Thought experiment...Which configuration you think is better?

Which one is better?

  • Option A: 240mm AIO at bottom

    Votes: 0 0.0%
  • Option B: 240mm mounted at 45°angle

    Votes: 0 0.0%
  • Option C: 140mm AIO on top exhaust

    Votes: 1 12.5%
  • Option D: Air cooling NH-D15

    Votes: 6 75.0%
  • I know a better solution

    Votes: 1 12.5%

  • Total voters
    8

M76

[H]F Junkie
Joined
Jun 12, 2012
Messages
14,006
So here is the premise: I have a Silverstone FT05 case an RTX2080Ti that must go into the first slot nearest to the CPU, and a 240mm AIO. There are two possible configurations this all fits together: (excuse the crude illustrations)

Option A: Remove the 200mm fans from the bottom and put the radiator there horizontally, 3x120mm fans to the bottom. This way there is about 10mm clearance between the bottom plate and the fans. The length of the GPU does not allow the fans to fit on the inside under the radiator.

option_A.jpg

Option B: Leave the factory air penetrator fans alone, mount the radiator on top of the cpu rotated 45 degrees, with two fans in push pull configuration as shown bellow:

option_B.jpg

Which option do you think would be better? Or you think there is an Option C? Let's hear it.
 
Option C: 50mm thick 120mm AIO cooler like the Corsair H80i

Really though that case is not at all designed for mounting radiators. I would stick with air cooling or get a new case. Air cooling is cheaper and more reliable. I have AIO's because it looks pretty in the case window lol
 
Is there no mounting the rad at the front?

Of the options given, I see nothing wrong with option A, but have many questions about option B. Can you not attach the rad to the fans on the bottom of the case and use them to push the air through it? It's been demonstrated before that there isn't a significant penalty for exchanging heat into the air on the way into the case, even though it seems like this would mean you're cooling the graphics card with air that's already been heated up by the CPU. In practice, it just doesn't matter.
 
Option C: 50mm thick 120mm AIO cooler like the Corsair H80i

Really though that case is not at all designed for mounting radiators. I would stick with air cooling or get a new case. Air cooling is cheaper and more reliable. I have AIO's because it looks pretty in the case window lol
I meant other options given the 240mm AIO.

This is a thought experiment because I already tried both option A and B, and I already tried a 140mm AIO fitted at the top, and one of the best air coolers that money can buy (NH-D15) which is more expensive than both the tested AIOs were. Maybe I should add those options to the poll as well.
 
Is there no mounting the rad at the front?
This is a bottom to top air flow case, like a more compact version of the RV02. The front, the back(except for psu intake) and the sides are completely closed.
Of the options given, I see nothing wrong with option A, but have many questions about option B.
what questions?
Can you not attach the rad to the fans on the bottom of the case and use them to push the air through it?
That is literally option A?! Or I don't know what do you mean.
 
This is a bottom to top air flow case, like a more compact version of the RV02. The front, the back(except for psu intake) and the sides are completely closed.

what questions?

That is literally option A?! Or I don't know what do you mean.
Your picture makes it look like it’s just kind of near the fans on the bottom and offset to the left a little. Is there a bracket that comes with the case for doing option b, or were you going to just lay the rad in there? Or were you going to make a bracket?

I’d do option A, myself.
 
I meant other options given the 240mm AIO.

This is a thought experiment because I already tried both option A and B, and I already tried a 140mm AIO fitted at the top, and one of the best air coolers that money can buy (NH-D15) which is more expensive than both the tested AIOs were. Maybe I should add those options to the poll as well.

I would go with option A because it seems like the easiest way to mount it. Having the fans pull air in through a thin space on both sides doesn't seem ideal but there are a lot of cases that have front panels designed that way so I wouldn't think it would be that bad.

Are you going to post your results later?
 
Your picture makes it look like it’s just kind of near the fans on the bottom and offset to the left a little.
Is that a question about option A or option B? If option B, then it is not offset it is as far forward as it goes, the hoses come out there so that needs space.

As for option A, the case has mounting holes for 120,140,240,280,360mm rads on the bottom, obviously I'm using the 240mm holes. But the factory 200mm air penetrator fans have to be removed if the radiator is mounted there, and that is a big con for the overall case airflow.

