Overclockers have always been a bit extreme – we take hardware many people are perfectly satisfied with and then push every last MHz we can out of it within our cooling limits. Some choose air cooling, some choose water cooling. There are a few brave souls that use water chillers, but mostly those are benching operations only.
Then there are the, well, for lack of a better phrase, there are the crazy people. I’m proud to be one and many members of our benchmarking team are too, as are a growing number of the overclocking crowd. You see, when you get into benchmarking, air and water just aren’t enough to satisfy your thirst for that extra 3D Marks or thousandth of a second. No, we truly go to extremes using things like dry ice (with acetone to help spread the cold) and liquid nitrogen.
It is to these overclockers k|ngp|n has dedicated a good portion of his craft. He started out making cooling pots quite a while ago, then turned his successful benchmarking ability and industry know-how into a career with EVGA. Thankfully he hasn’t stopped producing pots.
Today is dedicated to his latest work of art, the Dragon F1 Extreme Dark. It’s very functional of course and we’ll get to that, but this thing is a great looking piece of hardware. Don’t take my word for it though, have a look for yourself.
k|ngp|n Dragon F1 Extreme Dark
The Dragon F1 Extreme Dark (or F1 Dark) is milled out of a massive hunk of copper and retains a whole lot of mass itself. Its distinctive looking dark look comes courtesy of nickel plating on the exterior. Emblazoned with the F1 emblem many will recognize from his first F1 Extreme Edition (or F1EE, which was copper sans plating and has been discontinued for a long time now) and “KINGPINCOOLING”, branding has never looked better.
The plating stops at the interior lip and opens to un-coated copper. On the sides without markings, there is as close to perfectly smooth machining as you’re going to get.
The base is a mirror finish just like the rest of the pot.
On one side of the bottom, you’ll see a hole for a temperature probe.
For those unfamiliar, you put a K-type temperature probe in there with some thermal interface material and then run the cord up the side of the pot under the insulation so you can hook it up to your thermometer. This is absolutely necessary on any chip with a cold-bug (current- and previous- gen Intel), but also comes in handy for testing non-cold-bug chips (previous- and current- gen AMD) so you can monitor temperatures and ensure you’re as cold as you need to be.
Here’s a small shot of the interior to show the non-plated side walls. k|ngp|n requested we not show any internals for the review. They are out there if you look hard enough.
If you have ever seen one of the many photos around of the original F1EE, this design is very similar with some slight variations. The hole pattern is the same but there is more spacing between the holes, indicating there is slightly more mass than the original and that it may be a bit wider. Based on my experience with one previously, performance is very similar.
Temperature control and stability performance in cooling pots comes from two things – surface area and mass. The holes in the bottom of the pot give you the necessary surface area. There are pots with more that perform differently and are generally better for dry ice; the discontinued Koolance V2 pot is a good example. Those with more surface area tend to respond faster. The F1 Dark is built for more control and stability, generally with LN2, than those with more surface area.
That leads to weight. This pot is h-e-a-v-y, heavy. My former pot, Giraffe Pot, was no lightweight, but the F1 Dark is heavier than even it was. There is a lot of mass in the F1 Dark, which again lends itself to better control and stability.
There is a solid mounting kit included, with a black anodized aluminum back plate in which you screw four threaded rods. There are holes for most sockets currently benched including AMD sockets 754, 939, AM2 and AM3 as well as Intel sockets 775, 1156/1155 and 1366/2011.
IMOG ordered one of these shortly after they were released and showed one potential weakness of the mounting kit – the threaded rods can potentially bend in shipping. It’s not a huge deal and is still easy to use even with a slight bend, but the potential is there.
The hold-down bracket is solid with a slightly rubbery feel. It fits the pot snugly and is made to be plenty durable throughout many sessions. It ought to be – a replacement hold-down set is $55.00.
Not photographed is neoprene insulation that should be included. You get a motherboard-sized sheet of it, which is handy as a benching platform if you don’t have a benching station; you can see this in the photographs below, and a length of tubular insulation for the pot itself. If you find yourself needing more, the proper inner diameter is 2-5/8″ (by 1/2″ thick) and can be purchased will-call from Grainger for $10.50 per six-foot length, which is plenty…and then some.
F1 Extension – Full Pot Goodness
If you bench AMD much, you’ll quickly learn the value of full-pot benching. Fill-it-and-forget it (temporarily of course) is a very fun way to bench on a chip with no cold bug. To help with that you can also order (separately) an F1 Extension. These are slated to be anodized in the future but currently available models are raw aluminum, which is fine by me…they look good contrasting with the F1 Dark. The extension can be had for an additional $55.00.
$55.00 seems a bit steep for a relatively simple piece of aluminum, especially considering the pot’s cost, but where else do you plan on getting one?
