Thermalright Outs Assassin X 90 SE Compact Tower-type CPU Cooler

[btarunr]

Thermalright today introduced the Assassin X 90 SE, a compact tower-type CPU cooler designed to replace the stock CPU cooling solution with something that's quieter at stock frequencies. The aluminium fin tower-type cooler features a typical design for its kind—four 6 mm-thick nickel-plated copper heat pipes make direct contact with the CPU at an aluminium base; with their ends passing through an aluminium fin-stack that's capped off by a die-case top-plate. This is then ventilated by an included Thermalright TL-G9B 92 mm PWM fan.

The defining feature is the cooler's height of just 118 mm. With its fan in place, the cooler measures 94 mm x 77 mm x 118 mm, weighing 445 g. The included 92 mm fan features fluid dynamic bearing, takes in 4-pin PWM input, turns at speeds of up to 2,400 RPM, pushing up to 43 CFM of airflow at 1.51 mm H₂O pressure, and has a noise output of 27.3 dBA. Among the CPU socket types supported are LGA1700, AM5, AM4, and LGA1200. The company didn't reveal pricing.



View at TechPowerUp Main Site

 

[P4-630]

I see a review coming up from @crazyeyesreaper ...

 

[DeeJay1001]

Looks like a solid contender to the nh-U9s. I like the 92mm tower coolers for budget builds based on i5/Ryzen5 chips. They don't really need a massive 120mm tower and the 92mm size makes case compatibility and Ram clearance much less of a concern.

 

[ymdhis]

Now where have we seen this design before, hmmm.

 

[bonehead123]

Now where have we seen this design before, hmmm.

Like maybe, just maybe......... E*V*E*R*Y*W*H*E*R*E*

 

[Chrispy_]

Good, I'm pleased to see that the fan is as low as possible, making contact with the mounting screw.

There are too many 92mm coolers that aren't short enough because they waste space below the fin stack. IMO 92mm coolers that aren't shorter than the shortest 120mm coolers are completely pointless. You'd never pick one when you can also fit a 120mm tower in the same place...

 

[katzi]

Now where have we seen this design before, hmmm.

In pretty much every tower-style cooler ever?

 

[Mussels]

the shaved back heatpipe design just makes me cringe these days, solid base plate or no bueno

I feel like you'd be lucky to even get one heatpipe lined up on your CCX with AMD CPUs, since you usually cant rotate these designs you'd be screwed

 

[Chrispy_]

the shaved back heatpipe design just makes me cringe these days, solid base plate or no bueno

I feel like you'd be lucky to even get one heatpipe lined up on your CCX with AMD CPUs, since you usually cant rotate these designs you'd be screwed

Why? The IHS is effectively a base plate.
Sure, a baseplate adds a bit of thermal mass but it's not as if these are coolers aimed at 200W PPTs...

 

[Mussels]

Why? The IHS is effectively a base plate.
Sure, a baseplate adds a bit of thermal mass but it's not as if these are coolers aimed at 200W PPTs...

the gaps between the heatpipes
I made a big post about it in another thread, but lemme whip up another horrible MS paint example...

They transfer heat best basically directly in the center of the pipes - with the gaps between the pipes being the worst possible heat transfer

You can see how that can result in a cooler where only one of the 4 heatpipes is even contacting the CPU Cores in use, with users bitching about high temps at low loads, high idles and such - because maybe 2 of their 8 cores are even cooled directly, with only a single heatpipe doing it's full work crippling the cooler

A lot of these designs dont let you rotate the cooler, whatever direction they chose is what you're stuck with


So yeah, yay - at best 2 of the 4 heatpipes are cooling the actual CPU cores with the other 2 working on whatever heat they transfer sideways, if any

 

[Chrispy_]

the gaps between the heatpipes

Has this ever been tested, scientifically? I have serious doubts that your concerns are warranted, and heatpipe coolers are applied to the bare die in the GPU world all the time. If there was a "dead spot" between direct-contact pipes, they'd not be used in bare-die applications like laptops and GPUs - yet that's how it's been in the laptop and GPU industry for almost two straight decades.