Is there a bracket that comes with the case for doing option b, or were you going to just lay the rad in there? Or were you going to make a bracket?
There is no bracket, and there is no need for it either. There is exactly that much space there to fit the rad. It is not going anywhere, because it is wedged between the padded front of the case and the GPU's back heatsink. And the padded side of the case doesn't let it go that way either.
 
Your FT05 is one of the best airflow cases made. Only down side is it only has 162mm CPU clearance.

I would use an air cooler on CPU. Not sure NH-D15 will fit as it is 165mm tall and case CPU cooler spec is 162mm. But there are others as good costing less. Phanteks PH-TC14PE, Thermalright Macho Rev. B & Rev. C, Le Grand Macho RT, HR-22, Deep Cool Assassin & Assassin II to name a few.
 
Your FT05 is one of the best airflow cases made. Only down side is it only has 162mm CPU clearance.

I would use an air cooler on CPU. Not sure NH-D15 will fit as it is 165mm tall and case CPU cooler spec is 162mm. But there are others as good costing less. Phanteks PH-TC14PE, Thermalright Macho Rev. B & Rev. C, Le Grand Macho RT, HR-22, Deep Cool Assassin & Assassin II to name a few.
I do have an NH-D15 and have used it in this case, it fits.
This is a thought experiment because I want to see what people think. I already tried all solutions, and will reveal what I found out in 2020.
 
I do have an NH-D15 and have used it in this case, it fits.
This is a thought experiment because I want to see what people think. I already tried all solutions, and will reveal what I found out in 2020.
Look forward to seeing what you did and the results.
 
That case looks like a pita. Good luck whichever way you go.
It is actually my favorite case so far, and I'll probably keep it for the indefinite future as my main PC. And I have tried loads of cases.

There are five reasons I like it so much:
  1. Small footprint, but still fits a normal ATX MB
  2. IO ports on top so I don't have to crawl behind it to connect / disconnect things.
  3. Almost unrestricted airflow from bottom to top, which coincides with natural convection of heated air.
  4. Front and sides completely enclosed by thick cast alu and padded with noise absorbing foam, which does wonders for noise reduction.
  5. Completely minimalist design with zero flashiness and RGB nonsense on the outside.
It has some drawbacks as well, but I can live with those.

  1. There is very ltitle room to work with, so even disconnecting / connecting a CPU fan is a hassle.
  2. Limited GPU length and CPU cooler height.
  3. Only two USB3.0 ports on the front.
  4. Needs longer cables for perhipials
  5. Only a flimsy bracket for 3.5" drives and only for two of them. (I'm not even using it, I instead put my only remaining spinner behind the MB tray)
 
I read somewhere in the past that some GPUs (ones with extremely long coolers have problems wicking coolant from ends of long heatpipes back to cooler base because of the vertical lenght from end of pipe to 180* bend. Hope that makes sense. Have you ever heard or seen anything about this?
 
I voted for the nice air cooler, especially since you're not showing off -- but I'd want to fit two AIOs, one each for CPU and GPU. CPU AIO on the bottom because it's larger, and GPU AIO on the top. My current Define R5 will have a 280mm AIO in the front with the GPU AIO in the back.
 
I voted for the nice air cooler, especially since you're not showing off -- but I'd want to fit two AIOs, one each for CPU and GPU. CPU AIO on the bottom because it's larger, and GPU AIO on the top. My current Define R5 will have a 280mm AIO in the front with the GPU AIO in the back.
Why bigger rad on CPU mounted before smaller one on GPU? GPUs are often higher wattage (make more heat) than CPUs.
 
My guess is that:
A: - makes GPU run hotter as it looks like you will be feeding warm air from the rad to the GPU.
B: - maybe work ok but a pain to work on
C: - maybe work ok, but how is the warm air from the rest of the system going to affect the rad temps since warm air from the rest of the system is going out the top of the case?
D: - maybe just as good as any of the other options without having to worry about failure of the pimp or leaks.
 
