Introduction
“It’s Chilly inside and it’s chilly outside.” Which fan of ‘Get Shorty’ would possibly forget Ray ‘Bones’ Barbone’s flat joke about Chilly Palmer (John Travolta) at the beginning of the movie? Working with Asetek’s new Vapochill unit in cold Europe instantly reminded me of those words. So I say, “Look at me”, as I am giving you ‘the beef’ about this Danish super cooling solution for socket processors.
You certainly remember my initial article ‘Processors On The Rocks‘ from not quite a couple of months ago, where I described the technology behind Vapochill. I strongly recommend you read this article if you should not be familiar with Vapochill yet, because in this new article I will only focus on the actual Vapochill system without regurgitating the theory behind it’s function. However, understanding the basics behind this super cooling device will be very helpful for understanding all the issues in this piece.
In the initial article I was pretty impressed with Asetek’s technology and I certainly generated a lot of positive expectations. However, we know that the proof is in the pudding and before Vapochill can get a ‘thumbs up’ of ‘thumbs down’ the very unit has to go through rigorous testing. Tom’s Hardware received Vapochill several weeks ago and we have played around with it a lot. Let’s find out if Vapochill can indeed live up to the high expectations that I had after looking at its technology two months ago.
The Manual
Who in the world would ever start a review with the manual? Well, I would, but only if a manual is as good as Vapochill’s. You can have a look at it by yourself, because it is available as PDF-file at Asetek’s website. Now don’t get me wrong here, I am fully aware that a manual is just a manual, and in fact you will find out that Asetek has all reasons to supply a manual as detailed as possible, because there are several nasty traps when working with Vapochill. However, Asetek put a lot of dedication into this manual and each complaint of Kyle or me was addressed in there to ensure a smooth setup process for Vapochill customers.
No Front Bezel Yet – Archaic Look
The first thing that obviously jumps into your eye after unpacking Vapochill is the fact that right now the systems are shipping without the front bezel, because it is not available to Asetek yet. This is of course no serious issue, as Asetek will supply each customer with the missing front bezel as they are ready to ship it. However, as of right now, Vapochill doesn’t exactly come with the looks that would impress your buddies.
You can see where the cooling unit is. You are looking at the power converter of the compressor in the upper compartment of the case. On the right there are the reset and power-on buttons and what sticks out so nicely with its 3-digit display is the ‘ChillControl’. On the bottom you see the optional huge fan that is supposed to ensure a healthy climate inside the case for the other hot components like the hard drive and the graphics card.
No Front Bezel Yet – Archaic Look, Continued
After the removal of the cover you find that the whole case is painted black. I personally don’t associate black with ‘cool’ and would have considered it fancier if the case was artic white, but I guess it doesn’t really matter anyhow. You can see that the case cover is following the old fashioned and rather unpopular U-shape design, which is a pain to remove and put on again. We would obviously have preferred one of the new designs where you can remove the single sidewalls of the case. Asetek’s Andre told me that they searched high and low for a case that could fit the cooling unit and the only company that would alter their case for Asetek was CED, Germany. Supposedly it is extremely expensive to supply a completely Asetek-designed case, which is why we have to live with the awkward case cover for the time being. I cut my fingers twice when handling the cover, so it won’t get a whole lot of plus points from me.
No Front Bezel Yet – Archaic Look, Continued
The case itself is very spacious, leaving a lot of work and breathing space for the motherboard and expansion cards. However, due to the space taken up by the cooling unit, it only hosts three 5 1/4 inch bays and three 3 1/2 inch bays. This should be enough for the majority of users, but people with streamers, CDROM, CDR/W and a life drive will already run into trouble. The other slightly annoying issue is the location of the 300W power supply. Obviously it had to be removed from its usual position in the top/rear area of the case, because that is where the condensor of the cooling unit sits. Thus the PSU can be found right in front of the motherboard, blocking the view onto the CPU and the evaporator. You will always need to remove the power supply if you should want to change the CPU or work on the evaporator.