Heatpipes are phase-change devices; In a non-faulty one, the rate of water evaporation at the hot end is the limiting factor, provided the cold end is cooled adequately. The reason for this is pretty easy to visualise but the system reaches equilibrium when (ΔT above evaporation point)*(hot surface area)=(ΔT below condensation point)*(cold surface area). Since the length of heatpipe covered by fin-stack is FAR greater than the tiny hot-spot above the die, it's the evaporating end that we need to look at, and specifically the hot surface area of the internal sintering of the heatpipes.

Here's a cross-section of the join between two heatpipes in a continuous direct-contact heatpipe cooler:

That "dead spot" as you call it, actually has far more internally-sintered surface area above it than the middle of the pipes, which are effectively just flat planes of minimum surface area. The heat is conducted by the copper walls (and filler solder in this example, but often they're just smooshed together and ground flat) up to two walls, which means double the internally-sintered surface area. We know from the above equilibrium equation that it's going to be the internally-sintered surface area of the hot end that matters the most, so if anything, the gaps between heatpipes should have a cooling advantage, not be a "dead spot".

That's the science and maths of a simplified model. In the real world, I suspect many other factors serve to muddy the waters and the differences all come out in the wash; If it's relevant, at all. Widespread, almost exclusive industry adoption of this method for mainstream GPUs and laptop CPUs would indicate that no, it's not relevant.

As for the orientation thing:

Yeah, spreading the load over more heatpipes is definitely better, since even though the heatpipes themselves aren't being overwhelmed, more heatpipes uses means that the thermal load is being spread out over more of the fin stack. Adding a baseplate increases thermal mass slightly which is good for bursts of high load but also increases the distance of a temperature gradient which is bad for keeping temperatures down as low as they can be with heatpipe technology. The thickness of the baseplate does serve to dissipate heat over a wider area, but it's not like a baseplate can magically transfer heat sideways very far. If the baseplate is 2mm thick, it will transfer heat about 2mm further away from the die edge than a direct-contact heatpipe, that is all - because heat is like water and electricity in that it always takes the path of least resistance.

 

[trsttte]

the gaps between the heatpipes
I made a big post about it in another thread, but lemme whip up another horrible MS paint example...

They transfer heat best basically directly in the center of the pipes - with the gaps between the pipes being the worst possible heat transfer

View attachment 273501


You can see how that can result in a cooler where only one of the 4 heatpipes is even contacting the CPU Cores in use, with users bitching about high temps at low loads, high idles and such - because maybe 2 of their 8 cores are even cooled directly, with only a single heatpipe doing it's full work crippling the cooler

View attachment 273502


A lot of these designs dont let you rotate the cooler, whatever direction they chose is what you're stuck with


So yeah, yay - at best 2 of the 4 heatpipes are cooling the actual CPU cores with the other 2 working on whatever heat they transfer sideways, if any

Isn't the thermal paste there to fill those gaps?

I'd be curious to know if these direct heatpipe type coolers don't perform better than having a cold plate (which is doing pretty much the same as the IHS) since it's one less resistence layer added on the heat path

 

[Mussels]

Isn't the thermal paste there to fill those gaps?

I'd be curious to know if these direct heatpipe type coolers don't perform better than having a cold plate (which is doing pretty much the same as the IHS) since it's one less resistence layer added on the heat path

thermal paste is far FAR less effective than plain old copper

Chrispy: Yeah its called reviews, and you can find this with the CM212 range: all their budget variants use the shaved off direct heatpipes and they all perform worse than the older solid plate models
and they also released a new model with the solid base as their new premium variant, which to absolutely no surprise performs better than the shaved pipe models



Modern CPU's have heat density issues, and these 100% make it worse


Let me put it this way: See the hints of silver gaps?
Would you buy a cooler from any other brand if it has scratches that deep horizontally across the cooler or be screaming for a refund?