I read somewhere in the past that some GPUs (ones with extremely long coolers have problems wicking coolant from ends of long heatpipes back to cooler base because of the vertical lenght from end of pipe to 180* bend. Hope that makes sense. Have you ever heard or seen anything about this?
No, haven't heard about that, but gamersnexus just did a video debunking the myth that gravity affects heatpipes. At least on CPU coolers. The coolant is not returned to the evaporator by gravity it returns by capillary action.
 
So, I'll reveal what I found. I won't be as scientific as a pro test as I don't have the equipment to take accurate temperature measurements, or to control for room temperature variances. But the conclusions are obvious. I'll start from worst to best.

  • The worst solution: Option A
    This was quite a shock to me, I expected it to work well enough, but in reality this was abysmal, far worse than all other options, and I mean double digits worse in temperature for both CPU and GPU.
    At first it seemed OK, but the HEAT builds up inside the case as the radiator and low clearance restricts the fans too much, turning the case into an oven. Eventually the CPU was idling at over 50C° and would hit excess of 80C° even at the easiest loads.
    The GPU was also idling over 50C° as it was enveloped in the hot air emanating from the rad.

  • The second worst: Option C
    This is an actually valid solution, it works as expected, nothing out of the ordinary. CPU idling at high thirties, under load it would hit mid 70s. I've had a different GPU when testing this method of CPU cooling so there is no point in comparing GPU temps.
    The only reason this looses out to the next solution is noise. As with the top mounted rad the cpu fans are closer to me, and the noise they make is more pronounced.

  • The second best solution: Option D
    The good old air cooler does its job well enough, temps wise it is indistinguishable from option C any variance can be chalked up for measurement error. But the noise is less due to needing less beefy fans and being located in the middle of the case as opposed to the exhaust on the top.

  • The clear winner: Option B
    I was a bit surprised at first how well this works, but when I think of it it makes sense. The space behind the GPU almost forms a completely isolated chamber for the CPU cooler separating it from all other heat sources and the bottom to top airflow is still provided by the big fans that come with the case giving the GPU plenty of fresh air. And the °45 orientation of the radiator has no ill effects on the cooling performance of it or the fans, in fact it might even be beneficial IMO as the temps suggests. In my 20 years of experience with building PCs I've never seen a CPU idle temp bellow 30C° Now I did.
    Surprisingly the cpu idles at high twenties to low 30s, and under full load only goes up to the high 60s low 70s. The GPU idles in the mid 20s, barely 5C° above room temperature. I've never seen anything like it. With the air cooler the GPU would idle around 30C°
    So when compared to the NH-D15 the CPU idle temp is about 10C° better, the load temp is about 5-7C° better.

Conclusion? Don't cover the only intake the case has with an AIO radiator. You can do whatever else and still be relatively OK.
 
No, haven't heard about that, but gamersnexus just did a video debunking the myth that gravity affects heatpipes. At least on CPU coolers. The coolant is not returned to the evaporator by gravity it returns by capillary action.
IsGN testing was with heatpipes horizontal and vertical with tips up, and while not enough difference to care aobut, there actually is a slight difference. TheLab (Greek) tested same way for years with results showing with motherboard flat (desktop case) vs vertical (tower case) there was a slight difference in temps. Again, not enough to worry.

Both of above tests were with heatpipes horizontal and with cooling ends up .. meaning wick's capillary action is horizonal or with gavity.

But in your case (no pun) GPU cooler orientaion is cooling ends of heatpipes down .. meaning wick's capillary action is up against gavity. Cooler heatpipes are not designed for this orientation and it's very possible some of the extremely long GPU cooler designs could be problematic in this orientation.

What I'm talking about is the same thing as Noctua said about their NH-L12 cooler as shown below:
"When using top-flow coolers in a tower-style case, we strongly recommend installing the cooler with the heatpipes in horizontal position or with the bends of the heatpipes pointing downwards. Please avoid installing the cooler with the bends pointing upwards as this may result in reduced cooling performance. In desktop cases, the cooler can be used in any orientation."
top_flow_orientation.jpg

NH-U12 is 66x128x150mm (HxWxD), that's 150mm time to turn in heatpipe which is less than some GPU cooler heatpipes. ;)
 
So, I'll reveal what I found. I won't be as scientific as a pro test as I don't have the equipment to take accurate temperature measurements, or to control for room temperature variances. But the conclusions are obvious. I'll start from worst to best.