The Refrigeration Unit
You already spotted it in the upper compartment of the case, the refrigeration unit is obviously the most integral part of Vapochill, as it is responsible for the minus-degrees of your CPU. Here is a schematic view of it.
The next picture shows the actual refrigeration unit after it had been removed from the Vapochill case.
In the front right you can spot the condenser with the attached fan, then there is the receiver, the capillary tube and the actual compressor with its power converter. In the front left you see the Vapochill evaporator and the thermally insulated hose that leads the gaseous refrigerant back to the compressor and the liquid refrigerant from the capillary tube to the evaporator. If you want to understand how this unit works and you haven’t read my initial Vapochill article yet, then it might be a good time to do so now.
The Evaporator
The most sophisticated part of Vapochill is obviously the evaporator, which gets attached to the CPU and cools it down to those chilly temperatures we want. This is the schematic view of it.
The innards of the Evaporator are no black magic, but you can see that the whole setup is quite complex.
The Evaporator, Continued
Here’s an actual photograph of the evaporator. Asetek has certainly done a pretty eye-appealing design.
Inside the ‘clam shell’ you find a lot of insulation material and the actual copper-made evaporator itself plus a thermal sensor attached to it, which feeds the ChillControl with actual temperature information of the evaporator.
After installation the evaporator sits right on top of the processor die. It is just about as thick as a finger and about an inch high. Although that may not be big at all you can certainly still imagine how much larger it is than the CPU-die it is supposed to cool. This situation resembles one of the Vapochill’s current problems, because such a comparably huge piece of copper can easily smash a little processor-die if it is not handled carefully enough.
The ChillControl
If we say that the refrigeration unit is the muscle of Vapochill, the ‘ChillControl’ has to be seen as its brain. It processes the temperature read by the thermal sensor attached to the evaporator and switches the system according to it. The buttons let you set two different temperature limits, T1 and T2.
T1 marks the temperature at which you are afraid that your CPU might get too hot. It’s a safety limit and shuts down the motherboard as well as the heating element to guarantee that your processor won’t get too hot. It’s only needed right after you turned on Vapochill, because it takes some time until the refrigeration unit has built up enough pressure to initiate proper cooling. In those few minutes after the system has been turned on, the T1-limit makes sure that your CPU doesn’t reach any dangerous temperatures. T1 is set to +30°C / +86°F by default, which is reasonable and safe.
T2 is a very important limit. Once the temperature of the evaporator has dropped below the T2-limit, the ChillControl releases RESET and the system comes on. As soon as this limit is exceeded ChillControl will apply RESET and the system will be held at this stage until the temperature dropped lower than this limit again.
You can see that T2 should neither be too high nor too low. By default it is set too -5°C / +23°F, which seems like a good compromise. Very heavy overclocking might require a lower temperature limit, but only if Vapochill can actually reach it, because otherwise your system will either never come on or it will go into reset very easily once the temperature has been exceeded. You need to experiment with the T2-limit a bit, because it really depends on your processor and its clock speed. Celeron and Pentium III processors reach lower temperatures than Athlon and Duron CPUs. Thus the T2-limit can be lower for Celeron and Pentium III. Setting an Athlon/Duron-system to -15°C/+5°F is very unwise, because it might never come on or switch into reset very often.
Here’s a schematic picture of the ChillControl:
And that’s an actual photograph.
The glue you can spot on to of the housing has only been supplied to keep in a reasonable position as long as the front bezel hasn’t been supplied. It normally sits inside the front bezel and does not require this glue.
The Heating Element
You remember from the explosion diagram of the evaporator that between it and the processor there is a heating element that heats up the processor pins to keep water from condensing on them, which could cause short circuits. Please refer to my initial Vapochill article if you require more information.
Asetek supplies a few of those self-adhesive heating elements, which need to be hooked up to the ChillControl.