 

[claes]

This is why GN does measures contact pressure and “coldplate flatness”

 

[Mussels]

This is why GN does measures contact pressure and “coldplate flatness”

Indeed, like with the 120mm version of this cooler

I need to save these images for the other threads i'm discussing this in, you can literally see the lines of contact and the poop spaces between the heatpipes
yes you could try repeated mounts over and over, but it's super easy to imagine a CCX or part of one stuck in one of those huge gaps

They dont bag it out because its a super cheap cooler, but it's not great

These type of coolers are great at transferring heat FAST, but once the wattage goes past what a single heatpipe can transfer, performance tanks and they go from leading charts to losing them

They're fine for 65W CPUs (actual 65W, not the 10700)


Lower wattage: peachy! because even contact with just the one heatpipe gets the job done

but you increase that wattage and they fall behind, and this is on a dual CCX design now - a single CCX chip would be absolute ass

Are they a good budget cooler: yes
Are they great for low wattage CPUs: absolutely

Is it worth being aware that this design is why the web is flooded with posts like "why 5800x so hot" posts? absofeckinlutely.

 

[Grunchy]

I think there might be some basic misunderstanding about how coolers work to conduct heat away from the hot surface.
The heat pipes are copper tubes that have some pure distilled water inside of them.
That water gets into a convective flow, as the water heats up it rises away from the CPU surface, meanwhile cooler water is flowing back from the fins.
If that convection wasn't the most important transport mechanism then you could simply replace the heat pipes with much simpler solid copper rods. The manufacturer would save the money & get better performance, but nobody does that which tends to prove that the convection of water really is where the performance lies.

So to get optimum convection, you glue 4 heat pipes together and machine a flat face into them, that diminishes the conduction path between CPU and water.
There really is no better performance possible than the shaved heat pipes, because it provides the shortest distance between CPU and the cooling water.
So the function of the pad of metal is to hold the pipes steady so they can keep that flat face & provide even clamping pressure over the CPU die.
(Somebody pointed to some endurance heat tests, I've got news for you: after a few hours of such a test every single cooler becomes saturated with heat, no matter how big your heat sink is makes no difference no more. That's kind of the point of doing the endurance heat test?? I really hope that point isn't lost on you. A bigger "heat sink" does absolutely nothing in that circumstance.)

If you want better performance you would "lap" that machined surface to get it really smooth. I think the shaved and lapped heat pipe is the only thing that is better than the shaved heat pipe.
Next issue is how many heat pipes the cooler has & whether they branch out as a single tower or a double tower.

After that is the question of how well do the fins and fan blow that heat off the heat pipes, obviously the bigger the fins and fan the better, but 92mm diameter is what fits in my PC case so that's what I'm using.
Furthermore: you can make the cooler work better by just running the fan harder. It makes more noise but you know what? It's the 92mm fan that fits inside my case.

 

[claes]

I left it alone because it’s a complex issue and, as far as I’ve read, there are no actual studies (I found one, but it uses TIM with the baseplate test and runs bare with direct contact, assumes perfectly paired/flat surfaces, doesn’t assume something like an IHS from the heat source, etc), but I’ll play ball

I think there might be some basic misunderstanding about how coolers work to conduct heat away from the hot surface.
The heat pipes are copper tubes that have some pure distilled water inside of them.
That water gets into a convective flow, as the water heats up it rises away from the CPU surface, meanwhile cooler water is flowing back from the fins.
If that convection wasn't the most important transport mechanism then you could simply replace the heat pipes with much simpler solid copper rods. The manufacturer would save the money & get better performance, but nobody does that which tends to prove that the convection of water really is where the performance lies.

So to get optimum convection, you glue 4 heat pipes together and machine a flat face into them, that diminishes the conduction path between CPU and water.

