News New GPU hybrid with radiator in PCIe by Asetek

blubblob

SFF Lingo Aficionado
Jul 26, 2016
104
127
So actually it does perform significantly better than a traditional blower GPU (as referenced by Alienware's own in-house testing).

The reason being that normally the single blower fan would have to cool VRM, Vram and GPU die. However, using the Rad Card the traiditional blower can run at lower RPM and still cool Vram and VRM plenty. Meanwhile the AIO will grab all the heat from the GPU die, and transfer it into the Rad Card radiator, which is then cooled by another blower, which also doesn't need to run at full force. :)
That a cooling solution that has about 7cm "width" to work with outperforms one that has about 3cm and is generally considered to be inferior efficiency wise ("blower") isn't really much of a surprise.
The thing takes up about 2.5 liters by itself (by my estimation), so it better be at least as powerful/silent as a solution with a full cover block, pump and a slim 240 rad with 120x25 fans that's pulling air from a bottom vent - of a properly design case. ;) As that would still take up less space.

At the moment the "PCIe-Cooler" seems like a nice engineering solution for horrible case design - that would probably be outperformed by any decent existing GPU-cooling solution (especially the "AIO" ones with 240mm rads) in a case where airflow wasn't an absolute afterthought.
This could be "easily" checked by dropping a 2080ti variant of this into say a Chimera Cerberus and comparing it to an EVGA 2080ti kingpin. I'm sure you have one of those lying around - the AIO on those is an Asetek one too, right? :D
 
Last edited:

murderbymodem

Average Stuffer
Jun 6, 2019
60
55
I like this. The vast majority of ATX or Micro-ATX rigs built are single GPU systems that use no other expansion cards. It's just a waste of space. A big GPU radiator that can occupy expansion slots is a decent idea for people who are not utilizing that space for anything else.

In the future, I forsee M.2 SSDs becoming much cheaper and people will start utilizing those PCI-Express to 4x or 8x M.2 cards. but until the cost of that drops dramatically I don't see anything else that the average user would need to use extra expansion slots for, with dual GPU setups being effectively dead for gaming use-cases.
 
  • Like
Reactions: AseDen

Valantar

Shrink Ray Wielder
Jan 20, 2018
2,201
2,225
Yeah, I still find this quite intriguing. Designing a case around two of these (in a simpler shroud without the PCIe slot format etc.) on each end of an ITX motherboard + GPU sandwich could make for a very compact yet very effectively cooled solution. Definitely intriguing, if unrealistic.
 
  • Like
Reactions: AseDen

AseDen

Cable-Tie Ninja
Original poster
Asetek
Dec 7, 2018
194
295
www.asetek.com
That a cooling solution that has about 7cm "width" to work with outperforms one that has about 3cm and is generally considered to be inferior efficiency wise ("blower") isn't really much of a surprise.
The thing takes up about 2.5 liters by itself (by my estimation), so it better be at least as powerful/silent as a solution with a full cover block, pump and a slim 240 rad with 120x25 fans that's pulling air from a bottom vent - of a properly design case. ;) As that would still take up less space.

At the moment the "PCIe-Cooler" seems like a nice engineering solution for horrible case design - that would probably be outperformed by any decent existing GPU-cooling solution (especially the "AIO" ones with 240mm rads) in a case where airflow wasn't an absolute afterthought.
This could be "easily" checked by dropping a 2080ti variant of this into say a Chimera Cerberus and comparing it to an EVGA 2080ti kingpin. I'm sure you have one of those lying around - the AIO on those is an Asetek one too, right? :D

The 2080Ti kingpin is indeed using Asetek. The two technologies aren't really comparable, as they fill two different needs.
The 240mm will outperform the Rad Card, absolutely no doubt. It is also considerably larger than the Rad Card (both the radiator of it, as well as using two standard fans, rather than a single blower-style).

I think something that was lost (by our own fault) in the launch, is that this product is intended to solve a very specific problem (at the very least initially) - getting above average GPU performance in airflow restrained cases. The design is versatile, though, and can be adapted to airflow optimized cases (should there be a demand for such a solution as well).

Awesome video, it makes the benefits of this design more clear.