  • The worst solution: Option A
    This was quite a shock to me, I expected it to work well enough, but in reality this was abysmal, far worse than all other options, and I mean double digits worse in temperature for both CPU and GPU.
    At first it seemed OK, but the HEAT builds up inside the case as the radiator and low clearance restricts the fans too much, turning the case into an oven. Eventually the CPU was idling at over 50C° and would hit excess of 80C° even at the easiest loads.
    The GPU was also idling over 50C° as it was enveloped in the hot air emanating from the rad.

  • The second worst: Option C
    This is an actually valid solution, it works as expected, nothing out of the ordinary. CPU idling at high thirties, under load it would hit mid 70s. I've had a different GPU when testing this method of CPU cooling so there is no point in comparing GPU temps.
    The only reason this looses out to the next solution is noise. As with the top mounted rad the cpu fans are closer to me, and the noise they make is more pronounced.

  • The second best solution: Option D
    The good old air cooler does its job well enough, temps wise it is indistinguishable from option C any variance can be chalked up for measurement error. But the noise is less due to needing less beefy fans and being located in the middle of the case as opposed to the exhaust on the top.

  • The clear winner: Option B
    I was a bit surprised at first how well this works, but when I think of it it makes sense. The space behind the GPU almost forms a completely isolated chamber for the CPU cooler separating it from all other heat sources and the bottom to top airflow is still provided by the big fans that come with the case giving the GPU plenty of fresh air. And the °45 orientation of the radiator has no ill effects on the cooling performance of it or the fans, in fact it might even be beneficial IMO as the temps suggests. In my 20 years of experience with building PCs I've never seen a CPU idle temp bellow 30C° Now I did.
    Surprisingly the cpu idles at high twenties to low 30s, and under full load only goes up to the high 60s low 70s. The GPU idles in the mid 20s, barely 5C° above room temperature. I've never seen anything like it. With the air cooler the GPU would idle around 30C°
    So when compared to the NH-D15 the CPU idle temp is about 10C° better, the load temp is about 5-7C° better.

Conclusion? Don't cover the only intake the case has with an AIO radiator. You can do whatever else and still be relatively OK.

I don't know about your conclusion. I have a 240mm AIO as my only intake. It only draws fresh air from a 1" opening in the top and bottom of the front panel as the rest is solid. Single 120mm exhaust in the rear and my temps are good. There is a good inch of space between the fans and the front panel though. Your small space between the panel and fan might just be killing airflow.

I have also read about people getting terrible AIO performance when their radiator is bottom mounted. The thought was leaving the block/pump as the highest point in the system would allow air bubbles to collect there.
 
I have also read about people getting terrible AIO performance when their radiator is bottom mounted. The thought was leaving the block/pump as the highest point in the system would allow air bubbles to collect there.
Yes, I think that is also a likely contributor. This current AIO seems to have much more air bubbles in it than other ones I had. Whenever the orientation is changed it gurgles for minutes during first use.
 
Last edited:
IsGN testing was with heatpipes horizontal and vertical with tips up, and while not enough difference to care aobut, there actually is a slight difference. TheLab (Greek) tested same way for years with results showing with motherboard flat (desktop case) vs vertical (tower case) there was a slight difference in temps. Again, not enough to worry.
The slight difference was actually the temps were better in the "wrong" orientation. VM temps were worse because heat couldn't just float up in an upside down MB.

Both of above tests were with heatpipes horizontal and with cooling ends up .. meaning wick's capillary action is horizonal or with gavity.
I don't know what you've seen but GN cleraly tested the coolers with upside down orientation where the liquid had to "climb" up to the cpu.

But in your case (no pun) GPU cooler orientaion is cooling ends of heatpipes down .. meaning wick's capillary action is up against gavity. Cooler heatpipes are not designed for this orientation and it's very possible some of the extremely long GPU cooler designs could be problematic in this orientation.