The Power Supply
The 300W power supply is from ‘Fortron/Source’, model ‘FSP300-60GT’ and thus pretty basic. It does not only supply the system, but also the 12V compressor, which requires some 55W. Please realize that the power supplied by the compressor stands in no direct energy relation with the amount of heat the compressor is able to remove. A typical example to understand it would be if you cool down your frying pan with some water from the tab (don’t do that if there’s boiling oil in the pan or you’ll have a fiery surprise!). The energy of splashing water onto the pan (which is obviously minimal) is also in no way related to the huge amount of heat the water is removing from the hot pan.
The power supply is hooked up to the ChillControl and the compressor and one of the ‘power-good’-lines to the motherboard can be interrupted by the ChillControl, which happens if the T1-limit should have been exceeded.
Extra Fan
Optionally Asetek supplies a huge extra fan that can be mounted to the front opening of the case. This is for people who are particularly religious about the temperature inside their case or who have very hot graphics cards or hard drives.
It’s a Papst-Fan, model ‘4412 FGMLD’ and very quiet.
The Mounting Process
The manual does an excellent job of walking you through the mounting process, but I decided to publish this whole process here as well, because it can be very tricky.
First of all you obviously mount the motherboard into the case, which is done just as with any other average case. Then you place the insulation pad inside the processor socket.
Before …
… and after. Asetek supplies pads for Socket370 as well as Socket462. The one for Socket 370 is obviously a bit larger.
The Mounting Process, Continued
Asetek suggests that you cover the whole socket with heat conduction compound (included in the Vapochill shipment in large quantities) to reduce the risk of condensation. It’s a messy job and we decided against it, because it messes up the motherboard and the processor. However, we only ran Vapochill for a few weeks and not for several months. Therefore I suggest that if you are certain that you will stick to the CPU and the motherboard, you follow Asetek’s advice and smear the thermal compound all over the poor socket. We cannot supply any pictures of that, since I just couldn’t do that to any of my test motherboards. We did not run into any problems due to condensation, but we never ran the systems for more than a few days.
Now you place the CPU into the socket. You will need to push down the CPU while closing the lever, because the insulation pad is higher than the socket and pushes against the processor. Either you have attached the heating element to the processor already before its insertion, or you need to stick the little heating foil onto the processor now. It is probably wiser to do it after the insertion of the CPU, because otherwise you may fix the heating element to the CPU with the cable going into the wrong direction.
The cable should obviously not go across the socket lever, nor should it run towards the mounting notches of the socket where the clamps of the evaporator kit will be fixed. Therefore there’s really only one direction in which the cable should go, which is shown in the picure.
So far everything was very easy, but now there comes the tough and dangerous part. The mounting of the evaporator kit seems very easy, but it bears the high risk that you might damage the die of your processor, especially if it should be an Athlon or Duron CPU.
The Mounting Process, Continued
First of all you need to loosen the screws that hold the clamps at the retention clip as much as possible, ensuring that the clamps come down as far as possible. Then you need to align the retention clip according to the processor socket.
After putting some thermal compound onto the processor die or the evaporator, you place the evaporator kit onto the CPU, paying UTMOST attention to holding the evaporator perpendicular to the processor. In case of Celeron and Pentium III processors it is pretty easy, but with Durons and Athlons you should be extremely careful. Try to attach the clamps to the notches of the socket. Do this by gently applying pressure to the evaporator, still ensuring it is positioned perpendicular, because otherwise you will hear some lovely grinding noises from the die of your processor as the evaporator breaks off parts of its edges. Once the clamps are attached to the notches, you’ve deserved to take a deep breath, because the hardest part is over. You should see something like this now:
The evaporator sits on the CPU-die, which you unfortunately cannot see. The clamshell is still ‘hovering’ a quarter of an inch above the heating element. Make sure the clamshell is positioned properly, so that it only needs to move down towards the CPU, without the need of shifting it in any direction, as that will slide the evaporator along the processor-die, possible misaligning or damaging the die.