That was our intention. The initial messaging we did, didn't seem to properly highlight the purpose and benefits of the design clearly :)
 

blubblob

SFF Lingo Aficionado
Jul 26, 2016
104
127
The two technologies aren't really comparable, as they fill two different needs.
The 240mm will outperform the Rad Card, absolutely no doubt. It is also considerably larger than the Rad Card (both the radiator of it, as well as using two standard fans, rather than a single blower-style).
They are as much comparable as a blower to an open-air gpu - because they are blower and open-air AIOs - both keeping the gpu cool. ;)
In what way is the 240mm considerably larger than the rad card? From what I can tell the rad card takes up 2 slots for the actual GPU (4cm) + a bit more than 2 slots for the radiator (4.5cm? - I'm working of one render here, so I might be off by a bit). The 240mm is using a standard slim radiator (3cm) and standard fans (2.5cm) and the GPU portion seems to be about the same (4cm). So we are looking at a 8.5cm vs 9.5cm thickness. The radcard is longer than an ATX-Mainboard (so about 28cm?) and has a height of at least 11cm, more like 13cm if we calculate in space required for the tubes, which seems to match the rough dimensions of the 240mm hybrid gpu.
While the 240mm is indeed bigger, I would argue the size difference is marginal in comparison to the cooling alternatives. (We where comparing the performance to a blower GPU-cooler with less than half the volume just a few posts ago.)

[...] this product is intended to solve a very specific problem (at the very least initially) - getting above average GPU performance in airflow restrained cases. The design is versatile, though, and can be adapted to airflow optimized cases (should there be a demand for such a solution as well).
I agree on the first part. Though would argue the cooling solution is a kludge, and the case design of the alienware aurora should have and could have been easily improved for better airflow in the GPU area. There is for example about 5cm of space between case sidewall and "standard" PCIe height - enough for a corsair one like setup of a slim 240mm rad + slim fans, it would just require extending the vent pattern on the side a bit. Alternatively the bottom of the case is pretty much shaped for airflow ingress with the thick plastic standoffs and "hollow" bottom - just move the HDD trays to the swing arm and you have ample space for a full sized 240 rad with fresh air bottom intake. Of course that's on Dell/Alienware and Asetek had to do the best designing around those "constraints".

What I don't see is how this design can be applied in a situation other than a combination of ample space and horrible airflow while still maintaining an advantage over alternatives (Open Air Coolers with a hypothetical 4+ slot width - think Raijintek Morpheus x2, "standard" Hybrid GPUs, or just good open air coolers with a lot of extra space for air to move around).
I believe it has been established that blower are generally inferior to open-air designs when it comes to GPU cooling in good airflow conditions. It would surprise if that wouldn't hold true for Hybrid-AIOs. There are no inherent space savings associated with a blower radiator as opposed to an open-air one. One could easily manufacture a "frame" you could screw a standard radiator to, that slots into a PCIe slot, but having the air-ingress directly flush with the outside wall for large volume fresh air intake (or air-egress for directly dumping the heat outside) is obviously superior.
 
Last edited:

|||

King of Cable Management
Sep 26, 2015
775
759
They are as much comparable as a blower to an open-air gpu - because they are blower and open-air AIOs - both keeping the gpu cool. ;)
In what way is the 240mm considerably larger than the rad card? From what I can tell the rad card takes up 2 slots for the actual GPU (4cm) + a bit more than 2 slots for the radiator (4.5cm? - I'm working of one render here, so I might be off by a bit). The 240mm is using a standard slim radiator (3cm) and standard fans (2.5cm) and the GPU portion seems to be about the same (4cm). So we are looking at a 8.5cm vs 9.5cm thickness. The radcard is longer than an ATX-Mainboard (so about 28cm?) and has a height of at least 11cm, more like 13cm if we calculate in space required for the tubes, which seems to match the rough dimensions of the 240mm hybrid gpu.
While the 240mm is indeed bigger, I would argue the size difference is marginal in comparison to the cooling alternatives. (We where comparing the performance to a blower GPU-cooler with less than half the volume just a few posts ago.)


I agree on the first part. Though would argue the cooling solution is a kludge, and the case design of the alienware aurora should have and could have been easily improved for better airflow in the GPU area. There is for example about 5cm of space between case sidewall and "standard" PCIe height - enough for a corsair one like setup of a slim 240mm rad + slim fans, it would just require extending the vent pattern on the side a bit. Alternatively the bottom of the case is pretty much shaped for airflow ingress with the thick plastic standoffs and "hollow" bottom - just move the HDD trays to the swing arm and you have ample space for a full sized 240 rad with fresh air bottom intake. Of course that's on Dell/Alienware and Asetek had to do the best designing around those "constraints".

