Vapor Chamber Heatsinks on LGA115x

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
I have recently switched from an x99 + i7 5820K system to a Z370 + i7 8700K. I found the lack of good cooling options frustrating, as I understand many others do. With socket 2011, the standard cooler recommended here and elsewhere seems to be the Dynatron T318 heatsink, plus a fan of your choice. The T318 is a full-copper vapor chamber cooler, specified for up to 165W TDP. That's with server rack airflow, but even with tolerable noise levels this heatsink has pretty decent cooling abilities. For those not familiar with them, vapor chambers work on the same principle as heatpipes, but instead of transporting heat from one end to the other, they spread out heat across their opposite surface area. My layperson's understanding is that for space constrained applications, vapor chambers are sometimes better than heatpipes, as you don't have far to go anyway, and vapor chambers are great at spreading out your waste heat as evenly and efficiently as possible across whatever small heatsink you can fit in. Aside from select socket 2011 server heatsinks, vapor chambers are also used in some GPUs, especially reference blower models. That's 200, 300W of excess heat cooled by heatsinks no bigger than an L9i or C7.

So you would think a similar solution would be amazing for the relatively modest 95ish W TDP you get with most top-end socket 115x CPUs. Alas, I've looked everywhere from eBay to Digikey to Taobao, and I couldn't find a single vapor chamber cooler for socket 115x. I'm not quite sure why - might be that there's space constraints that make it infeasible in general, or maybe vapor chambers are too expensive to be worth it for relatively small cooling requirements. (Or maybe I'm misunderstanding the physics and engineering of theis, and vapor chambers are just not appropriate for this?)

Coming up empty, I decided to take matters into my own hand. After all, I still had the T318 lying around anyway, so why not try and see if I coud make it fit?

First, the T318, as well as presumably its square sibling, the R15, work, sort of. They both have a flat base, meaning that unless your mainboard has an exceptionally large area around the socket clear of any components that protrude farther than the top of the CPU, they won't fit directly. Still, on my board, nothing stuck out more than a milimeter or so above the CPU, so I could get the T318 to make contact via a copper shim. I left this approach at that stage as I didn't like the copper shim too much.

What I ended up with ultimately though is a heatsink from a GPU. That one has a square-shaped elevation of roughly 46x46mm on the base to make contact with the GPU die, which also fits pretty well for a cpu. That elevation is about 2.7mm from the rest of the base of the vapor chamber, so on my board the entire heatsink was clear of any other components without any shims. The heatsink has mounting holes with 58.4mm spacing, so I 3D-printed an adapter to fix it to the LGA 115x 75mm-spaced mounting holes, then attached a Noctua A9x14 on top.

I'm attaching pictures if people are interested. (Don't mind my DIY M.2 heatsink...)

I'm curious to hear what people think of the idea. I don't have another cooler to do temperature comparisons with, but just as some sort of reference, my stock 8700K sits at around 41 degrees at idle, with the fan at an inaudible 860rpm.
 

Attachments

  • DSC04606.jpg
    DSC04606.jpg
    193 KB · Views: 0
  • DSC04607.jpg
    DSC04607.jpg
    248.3 KB · Views: 0
  • DSC04613.jpg
    DSC04613.jpg
    342.8 KB · Views: 0
  • DSC04615.jpg
    DSC04615.jpg
    289.6 KB · Views: 0
  • DSC04620.jpg
    DSC04620.jpg
    213.4 KB · Views: 0
Last edited:

RosaJ

Limp Gawd
Joined
Mar 22, 2016
Messages
281
Um, as a T318 owner and user, screws are diagonal, I would be very interested in having one of these adapters too!
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
What are the temps under full load in Prime95?

