime temps, but they all added up to $1 in the end, getting a better balance of case airflow vs cpu and gpu temps overall.

And that was with some mediocre stock fans. If I'd had decent fans, that would have changed noise outputs, lowered rpms, and probably dropped the temps another 2 or 3°C in the process, an overall benefit everywhere, especially to my ears.

So there's give and take in everything, I could have dropped all the temps another 4°C by ramping the fans to max, but the temps were good enough as was to suit the needs, and my wants.

 

[Darkbreeze]

Well that's manual OC, I'm running stock. And it's an i7 not a 9900K. And I don't know if you run prime95 with small FFT and AVX ... becasue without AVX, of course temps are way lower. And then, all the CPU are kind of different so yeah, hard to compare.

This is your problem. You CANNOT run encoding or stress test applications with full on AVX, AVX2 or AVX512 enabled or you'll NEVER be able to remain thermally compliant. If you are running applications that you know will make extensive use of AVX instructions, and you want to be sure you are remaining within the thermal envelope, then you NEED to configure an AVX offset in the BIOS, because your 9900k is never going to manage full AVX operations especially when you don't have a CPU cooler that is capable enough for the 9900k, which that Macho rev.B is probably not. And I'm pretty sure I've said that before.

If you are going to run AVX extensively, AND insist on remaining on air cooling with it, then you are going to continue to run in circles until you install a more capable cooler like one of these.

Noctua NH-D14 (Replace stock fans with NF-A14 industrialPPC 2000rpm)
Noctua NH-D15/D15 SE-AM4
Noctua NH-D14 (With original fans)
Thermalright Silver arrow IB-E Extreme
Cryorig R1 Ultimate or Universal
Be Quiet Dark rock Pro 4
Noctua NH-D15s
Thermalright Legrand Macho RT
Phanteks PH-TC14PE (BK,BL, OR or RD)
Deepcool Assassin III

 

[Owterspace]

True Spirit 140 Power is also good for 50 bucks, as good as all of the coolers mentioned in the post above mine.

 

[Darkbreeze]

True Spirit 140 Power is an even weaker CPU cooler than the Macho Rev. B he has now. It would probably be advisable, for yourself and for this forum, if you took the time to learn which CPU coolers are suitable for which CPUs (True Spirit 140, direct or power, are definitely NOT suitable for a 9900k) and which coolers are better or worse than others, in general. Recommending the True Spirit 140 power for a 9900k, just looks bad and extremely lacking in experience in this area, for you.

So, not trying to shoot you down or insult you in any way, but clearly you need to gain a lot of knowledge in this area before you start shooting from the hip and throwing recommendations out there. Because you might do that in a thread where somebody actually believes what you're telling them and then goes an spends REAL money on real parts just to find out that it was bad advice that won't even remotely cool their device under normal conditions MUCH LESS while running extensive AVX workloads.

 

[Karadjgne]

The 9900k at just turbo running 100% all core, stock, could hit right at 200w. Just locking the cores at the 5.0GHz turbo put that at 250w.

That means you'll be hitting limits with any aircooler, any aio of 240mm, most custom loops with a single 240mm radiator..

Which means temps in the 90's at those kinds of loads, and breaching 100°C if trying to keep those speeds and Any AVX use.

Any of the x900 series cpus planning on high AVX usage should be on a 360mm AIO or at least a good 280mm custom loop at a minimum.

Air just doesn't cut it.

 

[Darkbreeze]

And they will almost certainly HAVE to configure an AVX offset in the BIOS. They only way you might get away with running full AVX, no offset, with the 9900k at full stock turbo boost profile specifications is with a rather large custom loop. Otherwise, you are GOING to have to offset for AVX. There just is no way around it. You can't run full AVX on the 9900k with air or smaller water. MAYBE, MAYBE, with a very good 360mm or larger AIO, but I doubt it, since custom loops offer much higher end cooling potential.

 

[Owterspace]

You are thinking of the TS140 Direct, if you aren't, then you have no idea what you are talking about. I own TS140P and LGMRT. Both are better on the same hardware then D14. But its all good man, think what you want.

 

[Darkbreeze]

Yes, I do have an idea of what I'm talking about. A very good idea in fact. And we have several members here who've put it to the test against larger heatsinks to prove it, without any lab conditions. Under lab conditions, it's a little better, but not that much. It is in NO WAY as good as or better than the D14 or D15. No, it isn't. YOU can think what YOU want, but it honestly doesn't change the outcome ESPECIALLY when it's being used on something with a lot of cores and a very high TDP.

Besides which, even in reviews, it fails to beat the D15 by about 8 degrees, and other coolers (Even a couple of much older ones) beat it easily, by a wide margin, in the few reviews of this cooler that are actually out there. And none of what IS out there is by anybody doing top notch reputable normalized testing anyhow. So it must be taken with a grain of salt in any case.

https://www.nikktech.com/main/artic...ue-spirit-140-power-cpu-cooler-review?start=5


But the results from user feedback is even worse than that would suggest.

 

[Owterspace]

Ok .. Well I have owned all three coolers, and that was my experience. I'm not going to argue with you about it.

 

[Karadjgne]

The best cooler Thermalright ever put out to the public was the Silver Arrow, but that's ancient now so good luck finding one. That cooler would go head to head with the Noctuas, but to do so basically required a double wide trailer for a case.

Point is, cooler wattage and cpu wattage and temps are all interlocked. And not on a 1:1 basis. 1:1 means at 100% you'll basically see 100°C. So the ideal cooling solution is 1.5-2 :1 ratio, where cooler wattage is closer to double (if not more) than cpu output.