What I don't see is how this design can be applied in a situation other than a combination of ample space and horrible airflow while still maintaining an advantage over alternatives (Open Air Coolers with a hypothetical 4+ slot width - think Raijintek Morpheus x2, "standard" Hybrid GPUs, or just good open air coolers with a lot of extra space for air to move around).
I believe it has been established that blower are generally inferior to open-air designs when it comes to GPU cooling in good airflow conditions. It would surprise if that wouldn't hold true for Hybrid-AIOs. There are no inherent space savings associated with a blower radiator as opposed to an open-air one. One could easily manufacture a "frame" you could screw a standard radiator to, that slots into a PCIe slot, but having the air-ingress directly flush with the outside wall for large volume fresh air intake (or air-egress for directly dumping the heat outside) is obviously superior.

You can see in the video that the physical version of the rad card is within the 2 slot width of the double-wide PCI bracket. So it is 39.5-40 mm thick.

The major thing is airflow. With this design, they ingest the air from the face where the fan is and they exhaust out the back through the PCI slot. There have been some media center styled PCs with minimal height above GPUs that have shown great benefit to using a blower style card in those sorts of situations.

The EVGA Kingpin radiator uses a 30mm thick radiator and 25mm thick fans, so right there you have an additional ~15mm of thickness, but then you need additional spacing either in from or or behind the card for ingest/exhaust air, not to mention air recirculation problems with hot air within the case.

They did a good job maximizing the thermal gradient of the working fluids in the cooling system.

I would agree with others sentiments that this could allow for some interesting SFF watercooling setups with the unique airflow path if this were able to be integrated into a custom loop (although an outstanding question of whether this radiator is copper or aluminum for galvanic corrosion compatibility remains).
 

AseDen

Cable-Tie Ninja
Original poster
Asetek
Dec 7, 2018
194
295
www.asetek.com
They are as much comparable as a blower to an open-air gpu - because they are blower and open-air AIOs - both keeping the gpu cool. ;)
In what way is the 240mm considerably larger than the rad card? From what I can tell the rad card takes up 2 slots for the actual GPU (4cm) + a bit more than 2 slots for the radiator (4.5cm? - I'm working of one render here, so I might be off by a bit). The 240mm is using a standard slim radiator (3cm) and standard fans (2.5cm) and the GPU portion seems to be about the same (4cm). So we are looking at a 8.5cm vs 9.5cm thickness. The radcard is longer than an ATX-Mainboard (so about 28cm?) and has a height of at least 11cm, more like 13cm if we calculate in space required for the tubes, which seems to match the rough dimensions of the 240mm hybrid gpu.
While the 240mm is indeed bigger, I would argue the size difference is marginal in comparison to the cooling alternatives. (We where comparing the performance to a blower GPU-cooler with less than half the volume just a few posts ago.)


I agree on the first part. Though would argue the cooling solution is a kludge, and the case design of the alienware aurora should have and could have been easily improved for better airflow in the GPU area. There is for example about 5cm of space between case sidewall and "standard" PCIe height - enough for a corsair one like setup of a slim 240mm rad + slim fans, it would just require extending the vent pattern on the side a bit. Alternatively the bottom of the case is pretty much shaped for airflow ingress with the thick plastic standoffs and "hollow" bottom - just move the HDD trays to the swing arm and you have ample space for a full sized 240 rad with fresh air bottom intake. Of course that's on Dell/Alienware and Asetek had to do the best designing around those "constraints".

What I don't see is how this design can be applied in a situation other than a combination of ample space and horrible airflow while still maintaining an advantage over alternatives (Open Air Coolers with a hypothetical 4+ slot width - think Raijintek Morpheus x2, "standard" Hybrid GPUs, or just good open air coolers with a lot of extra space for air to move around).
I believe it has been established that blower are generally inferior to open-air designs when it comes to GPU cooling in good airflow conditions. It would surprise if that wouldn't hold true for Hybrid-AIOs. There are no inherent space savings associated with a blower radiator as opposed to an open-air one. One could easily manufacture a "frame" you could screw a standard radiator to, that slots into a PCIe slot, but having the air-ingress directly flush with the outside wall for large volume fresh air intake (or air-egress for directly dumping the heat outside) is obviously superior.