Modern designs use copper powder and the like, but glad your first post begins with a condescending lecture — welcome to TPU!

There really is no better performance possible than the shaved heat pipes, because it provides the shortest distance between CPU and the cooling water.

In theory, sure, but in reality neither the cpu IHS or the heatsink is perfectly flat. Most IHS are concave, which is why the best performing heatsinks usually have a convex base. Ideally you’d just solder the heatsink to the IHS, but that’s not practical for obvious reasons. This is why companies like Noctua don’t use perfect mirror finishes, but instead have small grooves to make up for imperfections on the IHS, with the assumption that thermal paste is being used.

So the function of the pad of metal is to hold the pipes steady so they can keep that flat face & provide even clamping pressure over the CPU die.

Huh? They’re soldered to the base plate, just like they are with exposed pipes, for the reasons you mention in both cases. AFAIK, base plates are used in the best performing designs for the reasons mussels mentioned — there are too many inconsistencies in manufacturing (and due to ILM pressure in the case of IHS) on both surfaces to ensure a perfectly flat surface — and, probably more significantly, because a CPU die is typically much smaller than it’s IHS. Base plates are much more consistent (and cheaper) to manufacture than exposed heat pipes (of similar consistency), and, AFAIK/can guess, can have the added benefit of exposing more of the heat pipe’s surface area to the plate, which itself distributes heat across a larger area of the pipes. Even better than both are vapor chamber base plates, but they’re cost prohibitive and the advantages marginal.

If you want better performance you would "lap" that machined surface to get it really smooth. I think the shaved and lapped heat pipe is the only thing that is better than the shaved heat pipe.

That’d be a bad idea because the IHS is concave. You’d want to lap both surfaces for even contact. Typically extreme OC’s lap the die and leave the cooling device as is, but some go the extra mile and lap the concave surface of the cooler (typically pressure makes up for the loss there, but I’ve seen martinsliquidlab and ehume do it to some benefit). Better yet, just delid the CPU and make direct contact.

 

[Mussels]

heatpipes are used because they're cheaper than solid copper, not because solid copper isn't better

 

[claes]

Don’t heat pipe’s have 250x conductivity?

 

[Mussels]

Don’t heat pipe’s have 250x conductivity?

so do those carbon pads to be used in place of thermal paste, yet their performance is far worse since the heat transfer is only in a specific direction


Heatpipes are great, they work well and they're cost effective. They're just not magically the best thing ever in all categories, just as an overall solution.
Heatpipes can have too much heat applied and just stop working, since the liquid cant cool back down - it stays boiled up the top away from the heat source as an example

 

[claes]

Isn’t that why heat pipes are tailored for specific loads? Not being facetious this time I promise Why wouldn’t Noctua or similar just charge $200 for a heatsink with rods if the performance were better?

Aside, I was actually asking myself the same question about direct contact vs base plate coolers after re-reading this article I had totally erased from memory about direct contact’s performance on lapped CPUs — why add a base plate if you can achieve great temps with direct contact? This isn’t arguing against my agreement with you about direct contact but just a curiosity

Our Lapped CPU Heatsink Test Platform

Lapping the CPU in a heatsink test platform is probably a controversial move that’s bound to provoke reactions. Funny thing is, it was done a year ago, and photos of the CPU showing the copper top exposed by the lapping have been featured in many of our reviews. Yet, not a single comment. This...

silentpcreview.com

Think I answered my own question re: heat pipe conductivity/size vs copper in and of itself:

https://d1wqtxts1xzle7.cloudfront.net/47592637/2016_Analytical_Expression_of_heat_pipe-libre.pdf

above link might be broken: https://www.sciencedirect.com/science/article/abs/pii/S1359431116301855

https://en.m.wikipedia.org/wiki/Darcy's_law

HEAT PIPES

www.thermopedia.com

https://link.springer.com/content/pdf/10.1007/s002310050097.pdf