Good point, should have tested that first. Turns out it goes up to close to 100 degrees and then thermal throttles. Given that the heatsink is from a 250W GPU, I'd expect a litlle better than that. Could be partly because I didn't take too much care with applying thermal compound when mounting the heatsink, as I'm not done with the build yet and will have to remove and remount it a few more times anyway. I'll have to double check mounting pressure too, I think the frame of the 3d-printed mount adapter might be a fraction of a milimeter too thick. I've only just done all of this in the last few days, so it's very much still a work in progress.

Ok, I need one for AM4 asap!

I could have a look at making an adapter for AM4. Which board are you using? Are you sure a sufficiently large area is clear around the socket?

Um, as a T318 owner and user, screws are diagonal, I would be very interested in having one of these adapters too!

I might make an adapter for the T318 too, if only to compare against the GPU heatsink. The latter has the advantage that it makes direkt contact with the CPU, but the fins are aluminium. With the T318 you'd need a copper shim, but then the fins are full copper too. I'd be curious which one does better. What do you mean with diagonal screws? That the narrow-ILM mount is rectangular, not square?
 

rfarmer

[H]ard|Gawd
Joined
May 9, 2014
Messages
1,168
Couple questions. I am guessing your 8700k is not delidded? What overclock, if any, are you running. What voltage is your CPU at?

I have a delidded 8700k being cooled by a Big Shuriken 2 which is not a huge cooler. 4.4 GHz, 1.20v. With an ambient temp of 20C mine idles at 31C and will run Prime95 at 75C max.

Like you said I would check contact and pressure with your cooler and the biggest thing I can recommend for the 8700k is delidding. Makes a huge difference especially with small coolers.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
Couple questions. I am guessing your 8700k is not delidded? What overclock, if any, are you running. What voltage is your CPU at?

I have a delidded 8700k being cooled by a Big Shuriken 2 which is not a huge cooler. 4.4 GHz, 1.20v. With an ambient temp of 20C mine idles at 31C and will run Prime95 at 75C max.

Like you said I would check contact and pressure with your cooler and the biggest thing I can recommend for the 8700k is delidding. Makes a huge difference especially with small coolers.

No, not delidded yet. That is the main reason I did not bother with getting the thermal paste perfect yet, as I will delid the CPU in a week or so, once I get a copper IHS. I will probably put liquid metal between IHS and heatsink as well. Will report back of course.

The processor is all at stock settings right now, no overclock, stock voltage, except the RAM from 2666 to 3000.

And yeah, mounting pressure is almost certainly off, as are a few other dimensions on the adapter. Still quite happy with it for a first prototype. ;) I'll have another one printed soon.
 

Ej24

Gawd
Joined
Jun 11, 2016
Messages
628
No, not delidded yet. That is the main reason I did not bother with getting the thermal paste perfect yet, as I will delid the CPU in a week or so, once I get a copper IHS. I will probably put liquid metal between IHS and heatsink as well. Will report back of course.

The processor is all at stock settings right now, no overclock, stock voltage, except the RAM from 2666 to 3000.

And yeah, mounting pressure is almost certainly off, as are a few other dimensions on the adapter. Still quite happy with it for a first prototype. ;) I'll have another one printed soon.

Keep in mind that gpu's are direct contact. They don't have an IHS. So the IHS will definitely have an impact on performance. Also, I assume that heatsink is from a 1080 or 1080ti? It looks like the Nvidia reference vapor chamber. Anyway, those gpu's have a much larger die size than an i7 cpu iirc, so the gpu's have more surface area to transfer heat directly. I say definitely delid and let us know how it goes. Liquid metal under the IHS is a must, but between the IHS and heatsink? Meh. Just save yourself some headaches and use a good non conductive paste there. You'll get a huge drop in temps from liquid metal under the IHS.
 

Boil

[H]ard|Gawd
Joined
Sep 19, 2015
Messages
1,439
Thermal Grizzly Conductonaut (Liquid Metal) under the IHS...

(...Conductonaut is what Silicon Lottery is using on their de-lids now...)

Thermal Grizzly Kryonaut (Thermal Paste) over the IHS...
 

