Whats the point of a CPU liquid cooler that just sit on top of the CPU?

[Mtt]

I stumble upon this weird CPU cooler.
http://asetek.com/customers/workstations/hp/hp-z420-workstation.aspx

To my knowledge, the whole point of using liquid cooler is to have a larger cooling surface area and dump heat outside by transfering heat using liquid.

But this cooler just sit on top of the CPU.
Wouldnt heat pipes do a better job of transfering heat in this case?

A clearer picture of the cooler inside the case

 

[krnmastersgt]

Just utilizing water instead of standard heatpipes on some of the larger air coolers to transfer heat to the fin array. One benefit is how they can design the fins and pack them together but I'm not entirely sure what the complete pros/cons list would be, you'd have to ask one of the companies that's build both types of coolers to see what merits it can truly have.

 

[Topweasel]

Yeah I think krn has it right. Normal heat-sinks require the thermal compound in the heat-sinks to heat up to evaporation where it would travel up the heat pipes to the fins, then have the fans/fins dissipate/cool off the vapor so that it falls back down the pipes. This would be actively pushing that thermal compound (mostly water) around to the radiators.

Have you seen most work station and Desktop coolers from system builders like Dell and HP? They are huge and usually don't have attached fans. My guess is HP when creating a mid tower LGA2011 workstation, found that the cooling properties and noise level of this unit, were better then they would get from a standard HSF, that because of the memory location and compactness of the board their usuall HS wasn't going to work, and that it beat designing a very specific HSF for this system (or relying on very expensive retail channel parts).

 

[coffeejunkee]

The only advantage I see is that the memory slots aren't blocked. Then again, they are covered by all kinds of plastic covers so might still be hard to get at.

I'd like to see this cooler tested though. My guess is it will be beaten by a decent air cooler but would like to know for sure.

 

[mv2devnull]

The "plastic covers" are "ducts that tunnel the air by the memory". Allegedly.

In my experience, this Z420 cooler is way better than a model on some Z400's, where the radiator and fan actually are at the back of the case. A knit shirt is more smooth and reflective than the water block on the latter.

 

[dave_the_nerd]

The only advantage I see is that the memory slots aren't blocked. Then again, they are covered by all kinds of plastic covers so might still be hard to get at.

I'd like to see this cooler tested though. My guess is it will be beaten by a decent air cooler but would like to know for sure.

I remember Swiftech marketing something similar years ago. (A self-contained water-cooler that stayed inside the case, in more or less the same form factor as a large-ish HSF.)

IIRC, it performed well compared to most HSFs, but was beaten out by a few models. OTOH, (again, iirc) those models were more expensive.

 

[dma0991]

I find this method nullifies any benefit of watercooling, if typical AIO isn't crippling already. The radiator should have been placed at the back of the case, ensuring that any heat exhausted will not be recycled or increase the internal case temperature. Plus, this relieves much of the weight that the socket mounts have to bear and transfer that load(radiator) to the case.

According to its whitepaper, the reduction in noise was only 2 dBA to a comparable air cooling solution. For all its complexity, lesser reliability(pump failure) and higher cost, I'd rather implement a CM Hyper212 Evo with a slower fan.

 

[Topweasel]

I find this method nullifies any benefit of watercooling, if typical AIO isn't crippling already. The radiator should have been placed at the back of the case, ensuring that any heat exhausted will not be recycled or increase the internal case temperature. Plus, this relieves much of the weight that the socket mounts have to bear and transfer that load(radiator) to the case.

According to its whitepaper, the reduction in noise was only 2 dBA to a comparable air cooling solution. For all its complexity, lesser reliability(pump failure) and higher cost, I'd rather implement a CM Hyper212 Evo with a slower fan.

I think your missing a couple things including its compactness. And Evo or something similar would cover the memory modules. You don't want your techs to have to remove a HSF/CPU just to replace defective memory.

OEM cases are all about tunneling air. They usually have very specific boards built for them that allow for access to the memory while using a massive HS and channel the air from a fan from the front or connected to the baffle itself. It's obvious that this 2011 board is closer to a retail product. In this case going with a thin tower AIO (rather then have cabling they can't manage) they could still keep the system manageable by techs.

 

[dma0991]

I think your missing a couple things including its compactness. And Evo or something similar would cover the memory modules. You don't want your techs to have to remove a HSF/CPU just to replace defective memory.

Here's a picture of a CM Hyper212 Evo with a fan on a LGA2011 board. While it would have been true that an air cooler such as the Evo will hinder RAM replacement on a LGA115x board, this shouldn't be the case with a LGA2011 board on the HP Z420. The size of the LGA2011 socket would mean that RAM slots sit further away than it would on a LGA115x, providing ample clearance even for the closest slot.