That'll keep full loads in the 60-70°C area and still allow for mild AVX use. If you plan on Blender and higher AVX use, plan on overcompensating not only capacity, but efficiency. 9900k @ 5.1GHz locked core = 500-600w capable custom loop. No ifs, ands or butts. Or you will see temps looking at 80's+ with anything lesser.

All of the 900 series cpus, 9th or 10th Gen, are no joke for effective cooling, not when it comes to OC. It's simply a matter of go Huge or go home, because even Big doesn't cut it.

Gaming is chump change compared to Blender, folding @ home, or most other high core, high workload outputs.

 

[Owterspace]

I like Linpack Xtreme.

Le Grand Macho RT is pretty decent:

https://www.techpowerup.com/review/thermalright-le-grand-macho/6.html

And my TS140P is within 2-3c of my LGMRT. I could run my 3770K at 4500 with a fixed vcore of 1.25v and run linpack with no fan on the LGMRT

Its a good cooler. I cant do that with TS140P. But then again its not rated for 90w TDP passively like LGMRT is.

New project:

Tomorrow I will remove LGMRT and install TS140P and see what it can do semi passively with my 3600XT.

 

[Phaaze88]

Lovely... Someone's boasting about Thermalright coolers with low power draw cpus.
Here we go again...

So, a 6700K in the TPU review of the LGMRT, Owterspace's 3770K with TS140P, and this old leak with an SB-E on a bloody 4670K that someone bragged about, along with other TR coolers that 'beat Noctua' in a discussion of TPU's review of the CNPS20X...
My 7820X says no: Noctua still wins with ACTUAL high power draw cpus.

I tested NH-D15S, TS140P, and IBE-E Rev. B. 7820X was delidded by Silicon Lottery - it was OC'ed to 4.5ghz core, 2.7ghz cache, at the time.
I ran Cinebench R20 for 30mins to test thermals - it might not be quite as hard as P95, but thanks(?) to the OC, I was able to push 90C+ scenarios at times during testing.
TS140P is worse than both of the other coolers on this cpu, even when they're all on the same playing field. By playing field, I mean:

A)I took the TY-143 fans from the IBE-E and tested them on the other coolers. I also replaced all the chassis fans with IPPC 3000s. This cooler was meant to work with other high rpm fans; it actually does worse if the chassis fans can't match the speed of its own.
NH-D15S > IBE-E >> 140 Power. D15S beat IBE-E at its own game.

B)NF-A14/15 fans on all coolers. Took out the IPPCs and put regular Noctuas in.
Same deal as above.

C)Single fan test - TY147A on all coolers for the hell of it. No change.

The 140 Power really dragged its ass compared to the other 2. The IBE-E would be a few degrees behind the D15S, but the 140 Power would be 10C or more behind both.
I went though quite a bit of NT-H2 remounting it several times out of disbelief.
The common denominator appears to be the D15S's heavier heatsink. Even though the TS140P(725g) has those thick-ass 8mm pipes, it's still lighter than the other 2, which are 850g and 980g.

Oh, and here's a link to an older(before I had the TS140P) Overclock.net thread I'd created trying to figure out why the IBE-E wasn't 'killing' the Noctua: https://www.overclock.net/threads/t...e-rev-b-vs-noctua-nh-d15s-w-2-nf-a15.1751902/
[Unfortunately, the imgur links are broken. I inevitably deleted them - it was 6 months ago.]
I knew Thermalright coolers were popular with the folks over there, so they'd be able to help me get to the bottom of this...
It ended with the acceptance that the Noctua was handling high power cpus better - it then started going into discussion about my gpu cooler, LOL.

After all that:
-the 360w TDP BS, even though I know better regarding that, but I was still curious.
-the people talking about how both of these Thermalright coolers were 'Noctua Killers'... yeah, maybe with low power cpus, that's probably true. A 7820X doesn't 'purr' compared to a 6700K, 3770K, or a 4670K.
Yeah, I was disappointed. It wasn't even close.


A number of the cooler reviews out there - especially older ones - are flawed because their test systems are running kittens, when they should be using lions.
The Nikktech review Darkbreeze linked is pretty legit: a 3930K isn't a kitten.
Tom's recently started using a 10850K in cooler tests: a lion. Before that, it was a 5930K: another lion.

 

[Phaaze88]

You must be the same Owterspace in the thread I linked? Hello!

You expressed your opinion on select Thermalright coolers: Good. I don't have a problem with that - in fact, I agree with you there.
You were also stating facts from your own experience + posted a link to the LGMRT, as if these coolers are good everywhere, but there is a flaw: the cpus being used aren't very big on power.
I shared my experience in that these coolers are not as good as Noctua with cpus that pull a lot of power, and the 7820X is in another league from those 2 cpus.
I was really interested in the possibility of tossing Noctua to the curb and replacing my NH-D15S with one of these.
Perhaps if I'd gotten the LGMRT, thing's would've been different? I have my doubts after testing 2 of TR's 3 extreme high end cpu coolers. It's a little heavier(900g), but still less than a D15.

Mod Edit - Removed quote

 

[Owterspace]

Its all good man. It felt like I was being backed into a corner. Yes I do have a couple of light weight CPU's. I also have my X5690 ES. That CPU can easily pull 250-300w+ when you wind it up. Its what I was going to use for testing the Ice Giant before I got cold feet. My dinky little 3600XT and 3770K are nothing. My system running something like Linpack consumes less juice at load then that one does at idle lol.

Edit:

And that is why my x58 is in a box

Edit again

I'm sorry for my rude comment.

 

[Darkbreeze]

Indeed.

 

[Phaaze88]

Its what I was going to use for testing the Ice Giant...

Ooh, do let us know in another thread or something on how that goes. I was curious about that one, but I had already bought an Eisbaer Aurora 360.