Ej24

Gawd
Joined
Jun 11, 2016
Messages
628
Thermal Grizzly Conductonaut (Liquid Metal) under the IHS...

(...Conductonaut is what Silicon Lottery is using on their de-lids now...)

Thermal Grizzly Kryonaut (Thermal Paste) over the IHS...

Agreed. Winning combination.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
Yep, it's from a reference 1080Ti, found one (just the heatsink) on eBay for next to nothing. I think most reference vapor chamber heatsinks are more or less identical. At least, the vapor chamber seems to be a standard part, from Wakefield-Vette. Same for the Dynatron ones! The only one I've seen so far that uses a custom vapor chamber is the AMD Vega one.

Good point about die size. The Socket 2011 CPUs are a larger die as well I believe (plus soldered!), so you would expect somewhat less performance on 115x.

Still, I think with better contact pressure this should still work reasonably well. I've noticed as well that the adapter flexes a little bit, so I suspect with downward pressure from the screws in the corner the supports around the sides might flex upwards and press against the heatsink. But there isn't much space to make any of it any bigger for more stability; and printing / CNCing it in metal would be too expensive. I'll have to figure something out there. I might try just four small mount adapters, one for the screws in each corner.

Just out of curiosity, what's wrong with liquid metal between IHS and heatsink?
 

Ej24

Gawd
Joined
Jun 11, 2016
Messages
628
Just out of curiosity, what's wrong with liquid metal between IHS and heatsink?

Mostly diminishing returns, but also it's just a potential point of failure.

First, safety. The liquid metal under the IHS is relatively safe because it's contained under the IHS. On top of the IHS though, if you use even a tiny bit too much and it squeezes out, you're going to have a bad day if it drips down into the socket and pins, or on the motherboard. All it takes is the droplet the size of a pinhead to ruin things.

Second, reactivity. The IHS is nickel plated, the gallium alloy of liquid metal doesn't really react with nickel or silicon (iirc there's actually a silicon nitride coating on the surface of the die making it even less reactive to gallium), so under the IHS it rarely "dries out" and almost never needs to be reapplied. However it does tend to "soak in" to copper over time giving the impression that it dried out. That copper base gpu heatsink may soak up the liquid metal resulting in having to reapply ever year or so. It only penetrates a few atoms deep so it's not like it going to mess up the vapor chamber function. You might just have to reapply which means more risk for some spillage or leaking as mentioned above.

Third, viscosity. Normal non-conductive pastes are obviously, thick viscous pastes. They're more forgiving of uneven or low mounting pressure like you may experience. Liquid metal is about as viscous as water and has massive surface tension. It needs good, even pressure and a tiny gap between the surfaces to make good connection for thermal transfer.

You'll get huge drops in temps from liquid metal under the IHS, around 15C usually, but maybe only another 4-5C from liquid metal also on top of the IHS. Intel's Tim is the bottleneck (like 4W/mK), the normal tims we use between IHS and heatsink are way better and aren't even close to being used to their thermal conductivity limits (15W/mK) when used on a non-delidded cpu. Change out the Intel Tim and you remove the bottleneck. Maybe it'll be different with your setup, it's a freaking gpu heatsink after all. Can't hurt to try. Just thought you'd want to keep these things in mind.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
Quick update: I have done some more experimenting, and it looks like indeed there was an issue with mounting pressure; also with the thermal paste - I used some old no-name paste from eBay that's been open a couple of years, thinking it wouldn't make _that_ much of a difference for an initial test. Anyway, I have now tried with some Kryonaut and with manually applying pressure to the heatsink, and temperatures are a lot better. In Prime95, it stays around the mid-70s for large FFTs, and mid 90s for small FFTs; during small FFT tests I do see occasional spikes to 100 accompanied by thermal throttling. Mounting pressure will be tricky to figure out more permanently though. I've had some new mount adapters with extra supports printed, but they still feel relatively flimsy. And having anything made out of metal would likely be cost prohibitive. I might try adding a copper shim to the GPU heatsink as well, as I suspect the mount adapter might press against either the ILM or one of the mainboard components somewhere.
I've also tested the T318 with a copper shim again, and results are similar. Mounting it would be even more tricky though, as the mounting screws are so far apart I suspect any plastic adapter will bend too much. The R15 would probably work though.