OEM cases are all about tunneling air. They usually have very specific boards built for them that allow for access to the memory while using a massive HS and channel the air from a fan from the front or connected to the baffle itself. It's obvious that this 2011 board is closer to a retail product. In this case going with a thin tower AIO (rather then have cabling they can't manage) they could still keep the system manageable by techs.

I don't think this can even be considered tunneling air for this particular OEM model. The good ones from Dell OEM uses shrouds and server racks. The shroud and server rack both share a common similarity that they both channel air as straight as possible, which this HP does not. You obviously can make the air flow in a bend, but that is assuming that the shroud is sealed well enough with only an entrance and an exit.

Shouldn't be an issue with cable management when both the Evo(1 fan) and the HP's AIO needs only one 4 pin connector by default.

 

[Topweasel]

Here's a picture of a CM Hyper212 Evo with a fan on a LGA2011 board. While it would have been true that an air cooler such as the Evo will hinder RAM replacement on a LGA115x board, this shouldn't be the case with a LGA2011 board on the HP Z420. The size of the LGA2011 socket would mean that RAM slots sit further away than it would on a LGA115x, providing ample clearance even for the closest slot.


I don't think this can even be considered tunneling air for this particular OEM model. The good ones from Dell OEM uses shrouds and server racks. The shroud and server rack both share a common similarity that they both channel air as straight as possible, which this HP does not. You obviously can make the air flow in a bend, but that is assuming that the shroud is sealed well enough with only an entrance and an exit.

Shouldn't be an issue with cable management when both the Evo(1 fan) and the HP's AIO needs only one 4 pin connector by default.

I am not talking about this HP just Dell and HP Workstations in general. They tend to have mothboards designed specifically for them so that they can use a Tower HS, that doesn't block memory and then they shroud the HS and have a fan that is tunneled from the front to the shroud or at the very end of the shroud.

This unit uses a very retail design for CPU and memory layout. Probably HP's lowest and cheapest LGA 2011 system and using some variation of mobo manufacturers retail channel board. Because of that they wouldn't be able to use an EVO or any Tower cooler because it would kill access to memory slots. The AIO in the picture allows memory access, doesn't have thick hard to manage cables, and is probably quieter than a cooler that would support a 140w CPU that didn't extend over the memory. As for cabling I was refering to using a better AIO that would be mounted toward the rear. Sure they could get coolit or such to work out cabling that would work. But again this system seems to really push the "off the shelf" parts list to keep costs down. Probably because margin is low and production is low as well.

Edit:
Just looked at he response to the other user. You might be right on the single fan Evo setup. Though I would still say that this AIO would be a lot quieter. CM fans are some of the loudest I have seen.

 

[Mtt]

If memory access is an issue, they can have heatpipes extend to the left and have a tower like heat sink there, instead of the usual straight tower blocking memory access. Like how laptop cooler can make turns.

Edit: Never mind, I didnt notice there are memory sticks on the left also.

 

[krnmastersgt]

I think it's more linked to how dense they can make the Finn array for the cooler as using the water channels in a radiator allow for more fins per inch than a heatpipe based cooler. The fan that they stick on is going to have to be decent to warrant it but that's one other reason for easter over heatpipes.

 

[dma0991]

This unit uses a very retail design for CPU and memory layout. Probably HP's lowest and cheapest LGA 2011 system and using some variation of mobo manufacturers retail channel board.

There definitely are non-standard motherboard layouts for OEM desktops. However, this layout between the CPU socket and RAM slots won't differ much as the copper traces that runs from the CPU socket to the memory slots are usually as short as possible, depending on the size of the socket.

Its not necessary to have a heatsink the size of the Evo to handle such a CPU at stock. The tower heatsink that Intel produced is smaller than the Evo and should be adequate at stock settings, if clearance for easy memory removal is a must.

 

[Topweasel]

There definitely are non-standard motherboard layouts for OEM desktops. However, this layout between the CPU socket and RAM slots won't differ much as the copper traces that runs from the CPU socket to the memory slots are usually as short as possible, depending on the size of the socket.

Its not necessary to have a heatsink the size of the Evo to handle such a CPU at stock. The tower heatsink that Intel produced is smaller than the Evo and should be adequate at stock settings, if clearance for easy memory removal is a must.

No one is going to use an Intel HS for this. It also won't cool as well and be noisier. That's the point. It isn't a competitive for it's price cooler, but its better than 90% of the coolers out there, in cooling and noise, that don't cover the memory slots.

And no they are not. Hell most OEM systems and specially Workstations like the LGA2011 system, have the whole board rearranged, a lot have the the CPU at the bottom, some even add a L shape to the board to support it and the memory is placed in a way that even though the trace route is similar they can get air tunneled through a large HS, where even if the shroud covers the memory the HS doesn't. Sometimes it means having the memory above and below. Some times its to the left and right. But by moving the CPU away from it's normal ATX resting spot they aren't restricted in cooling options.