For the size of the radiator etc. I'll refer to "|||" who made some great points about the considerations that come into play.
I wouldn't go and claim that blowers are superior to regular fans, as they're simply not. Except for cases like this, where you have a unique situation, that allows for blowers to become highly efficient while (with the right tuning) also can be quiet.

You can see in the video that the physical version of the rad card is within the 2 slot width of the double-wide PCI bracket. So it is 39.5-40 mm thick.

The major thing is airflow. With this design, they ingest the air from the face where the fan is and they exhaust out the back through the PCI slot. There have been some media center styled PCs with minimal height above GPUs that have shown great benefit to using a blower style card in those sorts of situations.

The EVGA Kingpin radiator uses a 30mm thick radiator and 25mm thick fans, so right there you have an additional ~15mm of thickness, but then you need additional spacing either in from or or behind the card for ingest/exhaust air, not to mention air recirculation problems with hot air within the case.

They did a good job maximizing the thermal gradient of the working fluids in the cooling system.

I would agree with others sentiments that this could allow for some interesting SFF watercooling setups with the unique airflow path if this were able to be integrated into a custom loop (although an outstanding question of whether this radiator is copper or aluminum for galvanic corrosion compatibility remains).

This is produced in aluminum, as we're using fluids that deal with the issues of galvanic corrosion :)
However, a copper version is on the list of things we're looking at ;)
 

blubblob

SFF Lingo Aficionado
Jul 26, 2016
104
127
You can see in the video that the physical version of the rad card is within the 2 slot width of the double-wide PCI bracket. So it is 39.5-40 mm thick.

The major thing is airflow. With this design, they ingest the air from the face where the fan is and they exhaust out the back through the PCI slot. There have been some media center styled PCs with minimal height above GPUs that have shown great benefit to using a blower style card in those sorts of situations.

The EVGA Kingpin radiator uses a 30mm thick radiator and 25mm thick fans, so right there you have an additional ~15mm of thickness, but then you need additional spacing either in from or or behind the card for ingest/exhaust air, not to mention air recirculation problems with hot air within the case.

[...]
So I was off by 0.5cm - the Kingpin is 19% bigger than the Rad-Card instead of 12%. They are very much comparable size-wise.
The Rad-Card is 100% bigger than a standard 2-slot blower or open air GPU. Those are very much NOT comparable size-wise.

A setup with the Kingpin radiator drawing air from inside the case and exhausting it to the outside via a bottom- or side-vent has exactly the same constrains regarding recirculation and fresh air intake as the Rad-Card. AIOs in general should never have issues with air recirculation, so that's not something you can differentiate both solutions on. The minimum spacing requirements seem to be the same, though the blower style would suffer less from tighter spacing due to higher static pressure. But even a blower can't magically generate fresh air inside the case and needs to draw it from somewhere, requiring a gap on the intake-side.
Such a setup could have been possible size- and geometry-wise in the very case the Rad-Card solution was designed for - but wasn't due to silly case design decisions. Which is (again, just to reiterate) not an issue with the cooling solution but the horrible case design.
One could argue (this would have to be tested) that drawing fresh air from the outside through the main radiator and exhausting it via the blower portion is an even more efficient orientation (especially considering the cooling requirements and thermal tolerances of the vrm) of the Kingpin solution, creating a full air cycle. An option that's not possible with the Rad-Card.

As far as I can tell a standard blower design excels in a very narrow subset of situations where there is no space whatsoever for airflow, radiators or extra fans to move heated air, that require the card to run an inefficient double duty of both cooling the GPU and removing the heated air from the case over a long, narrow airpath.
With the size of the Rad-Card this constraint obviously can't be the case anymore. There has to be at least enough free space in the case for another 2 slot PCIe card (realistically more due to fixed PCIe spacing and intake air gap requirements). Space that could be used for radiators, intake/exhaust fans or airflow.
You can't argue that the Rad-Card is awesome because a blower works in a specific situation and then disregard that the blower to a big part works well in those specific situations because there is not enough space to fit anything else, including a bigger Rad-Card.

I would like to see those "media center styled PCs" that have enough space to fit a Rad-Card like setup, but no kingpin or even 3-slot open-air + exhaust slots/fans setup or at least couldn't be made to fit those with minimal design improvements.
I'm genuinely curious what you believe those unique situations are, where this can be the superior cooling solution. The Aurora (at least to me) clearly isn't it. It's just a badly designed case that is saved by a well-engineered kludge.
 