Couple of questions:
1. Does anyone have an idea of what sort of temperatures one should expect with an 8700K - stock settings, not delidded, with similarly small SFF heatsinks? Personally I'm fairly happy with the vapor chamber heatsink, but I'm struggling to place the performance in relation to other SFF coolers. I've seen some comments regarding the earlier quadcore i7s in SFF systems, but haven't found anything on the 8700K.
2. Especially for dondan, since you asked: What Prime95 settings do you use for temperature tests? With the default Blend torture test temperatures vary hugely between different tests for me, as I mentioned above.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
Nevermind. Curiousity got the better of me, and I ordered myself a Noctua NH-L9i for comparison. Results are mixed. For most of my tests, I compared different heatsinks (GPU heatsink with or without copper shims, T318 with one or two copper shims, and the L9i heatsink), all just lying on top of the CPU. That's not real-world conditions - on the one hand this was open-air, no case; on the other hand there wasn't as much mounting pressure as there would be had the heatsinks been screwed in. I've tried heatsink just lying there, or manually pressing down on them. I'm guessing that that would affect all of them roughly in the same way, but obviously a big source of uncertainty. Now, for all of my DIY heatsinks, results varied *hugely* between tests. I looks like the exact position makes a big difference. My hypothesis is that especially without a copper shim (or on the T318 with just one copper shim), there is some spots where the bottom of the vapor chamber might *just* make contact with some of the mainboard components - or not. Things got a little more consistent with an extra copper shim for more clearance, but there was still a lot more variance than with the L9i. I've also tried soldering a copper shim onto the base of the GPU vapor chamber to see if that made a difference compared to just sticking it on with thermal paste. (It didn't.) I did try one test with the best (sturdiest and best fitting) 3D-printed adapter I have managed so far, only to find that (a) performance was worse than without any mounting at all, and (b) after the test I found that the adapter had a crack, and fairly easily snapped in two when I tried to pull on one of the corners. Presumably the force from tightening the screws went into that, instead of into pressing the vapor chamber onto the CPU.

To get to the point: On the best runs, the DIY vapor chamber heatsinks did about as well as the L9i, giving temperatures in the mid 80s with my (now delidded) stock 8700K, in Prime95 Small FFT test. On other runs, the DIY heatsinks got mid-90s, and in a couple of extreme cases 100C and thermal throttling - in one of those tests I then wiggled the heatsinks slightly and the temperature dropped to mid 80s. So, clearly, there is some issue about proper contact between the heatsink and CPU. Extra copper shim, or soldered vs. thermal paste'd shim did not make much of a difference. (For comparison, with proper mounting pressure, i.e. screwed into the mainboard, the L9i gives me mid 70s in Prime95 Small FFT.)

So, overall this is mixed results. On the one hand this is not too bad I think, given that the GPU heatsink looks to be able to match one of the best low-profile coolers. On the other hand, performance was very inconsistent, and more imporantly, I can't see a way to mount any of these vapor chamber heatsinks properly. For one, there is barely any space around the CPU socket for a mount adapter. Even more crucially, none of my attempts to 3D print this in various different plastics were nearly stiff enough to provide decent mounting pressure. And anything metal is way too expensive to even think about seriously.