Quick Giraffe Pot Side-by-Side
Like the heading says, here is a photo of the two side-by-side.
You can see why it’s was dubbed Giraffe Pot. The internals of Giraffe pot offer more surface area (reference this post) than the F1 Dark, so it probably pulls down a little faster with DICE. The F1 Dark weighs more and it shows when benching with a full pot, though to its credit, a full Giraffe pot lasts longer than a full F1 Dark, for obvious reasons.
With regard to installation, the F1 Dark comes out ahead by far, it was much easier to install and the hold-down bracket material is much more solid and able to cope better with the extreme temperatures at which it will be used. Giraffe Pot’s bracket literally shattered in four pieces and is super-glued together.
Installation is pretty straightforward. Find the holes that match up with your desired socket and thread the rods into the backplate. Then you put that through the holes in the board (put some insulation in between the bracket and the board!).
Then you use black plastic inserts to guide the springs on the top and bottom. Four supplied thumbscrews complete the job.
Using the extension is similar, but you put the extension on before the hold-down bracket. The bracket fits the extension the same as it does on the pot.
Obviously I didn’t install insulation here. That is absolutely necessary to bench sub-zero. These are just example pictures. For full installation, we’ll take a moment and go through a quick tutorial on how I insulate.
Installation With Insulation – a Small Tutorial
One of our senior members – rdrash – mentioned he uses the method I use a long time ago. When researching which method would work best for me, I settled on his example for reasons we’ll get into later. Everyone is different and there are many available insulation options, but this is the method I go with.
First, go to a hardware store (or web site) and grab a roll of 12″ x 1/8″ x 15′ Frost King adhesive foil and foam pipe insulation. One roll will last a long time.
To use this stuff you have to use a heat gun (or hair dryer) and separate the foil from the insulation. It’s actually a relatively painless process. After that you have a nice blank slate of 1/8″ thick insulation to work with. For Intel I remove the socket retention mechanism (AMD doesn’t have one of these to worry about), so be very careful to hold the CPU in place when you’re working with this stuff, lest you bend socket pins.
- I cut three layers of insulation with all of my installs.
- The first goes around the socket fully (outside the plastic part).
- Then I place a very thin bit of artist’s eraser to seal the area between IHS of the CPU and the edge of the socket, which is probably unnecessary but better safe than sorry.
- Cut a second layer to go on top of the first and over the plastic part of the socket on which you just placed eraser, right up to the IHS on the CPU.
- Note the first two layers are usually harder than third because you have to account for stray capacitors, fan headers, etc. around the socket area.
- Then cut a final layer to go over the other two. Don’t cut holes for the stray caps/fan headers on this one, it goes on top.
Sorry there are no interim photos, but here is what it looks like with all three layers.
Then cut one or two (depending on how much room you have between your board and benching station) layers to go underneath the socket between the board and the backplate.
Once you’ve done that, it’s time for a quick paper towel layer just in case something thaws enough to let off moisture. I put one folded-over paper towel layer between the backplate and the insulation and two folded-over layers over the top insulation.
After that, it’s time to clean the CPU & pot base and install your chosen TIM. I go with Arctic Silver Ceramique (or, in this case, Ceramique 2) because it tolerates low temperatures and is dirt cheap in giant tubes. It isn’t the best but it gets the job done.
It’s okay if there is a bit too much. You want to ensure the whole CPU is covered. Excess will get squeezed out by mounting pressure.
Important side note about Ceramique – after you’re all set up and ready to bench, before dropping the temperature you need to put a load on the CPU and pull the pot up to 60 °C to allow the TIM to be viscous enough to form the best possible bond. This also acts as a good test to ensure proper contact with the CPU.
Now, install the pot!
You’ll notice there is blue tape around the pot. I install the thermal probe with TIM in its hole, then use painter’s tape to hold the wire in place as it runs up the pot. This serves double duty in that it also keeps the sides of this great looking pot clean. Now, install the included neoprene insulation.
After that, it’s time to install the hold-down bracket and screw it down.
Once you’re done, install all your hardware on your favorite benching station (or box) and you’re about ready.
After everything is set up (power cables installed, RAM installed, etc), you need to put some paper towels or a shamwow -or even toilet paper for some people – around the whole assembly to soak up any moisture that may work loose. I usually use two folded-up layers total around the whole thing, then insert some in the gaps between the RAM/PWM heatsinks and the pot. Within reason, more doesn’t hurt. It never hurts to have a little healthy paranoia.
This method is my favorite because it actually insulates. Vasoline, conformal, eraser…all of these transfer the cold (though conformal is pretty bullet proof). This method insulates the board (and other components) from the cold CPU socket. Cold doesn’t ‘creep’ like it does with other methods.