Valantar

Shrink Ray Wielder
Jan 20, 2018
2,201
2,225
So I was off by 0.5cm - the Kingpin is 19% bigger than the Rad-Card instead of 12%. They are very much comparable size-wise.
The Rad-Card is 100% bigger than a standard 2-slot blower or open air GPU. Those are very much NOT comparable size-wise.

A setup with the Kingpin radiator drawing air from inside the case and exhausting it to the outside via a bottom- or side-vent has exactly the same constrains regarding recirculation and fresh air intake as the Rad-Card. AIOs in general should never have issues with air recirculation, so that's not something you can differentiate both solutions on. The minimum spacing requirements seem to be the same, though the blower style would suffer less from tighter spacing due to higher static pressure. But even a blower can't magically generate fresh air inside the case and needs to draw it from somewhere, requiring a gap on the intake-side.
Such a setup could have been possible size- and geometry-wise in the very case the Rad-Card solution was designed for - but wasn't due to silly case design decisions. Which is (again, just to reiterate) not an issue with the cooling solution but the horrible case design.
One could argue (this would have to be tested) that drawing fresh air from the outside through the main radiator and exhausting it via the blower portion is an even more efficient orientation (especially considering the cooling requirements and thermal tolerances of the vrm) of the Kingpin solution, creating a full air cycle. An option that's not possible with the Rad-Card.

As far as I can tell a standard blower design excels in a very narrow subset of situations where there is no space whatsoever for airflow, radiators or extra fans to move heated air, that require the card to run an inefficient double duty of both cooling the GPU and removing the heated air from the case over a long, narrow airpath.
With the size of the Rad-Card this constraint obviously can't be the case anymore. There has to be at least enough free space in the case for another 2 slot PCIe card (realistically more due to fixed PCIe spacing and intake air gap requirements). Space that could be used for radiators, intake/exhaust fans or airflow.
You can't argue that the Rad-Card is awesome because a blower works in a specific situation and then disregard that the blower to a big part works well in those specific situations because there is not enough space to fit anything else, including a bigger Rad-Card.

I would like to see those "media center styled PCs" that have enough space to fit a Rad-Card like setup, but no kingpin or even 3-slot open-air + exhaust slots/fans setup or at least couldn't be made to fit those with minimal design improvements.
I'm genuinely curious what you believe those unique situations are, where this can be the superior cooling solution. The Aurora (at least to me) clearly isn't it. It's just a badly designed case that is saved by a well-engineered kludge.
You do have some points, but you are blatantly ignoring the airflow path and mounting geometry possibilities opened by a radiator where the air passes through lengthwise rather than the typical orientation. You say it requires "space that could be used for radiators, intake/exhaust fans or airflow", but seem to ignore the fact that this space is likely squeezed in between the GPU or GPU riser and the table the PC is standing on, meaning it's a highly restrictive area with large flat planes blocking airflow on both sides, likely with very small gaps around the edges. As such, you have no guarantee that a conventional radiator in that type of space will perform any better, and depending on the fans, it might very well do worse. This also forces you to have a very messy airflow pattern in your case, with any type of front-to-back flow being impossible. On the other hand, using something like this "rad card" (though without the portions making it PCIe slot mountable) as a radiator you could make a compact sandwich of the motherboard and GPU, with a full cover block on the (now 1-slot) GPU and something like an Eisbaer Solo LT on the CPU, bookend the each side (not front or back) of this sandwich with radiators, put a couple of small intake fans in the front to provide overall airflow including across the boards for VRM cooling, and you'd have an extremely compact fully watercooled PC. Sure, you'd need to figure out a PSU solution, but a FlexATX unit or something like a MeanWell AC-DC unit would fit perfectly next to either of the radiators without blocking any airflow whatsoever, or even as an extra layer in the motherboard+GPU sandwich if that fits better with the geometry. My point is: a radiator like this gives us new options for layouts that have previously been either extremely impractical or plain impossible. Does that mean it is the best radiator in a vacuum? Of course not! (Cue joke about radiators not working in a vacuum, I guess?) But if "what is the best cooling solution" was the question we asked here at SFF.network, we would all be building 2-3-4 radiator full tower builds. This can let us get more cooling capacity into a smaller volume than what has previously been possible. And that is great. This is of course discussing a potential DIY solution rather than the pre-built rad card itself, but I think we've all seen enough projects here ripping apart AIOs and using parts of them to know this is by no means exotic here.