The Mounting Process, Continued
Now you carefully and slowly tighten the screws. Don’t tighten one completely and then the other, but tighten the screws a little bit alternately so that the clamshell moves down onto the heating element slowly and equally. Don’t over-tighten the screws, because you could damage the socket by breaking off the notches.
Now the evaporator is fixed to the socket. It should not move around, but be tightly fixed to the motherboard.
Finally you need to reattach the power supply. Don’t forget to plug in the power supply connector, because you might have to remove the power supply again to do that.
Voila! Now you can attach all your expansion cards and the cables and fire Vapochill up.
Thermic Acceleration Results
Asetek doesn’t like the term ‘overclocking’ in combination with its products and calls it ‘thermic acceleration’ instead. I know that many of you will be disappointed, but I decided against posting any benchmarks with the ‘thermicly accelerated’ processors, as this is not the point of this article. If you want to see some benchmark scores of an Athlon running at 1400 and 1466 MHz, you should have a look at my recent article ‘Pentium 4 – Final Recount‘.
Let’s see how far we could get the different processors with Vapochill. To produce successful results you always need to increase the processor voltage. This is not as dangerous as usual, because it is impossible to burn a CPU that is cooled down to sub-zero temperatures. We did not modify any SocketA-motherboards to achieve voltages in excess of 1.85 V, and we doubt that the results would have improved. Modifying the voltage regulator on a motherboard goes far beyond the installation efforts required for Vapochill.
We tested with the following motherboards:
- Asus CUSL2 – Intel i815
- Asus A7V133 – VIA Apollo KT133A / VIA VT82686B (pre-release)
- Gigabyte GA-7DX – AMD 761 / VIA VT82686B (pre-release)
- Iwill KA266 – ALi MaGiK1 (pre-release)
- MSI MS-6341 – AMD 761 / VIA VT82686B (pre-release)
You can see that we only used SocketA-motherboards that allow a bus clock of 133 MHz, because otherwise it gets very difficult and pointless to overclock Athlon or Duron. The KT133A boards such as the Asus A7V133 will become available very soon.
We used the following CPUs:
- AMD Duron 750
- AMD Duron 800
- AMD Duron 850 (pre-release)
- AMD Athlon 1000
- AMD Athlon 1100
- AMD Athlon 1200
- AMD Athlon 1200/133
- Intel Celeron 566
- Intel Celeron 600
- Intel Celeron 667
- Intel Celeron 700
- Intel Pentium III 850E
- Intel Pentium III 866EB
- Intel Pentium III 933EB
- Intel Pentium III 1000EB
Results
- Intel Celeron
We tried all kinds of different Celeron processors, but never managed to get any further than 1100 – 1150 MHz. It turned out that the Celeron hits some magical barrier at this clock, which it is not able to pass. The voltage was raised up to 2 V, the evaporator temperature was about -15°C / 5°F at 1133 MHz. - Intel Pentium III
We ran into the same problems as with Celeron. 1150 MHz is the limit, regardless which CPU is used. Therefore it seems as if a Pentium III 850E is the best Pentium III processor for usage with Vapochill, as it runs reliably at 1133/133 MHz with an evaporator temperature of about -14°C / 6.8°F. The voltage was raised to 2 V. - AMD Duron
It turned out that current Durons are by far less overclockable than their big brothers. Duron goes up to 1066/133 MHz and that’s it. Even 1100 MHz would never run reliably, more was simply impossible. The voltage was raised to 1.85 V, the evaporator temperature was about -9°C / 15.8°F at 1066 MHz. - AMD Athlon
Athlon was clearly the processor with the highest potential. We were able to run Athlon 1200 processors at up to 1485 MHz reliably. It seems to hit a magical barrier at around 1500 MHz. Some Athlon 1100 or even Athlon 1000 processors were able to reach 1333/133 or even 1400/133 MHz as well. The voltage was set to 1.85 V, the evaporator temperature was about -7°C / 19.4°F at 1485 MHz. - Pentium 4
We managed to run Pentium 4 successfully at 1888 MHz with a slightly modified bracket. I am sure that Pentium 4 would have even reached 2 GHz if the multiplier would not be restricted to x16. Therefore I had to increase the clock of the quad-pumped processor bus to 118 MHz. At 1920/120 MHz Pentium 4 would crash in Windows, at higher speeds it wouldn’t even reach POST. The PC800 RDRAM was clocked at only x3 = 354 MHz, because it would not run at x4 = 472 MHz. The speed was increased to 1.85 V, the evaporator temperature was about -10°C / +14°F at 1888 MHz.