The drawback to this method is that initial setup takes a little longer because you have to cut insulation to fit every motherboard. It’s not one-size-fits-all and is tailored to each installation. The good part is, if you want to bench multiple CPUs or multiple times with the same CPU/board combo, installation once you have everything cut is a snap and much faster than removing/re-applying eraser.
It’s also very clean, leaving no residue at all. If you want to repurpose your board for 24/7 use, removal is quick, painless and you can’t tell it was ever frozen.
To test out the pot I grabbed 30 L of LN2 and went to town for about six hours one day. First, I tested control using an i7 3960X. This particular chip is an engineering sample and seemed to dislike anything below about -20 to -25 °C. Thus it was a good exercise in control to keep the CPU right around -15 °C.
I didn’t keep it there for too long because the chip is a disappointment and doesn’t go any further on LN2 than it did on water (5.1 GHz brick wall), but the pot held very easily with as little intervention as you could expect for keeping a narrow, relatively high temperature range. Two thumbs up on tight control, with a very similar feel to its older F1EE brother. Temperature variations were minimal and easily controlled with drops of LN2 now and then when running heavily multi-threaded benchmarks.
AMD, however, is much more forgiving and more fun to bench on LN2. For torturing an FX-8150 I put the extension on and went balls-to-the-wall for several hours to torture every last MHz out of that chip. The pot held perfectly. Even benching wPrime 1024M – the hottest timed benchmark, running full threading for extreme lengths of time – at 2.0+ volts, there was a meager 2 °C variance, ticking up to 3 °C for only a few seconds a couple of times and settling back down to 2 °C. Aside from that, unless you let the LN2 run low, it didn’t really move much.
Compared to Giraffe Pot, the F1 Dark is a better performer. The F1 Dark has less surface area and more mass, which ends up with a more easily controlled and stable experience, holding temperatures a little better than Giraffe pot could. Holding temperatures with Giraffe Pot at 1.8 V on this chip was similar to the F1 Dark at 2.0+ V.
After several hours at -196 °C you do end up with quite a lot of frost buildup. This is where due diligence in insulation will help.
If you have fans circulating air around at the right angles, you can even end up with some frosty art for your viewing pleasure.
While the F1 Dark performed admirably, the Bulldozer chip didn’t give the MHz boost I wanted. Eight GHz was the goal and it was going to reach that or die trying. After benching everything you can bench at the highest clocks the CPU was capable of, I aimed for the stars. Alas, the chip died trying at 2.3 V. RIP FX-8150.
If you’re interested, here is the full list of results from the day of benching:
3D Bonus Benches
- SuperPi 1M: 11.468s
- SuperPi 32M: 11:54.813
- PiFast: 21.36s
- UC Bench: 1670
- WPrime 32M: 4.765s
- WPrime 1024M: 151.172s
- CPUz: 7651.2MHz
Final Thoughts & Conclusion
The Dragon F1 Extreme Dark is a stellar performer for extreme benchmarking, especially for anyone that uses liquid nitrogen. Tight temperature control is unparalleled and the sheer mass allows it to hold extremely cold temperatures very well under heavy loads.
If you use dry ice more extensively or exclusively, a pot that responds a little faster with some extra surface area may be better suited for your purposes. k|ngp|n’s Dragon F1 Gemini with a dry ice base (supposed to be available soon) is right up your alley, or the EoL Koolance V2 if you can find one floating around in a classified listing somewhere.
That’s not to say you can’t use dry ice with the F1 Dark; you certainly can. The main drawback will be that it takes a long time to pull down to dry ice temperatures due to the amount of surface area. Once it gets there, it will hold just well as it does with LN2. The F1 Dark would make a fine DICE pot, but its true calling is LN2.
Of course, the F1 Dark isn’t going to come cheap. In fact, it’s one of the more expensive pieces of cooling hardware at $265.00 plus shipping. In that price though is the full package – a cooling pot that will last forever (a solid chunk of copper doesn’t really need replacing. Ever.), a very solid mounting kit that works on pretty much every socket you could want from old to new and insulation to make it work for you.
At some point in its life you’ll need to get more insulation, but that’s cheap and applies to all pots. The mounting hardware is very solid and should never need replacing unless you have to buy a bracket for a new socket down the road. So consider this price an investment for the long term. Once you get an F1 Dark, you won’t need another pot unless you just want to try something else for the heck of it.
There are cheaper pots out there, mostly used and residing in classified sections out there. However, there is a reason F1EE’s are so hard to find in those same classifieds. People love them because they’re the best. I’m pleased to say the F1EE is back and better than ever, in the form of a new and improved Dragon F1 Extreme Dark. If you want the most out of your sub-zero benching experience, the F1 Dark is a worthy investment.
– Jeremy Vaughan (hokiealumnus)