I do wonder if the vapor chamber approach in principle has potential and it's just that the GPU heatsink is not very optimised for CPU cooling, or if there's something fundamental that makes heatpipes more suited to CPU cooling that vapor chambers. One thing that occured to me is that the L9i has much closer fin spacing than either the GPU heatsink or the T318. I imagine that might be because both of those are meant to have airflow front-to-back instead of top-down, which I imagine would equate higher air resistance for a given fin spacing. The T318 on top of that is meant to be used with only case fans providing airflow, so probably relatively little static pressure. In any case, the denser fin spacing gives the L9i comparatively more surface area to work with, so maybe that is one reason why the vapor chamber heatsinks can barely match it. Part of me is tempted to try and solder a denser fin stack onto a vapor chamber, but I don't think I could do that with the required precision.

I will leave this be for now, and will just stick with the L9i myself. It's clearly good enough for even an 8700K after delidding, so no need to overdo it. Plus I've procrastinated more than enough with this already. ;) Thanks everyone who gave comments and advice!
 

Ej24

Gawd
Joined
Jun 11, 2016
Messages
628
Nevermind. Curiousity got the better of me, and I ordered myself a Noctua NH-L9i for comparison. Results are mixed. For most of my tests, I compared different heatsinks (GPU heatsink with or without copper shims, T318 with one or two copper shims, and the L9i heatsink), all just lying on top of the CPU. That's not real-world conditions - on the one hand this was open-air, no case; on the other hand there wasn't as much mounting pressure as there would be had the heatsinks been screwed in. I've tried heatsink just lying there, or manually pressing down on them. I'm guessing that that would affect all of them roughly in the same way, but obviously a big source of uncertainty. Now, for all of my DIY heatsinks, results varied *hugely* between tests. I looks like the exact position makes a big difference. My hypothesis is that especially without a copper shim (or on the T318 with just one copper shim), there is some spots where the bottom of the vapor chamber might *just* make contact with some of the mainboard components - or not. Things got a little more consistent with an extra copper shim for more clearance, but there was still a lot more variance than with the L9i. I've also tried soldering a copper shim onto the base of the GPU vapor chamber to see if that made a difference compared to just sticking it on with thermal paste. (It didn't.) I did try one test with the best (sturdiest and best fitting) 3D-printed adapter I have managed so far, only to find that (a) performance was worse than without any mounting at all, and (b) after the test I found that the adapter had a crack, and fairly easily snapped in two when I tried to pull on one of the corners. Presumably the force from tightening the screws went into that, instead of into pressing the vapor chamber onto the CPU.

To get to the point: On the best runs, the DIY vapor chamber heatsinks did about as well as the L9i, giving temperatures in the mid 80s with my (now delidded) stock 8700K, in Prime95 Small FFT test. On other runs, the DIY heatsinks got mid-90s, and in a couple of extreme cases 100C and thermal throttling - in one of those tests I then wiggled the heatsinks slightly and the temperature dropped to mid 80s. So, clearly, there is some issue about proper contact between the heatsink and CPU. Extra copper shim, or soldered vs. thermal paste'd shim did not make much of a difference. (For comparison, with proper mounting pressure, i.e. screwed into the mainboard, the L9i gives me mid 70s in Prime95 Small FFT.)

So, overall this is mixed results. On the one hand this is not too bad I think, given that the GPU heatsink looks to be able to match one of the best low-profile coolers. On the other hand, performance was very inconsistent, and more imporantly, I can't see a way to mount any of these vapor chamber heatsinks properly. For one, there is barely any space around the CPU socket for a mount adapter. Even more crucially, none of my attempts to 3D print this in various different plastics were nearly stiff enough to provide decent mounting pressure. And anything metal is way too expensive to even think about seriously.

I do wonder if the vapor chamber approach in principle has potential and it's just that the GPU heatsink is not very optimised for CPU cooling, or if there's something fundamental that makes heatpipes more suited to CPU cooling that vapor chambers. One thing that occured to me is that the L9i has much closer fin spacing than either the GPU heatsink or the T318. I imagine that might be because both of those are meant to have airflow front-to-back instead of top-down, which I imagine would equate higher air resistance for a given fin spacing. The T318 on top of that is meant to be used with only case fans providing airflow, so probably relatively little static pressure. In any case, the denser fin spacing gives the L9i comparatively more surface area to work with, so maybe that is one reason why the vapor chamber heatsinks can barely match it. Part of me is tempted to try and solder a denser fin stack onto a vapor chamber, but I don't think I could do that with the required precision.