What I'm talking about is the same thing as Noctua said about their NH-L12 cooler as shown below:
"When using top-flow coolers in a tower-style case, we strongly recommend installing the cooler with the heatpipes in horizontal position or with the bends of the heatpipes pointing downwards. Please avoid installing the cooler with the bends pointing upwards as this may result in reduced cooling performance. In desktop cases, the cooler can be used in any orientation."
View attachment 212268
NH-U12 is 66x128x150mm (HxWxD), that's 150mm time to turn in heatpipe which is less than some GPU cooler heatpipes. ;)
That is the exact image GN also used to show, and then they demonstrated that fitting the cooler upside down has no meaningful impact on CPU temps. This seems more like a precaution on Noctua's part, not something you actually need to worry about.
As for my GPU, and all other GPUs I used in this exact orientation over the years, they all say "thank you very much we're fine".
 
The slight difference was actually the temps were better in the "wrong" orientation. VM temps were worse because heat couldn't just float up in an upside down MB.


I don't know what you've seen but GN cleraly tested the coolers with upside down orientation where the liquid had to "climb" up to the cpu.


That is the exact image GN also used to show, and then they demonstrated that fitting the cooler upside down has no meaningful impact on CPU temps. This seems more like a precaution on Noctua's part, not something you actually need to worry about.
As for my GPU, and all other GPUs I used in this exact orientation over the years, they all say "thank you very much we're fine".
GN images of cooler orientation do not have heatpipes in same orientation as Noctua says not to use them in.
GN image
f7be59fe3d7f657e0155734831cd8c56_XL.jpg


Noctua image with green check marks on heatpipe orientations that work and red no symbol on bad heatpipe orientation.
top_flow_orientation-jpg.jpg

For GN to test heatpipes in this bad orientation the cooler and motherboard would have to be turned over so cooler is upside down below motherboard. Can you show me an image from GN of CPU cooler with heatpipes in that orientation?



VM temps are different because airflow patterns change with cooler (and fan) orientation, so airflow is different depending on cooler orientation .. and this has nothing at all to do with heatpipe orientation.

Can you post picture or link to picture showing coolers upside down orientation where liquid has to climb up to CPU?

The fact your GPUs have worked well is proof that your GPU heatpipes had wicking that was able to move liquid up against gravity.

There are several kinds of wick used in heatpipes like sintered, screen and grooved, and their capillary action varies.

Bottom line is what Noctua said about cooler orientation for good heatpipe capillary action/performance .. and that Noctua has more credibility than GN.
 
GN images of cooler orientation do not have heatpipes in same orientation as Noctua says not to use them in.
GN image.
Sorry, you are looking at the wrong place. They might not have images in that orientation but in their video, they clearly tested with the crossed out orientation, and the results were no worse except for VM temps for the aforementioned reasons.
 
Sorry, you are looking at the wrong place. They might not have images in that orientation but in their video, they clearly tested with the crossed out orientation, and the results were no worse except for VM temps for the aforementioned reasons.
Please supply link and minutes into video they show it.

GamerNexus heatpipe orientation first 2 paragaphs:
"Back when Ryzen 3000 launched, there was reasonable speculation founded in basic physics that the asymmetrical die arrangement of the CPUs with fewer chiplets could have implications for cooler performance. The idea was that, at the root of it, a cooler whose heatpipes aligned to fully contact above the die would perform better, as opposed to one with two coolers sharing vertical contact with the die. We still see a lot of online commentary about this and some threads about which orientation of a cooler is “best,” so we thought we’d bust a few of the myths that popped-up, but also do some testing on the base idea.


This is pretty old news by now, with much of the original discussion starting about two months ago. Noctua revived the issue at the end of October by stating that it believed there to be no meaningful impact between the two possible orientations of heatpipes on AM4 motherboards, but not everyone has seen that, because we’re still getting weekly emails asking us to test this hypothesis."

The article is about heatpipe orientation to AM4 die, not about heatpipe orientatiion to gravity.
 
Last edited:
Sorry, I got tired of shouting at a wall so I never checked back.
Here is the video

View attachment 216201
Yeah, I saw it somewhere else. If memory serves temps were consistantly slightly higher, but they tried to explain it off as within margin of error .. which it is, but as all coolers gave slightly higher temps when upside down that is a tread indicating it's more than just margin of error.

That and as I said before, I believe Noctua says not to mount in this orientation because heatpipes wicking isn't as effective working against gavity.
 
Back
Top