I was just informed by Asetek that they have successfully mounted Vapochill on Pentium 4 as well.This is a first Vapochill/Pentium 4 picture.
When I say that a processor runs ‘reliably’ I don’t mean that I can just about reach the Windows desktop or even only POST. Each ‘thermicly accelerated’ processor had to go through the same tough validation as Intel’s good old Pentium 1.13 GHz. I know that there is a lot of youngsters out there that consider it successful overclocking once they managed to take a screen shot of the POST-screen. What we do here is real stuff.
Good Things
Let’s look at the positive sides of Vapochill first:
- Thermic Acceleration
This is obviously the very reason why people would buy Vapochill and it is clearly its main benefit. Right now it provides almost 25% acceleration to the fastest AMD processor and it will soon accelerate Intel’s Pentium 4 processor as well. - Very Low Noise
Vapochill is quieter than any normal PC unless it should be particularly noise reduced. The compressor is almost inaudible and the fan in front of the condenser is a quiet and high quality Papst-fan as well. Even the optional Papst-fan for the front of the case adds only very little noise. - VersatilityM
It is certainly a big plus for Vapochill that you can use it on a wide variety of motherboards and processors. However, providing a universal product bears a lot of risks too. - Weight
Vapochill is no light weight, but its 20 kg / 44.5 lbs are nothing in comparison to Kryotech’s monsters. As long as you are of average strength you should be well capable of carrying Vapochill around and e.g. bring it to LAN-parties.
- Thermic Acceleration
Problems
There’s always a dark side and Vapochill has got a few problems too.
- Mounting – Details
- Evaporator not sitting properly
It happened to me quite often that I would fire up Vapochill and then be unable to reach any reasonable ‘acceleration’-results. This is often due to a badly aligned evaporator. Unfortunately it is impossible to see if the copper-piece sits properly centered on the CPU-die. I found out that removing the evaporator kit and putting it back on more carefully would solve the problem. However, each removal/installation procedure of the evaporator bears the high risk of damaging the processor-die. - Risk of destroying the CPU
It is a very serious problem that Vapochill can easily damage the edges of the processor die especially of Athlon and Duron processors. I had a whole lot of discussions about this issue with Asetek and we came up with several solutions that should make this touchy mounting process a whole lot easier. One of the ideas to save the die is this one:
Those little pieces of plastic or (as in this case) insulation material can save the CPU-die. Vapochill is currently experimenting with them and will hopefully include them soon. - Evaporator falling off
In a worst case scenario you are taking the risk of the evaporator falling off the CPU if the brackets don’t sit well enough or if you over-tightened the screws and the notches break off the socket. In this case an Athlon/Duron processor would die instantly. Asetek is now putting those little spacers between the retention clip and the brackets, to make sure that the screws don’t get over-tightened.
- Evaporator not sitting properly
It happened to me that the MSI MS-6341 motherboard would not boot properly, because each time the compressor came on (a few seconds after you start Vapochill) the board would shut down. I managed to run the board in Vapochill by disconnecting the RESET-connector that comes from the ChillControl. As soon as the compressor had started I was able to reconnect the RESET-plug and the system booted normally. The reason behind this strange behavior turned out to be the power supply. It was unable to keep the voltage steady enough when the compressor started and the MSI-board seems to be very sensitive to little voltage spikes in the RESET-line.