I will leave this be for now, and will just stick with the L9i myself. It's clearly good enough for even an 8700K after delidding, so no need to overdo it. Plus I've procrastinated more than enough with this already. ;) Thanks everyone who gave comments and advice!

I'm amazed the nh-l9i can handle the 8700k even with delidding.

The only differences with gpu's and cpu's that I can think of are, direct contact vs IHS and gpu's dies are as much as 2-3 times larger, so more surface area to transfer heat. Maybe cpu is more spikey in performance/heat output? Maybe gpu's are more steady, they're either going hard or at idle.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
I'm amazed the nh-l9i can handle the 8700k even with delidding.

The only differences with gpu's and cpu's that I can think of are, direct contact vs IHS and gpu's dies are as much as 2-3 times larger, so more surface area to transfer heat. Maybe cpu is more spikey in performance/heat output? Maybe gpu's are more steady, they're either going hard or at idle.
Yeah, I was surprised too, from everything I had heard I thought none of the existing off-the-shelf coolers could - hence my interest in the vapor chamber coolers to begin with.

Direct contact and die size are the two main things I can think off too. Not sure about spikey heat output on CPU - in Prime95 I think it's steadily full on. ;) Maybe also the front-to-back airflow in reference cards? Though my Vega Nano also has a vapor chamber heatsink, and a top-down fan. Regarding die size, I'd almost think that the smaller die would benefit _more_ from a vapor chamber, since the whole point of the vapor chamber is to spread the heat across a larger surface. I do suspect that it's more my execution here that's the issue, than the vapor chamber concept itself. That's all just educated guesses though.
 

thehack

Limp Gawd
Joined
Jan 16, 2016
Messages
155
I'm amazed the nh-l9i can handle the 8700k even with delidding.

The only differences with gpu's and cpu's that I can think of are, direct contact vs IHS and gpu's dies are as much as 2-3 times larger, so more surface area to transfer heat. Maybe cpu is more spikey in performance/heat output? Maybe gpu's are more steady, they're either going hard or at idle.

I think gpu also spread the heat more evenly on their already larger die.

GPU loads are highly parallel so in general the load and thus heat is evenly distributed.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
I think gpu also spread the heat more evenly on their already larger die.

GPU loads are highly parallel so in general the load and thus heat is evenly distributed.

Hm, with all the GPU talk, I'm actually getting tempted to try putting the L9i on my GPU, just to see what happens the other way round.... ;)
 

Ej24

Gawd
Joined
Jun 11, 2016
Messages
628
Hm, with all the GPU talk, I'm actually getting tempted to try putting the L9i on my GPU, just to see what happens the other way round.... ;)

Saw a random video on YouTube a while back. Someone put a nh-d15 (or other large tower cooler) on a gpu. Absurdly low temps. It's doable. Just gotta remember to cool the vrm too.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
Saw a random video on YouTube a while back. Someone put a nh-d15 (or other large tower cooler) on a gpu. Absurdly low temps. It's doable. Just gotta remember to cool the vrm too.
Interesting. Taking that as true fact for a moment (I can't be bothered to take everything apart again to try myself at the moment), this would suggest two things to me. One, GPUs are indeed just easier to cool, and the performance of the vapor chamber heatsink on the CPU is as good as could be expected. Two, there must be another reason why vapor chambers are used then. With reference GPUs (and 1U server CPU heatsinks), I can imagine that front-to-back airflow may be the main reason, plus very tight space constraints. Heatpipes would probably just be in the way. But there's also some (few) other instances where vapor chambers are used, e.g. in Sapphire's limited edition Vega cards. There, as far as I understand, a vapor chamber makes contact with the GPU die to spread the heat. This is to give them a larger surface to which they can then attach heatpipes, which transport it away to a large fin stack. Now, assuming this isn't just a marketing gimmick, there does seem to be a benefit to vapor chambers even when used in conjunction with heatpipes. That does make me wonder if there's be scope for using vapor chambers in CPU cooling, whether in SFF or elsewhere. As far as I can tell, there is not a single vapor CPU chamber cooler on the market outside of socket 2011/2066 1U/2U server heatsinks.

Don't mention GPU VRMs to me - at this point they are the only source of audible noise in my system. (Grrrrrrrrrr. Though that's a different story.)
 

Ej24

Gawd
Joined
Jun 11, 2016
Messages
628
Interesting. Taking that as true fact for a moment (I can't be bothered to take everything apart again to try myself at the moment), this would suggest two things to me. One, GPUs are indeed just easier to cool, and the performance of the vapor chamber heatsink on the CPU is as good as could be expected. Two, there must be another reason why vapor chambers are used then. With reference GPUs (and 1U server CPU heatsinks), I can imagine that front-to-back airflow may be the main reason, plus very tight space constraints. Heatpipes would probably just be in the way. But there's also some (few) other instances where vapor chambers are used, e.g. in Sapphire's limited edition Vega cards. There, as far as I understand, a vapor chamber makes contact with the GPU die to spread the heat. This is to give them a larger surface to which they can then attach heatpipes, which transport it away to a large fin stack. Now, assuming this isn't just a marketing gimmick, there does seem to be a benefit to vapor chambers even when used in conjunction with heatpipes. That does make me wonder if there's be scope for using vapor chambers in CPU cooling, whether in SFF or elsewhere. As far as I can tell, there is not a single vapor CPU chamber cooler on the market outside of socket 2011/2066 1U/2U server heatsinks.

Don't mention GPU VRMs to me - at this point they are the only source of audible noise in my system. (Grrrrrrrrrr. Though that's a different story.)

There was one vapor chamber cpu cooler a few years ago but it was not very good according to most tests. The ID Cooling IS-VC45. People bring it up every now and then but it's just never performed well. Presumably not all vapor chambers are created equal. Gpu vapor chambers are built to handle 250W+ after all.
 

sdfewfe

Weaksauce
Joined
Dec 16, 2015
Messages
92
There was one vapor chamber cpu cooler a few years ago but it was not very good according to most tests. The ID Cooling IS-VC45. People bring it up every now and then but it's just never performed well. Presumably not all vapor chambers are created equal. Gpu vapor chambers are built to handle 250W+ after all.
Yikes. Yeah, reviews of that are terrible. It looks like they also have a few tower coolers with a vapor chamber base plus heatpipes on top of that, similar to the Sapphire limited edition system. And some even with "heatpipes" directly integrated into the vapor chamber. Alas, none of those have particuarly good reviews either. That's surprising - with the IS-VC45 I could imagine that perhaps that design just doesn't work so well with top-down fans, same as the T318 and the GPU heatsinks, but with the tower coolers that wouldn't apply. So perhaps there really is some fundamental reason why vapor chambers don't work well with CPU cooling.
 

Hanakuso

[H]ard|Gawd
Joined
Mar 3, 2011
Messages
1,463
Yikes. Yeah, reviews of that are terrible. It looks like they also have a few tower coolers with a vapor chamber base plus heatpipes on top of that, similar to the Sapphire limited edition system. And some even with "heatpipes" directly integrated into the vapor chamber. Alas, none of those have particuarly good reviews either. That's surprising - with the IS-VC45 I could imagine that perhaps that design just doesn't work so well with top-down fans, same as the T318 and the GPU heatsinks, but with the tower coolers that wouldn't apply. So perhaps there really is some fundamental reason why vapor chambers don't work well with CPU cooling.
The IS-VC45 performs much better once it's lapped but still not the best.
 
Top