In some cases the processor-socket is so far away from the refrigeration unit that the hose to the evaporator kit is not long enough to reach the processor. In this case you cannot use Vapochill, at least as long as you leave everything in the case. Asetek pointed out to me that they do not say that they support every motherboard, but a large variety of motherboards. It is definitely wise to be aware of the hose-length issue before choosing the wrong motherboard for Vapochill. Vapochill’s website supplies a list of supported motherboards.
I discussed this issue with Andre and he suggested to offer two different hose sizes. We have to realize that it is no solution to simply supply a longer hose. If the hose is too long it gets bent and pushed against the housing, which compromises its insulation and thus bears the risk of condensation.
Overall Valuation
Vapochill is certainly a very fascinating solution to run a processor way beyond its spec. However, it is most certainly not a solution for everyone.
The installation of Vapochill is for technically skilled people only. Clumsy guys should better leave their hands off it. It is most important that Asetek reduces the risk of processor damage. The PSU-issue with the MSI motherboard also shows that Asetek has to do a whole lot more compatibility testing. The customer should not be burdened with that. Asetek was very courageous to position Vapochill as a generally useable product. It is obviously much easier to tailor such a complex cooling solution for one special platform only. We will see if Asetek will be able to solve all the problems and keep their customers happy. There is still quite a way to go and there are quite a few big stones in Asetek’s way.
The price of $650 is not bad but still pretty high. Interested customers should know exactly what they want to do with it. Let’s have a look at the different processors types and the sense or nonsense of purchasing Vapochill in combination with them. There is no point whatsoever in buying Vapochill to overclock Celeron or Pentium III processors. An Athlon 1200 processor is cheaper than Vapochill and faster than even the most overclocked Celeron or Pentium III. You simply can’t get those two processors to 1200 MHz or more and the simple purchase of an Athlon 1200 and a normal fan would be cheaper as well as faster than Celeron or Pentium III plus Vapochill. The same is valid for Duron, which lags behind Athlon in terms of ‘acceleratability’ as well as performance too. This leaves only AMD’s Athlon. However, even in case of a Vapochill/Athlon combination you require a platform that lets you use an FSB-clock of 133 MHz and beyond. If you want to round that up with top system performance you will require DDR-SDRAM support as well. Thus the most sensible Vapochill-platform right now is Athlon on AMD760 plus DDR-SDRAM, followed by ALi MaGiK1 plus DDR and then VIA Apollo KT133A plus SDR.
Things could look different once Vapochill supports Pentium 4 as well. We already know of the huge overclocking potential of Pentium 4, which I tested at 1728 MHz with air-cooling already. The 1888 MHz we achieved with our own modification to Vapochill’s mounting bracket aren’t shabby either. However, Intel will most likely lock Pentium 4’s multiplier and it remains to be seen how far we can overclock Pentium 4’s quad-pumped processor bus to reach clock speeds that are worth the $650 for Vapochill. At least there is no risk of damaging the processor die when mounting the Vapochill evaporator on a Pentium 4.
For me as hardware tester Vapochill is a gift from the gods. Once you have installed the evaporator more than 50 times you can do it within a few seconds. It enables me to test processors in completely different ways. I love the low noise, which was only made possible by Asetek’s decision to use high-quality fans and an excellent compressor.
All in all I suggest a mature decision after a thoughtful consideration of all the facts that I presented you here. What I can assure you of is that Asetek is a very professional and very dedicated company that has a lot of future potential. In my first article about Kryotech I already told you that in my expectation future processors will soon require this kind of cooling, because air-cooling won’t be sufficient anymore to remove several hundred Watts of heat. Right now Asetek is at least one step ahead of Kryotech.
Kyle Bennett from HardOCP received a Vapochill evaluation sample as well. His review should be available soon too. Kyle is a hardcore overclocker and might have a different view of Vapochill.