Table of Contents
It’s not often that we see new products developed and marketed exclusively for the subzero extreme overclocking world. There are only a handful of well-established brands that cater to the extreme overclocking community, so the overall market is quite small and generally slow to change. Today. we are excited to evaluate a new retail product from Bitspower International, which has been designed to be used exclusively with extreme cooling solutions, such as liquid nitrogen or dry ice.
If you’re a water-cooling enthusiast, then you probably know and love them. Since its entrance into the computer water cooling market in 2005, Bitspower International has become synonymous with high-quality and dependable products. They focus on simple designs which are highly effective and also pretty to look at.
Bitspower entered the extreme cooling market with two different offerings. Although we are only evaluating the RTX 20 series container today, they also launched a Universal Memory LN2 Container container at the same time.
For serious overclockers, liquid nitrogen is the only solution in order to push their hardware to the next level and get the highest benchmark scores. And for them, Bitspower presents the LN2 pot for NVIDIA GPU. Made out of a solid block of high quality copper, the Bitspower LN2 pot is precisely sculpted via CNC for a clean channel routing and most efficient cooling capabilities. Finished with nickel plating and a matte black anodized collar, the Bitspower LN2 pot has a great look with greater performance. –Bitspower
In the table below are the particular details of the LN2 container being evaluated today.
|eBitspower GeForce RTX 20 Series VGA LN2 Container|
|Compatibility||All RTX Series GPU’s and Nvidia G92|
|Color||Black with silver logo|
Retail Packaging and Accessories
The packaging for the Bitspower GeForce RTX 20 Series VGA LN2 Container is stripped down and simplistic. Other than the part number on the back, there really is not much on the outside of the box. While it may have been nice to include a bigger box with die-cut foam inside and glossy images on the outside, they clearly save you some money with this minimalist approach.
Meet the container
Once out of the box we get an immediate impression of quality. Many extreme overclocking cooling products don’t go the extra mile when it comes to fit and finish, because they know overclockers mainly care about results and not style. However, Bitspower clearly wants to give their customers both! The mating surface of the container has a protective film so no scratches occur during shipping and storage.
As implied by the name, the copper mass has been ‘tuned’ for the Nvidia RTX series of GPUs. This effectively means that once the copper has reached the desired temperature, introducing the 3D load won’t drastically change the overall temperature. Models such as EVGA’s 2080 Ti Kingpin Edition have been known to consume between 700 and 800 watts with heavy 3D loads under liquid nitrogen. If the container has been tuned correctly, it should handle the transition from 100 W (idle) to 750 W (3D load) without radical swings in temperature.
There is no exact recipe for tuning the load response of the container. Generally speaking, most overclockers can easily manage a 5-degree temperature delta. When the 3D load starts and the copper temperature instantly drops more than 5c in temp, it could then be considered unstable and difficult to manage.
As seen in the pictures below, the GPU mating surface is bare copper. This is something that we rarely see with extreme overclocking products, but often it’s a modification that the end-user will do. The process of electroplating can leave the surface uneven. In some extreme cases, the peaks and valleys of the surface can be up to 0.5mm difference. Bitspower has gone the mile and left the GPU contact surface as raw milled copper so as to avoid any unnecessary surface imperfection.
There is potentially a second reason they have taken these steps, and it has to do with thermal conductivity. The average nickel plating has a thermal conductivity rating of 52.4 Btu/(hr-ft-F). Pure copper, on the other hand, has a thermal conductivity rating of 231 Btu/(hr-ft-F). Even though the nickel plating layer is very thin, it acts as an insulator and does not provide for optimal thermal transfer. The reality is that a small nickel layer won’t hurt the realized maximum MHz too much, if at all. However, that didn’t stop Bitspower from giving you the best possible condition for optimal thermal transfer.
Looking closer at the container we see that it’s composed of two parts. The top part is a black anodized aluminum extender piece. The sole function of the extender is to provide more volume to contain liquid nitrogen. The extender can be removed by unscrewing three screws using the provided hex key tool. The total weight of the base and extender is 5.45 Lbs.
The extender is not sealed to the base in any way. Individuals using methods like dry ice will need to seal the two pieces. Dry ice requires a liquid component such as isopropyl alcohol or acetone, and thus, this container would leak with such a cooling method.
The rate at which the liquid nitrogen can cool the copper is directly proportional to the mass and the surface area. Those two factors are critical in determining how the container will respond to the GPU load. The machined channels in the copper provide the surface area that allows the liquid nitrogen to quickly and effectively bring down the overall copper temperature.
Bitspower crafted the mounting plates out of 3 mm thick stainless steel. Out of all the mounting hardware we have seen for extreme cooling solutions, this is by far the thickest and most sturdy. Furthermore, they include a back-plate with washers to better support the weight of the container.
While the container is marketed as being compatible with RTX series GPUs only, it will fit a vast variety of different GPUs using the included G92 hole spacing. The simple addition of a few more holes would have made this container nearly universal, in terms of GPUs that could be mounted to it. Adding 4 holes for the Nvidia G80 spacing would be a simple modification which anyone can do and would make it compatible with most Nvidia and ATI graphics cards.
The relative flatness of the mating surface is a critical aspect of its overall thermal cooling performance. In the picture above, we used Mitutoyo digital calipers to get an approximation of the flatness. It’s clear that Bitspower went the extra mile to ensure this copper is as flat as possible, with no peaks or valleys observable using this method.
The overall objective is to overclock the graphics card memory and core as far as possible. Starting out with a brand new EVGA RTX 2080 Ti XC Black, we heavily modified it to be used with liquid nitrogen. While it does not fit within the scope of this review to detail the modification process, the GPU has a hard-wired volt-mod for the core voltage as well as extensive work done to remove all of the power limitations. No bios flashing or editing was done to the card, only physical modifications.
In the table below is the particular supporting equipment we are using.
|CPU||Intel Core i5-7640X|
|CPU Cooler||Alphacool Eisblock XPX CPU Block (Custom Loop)|
|Water GPU Cooler||XSPC Universal VGA Waterblock|
|Motherboard||ASRock X299 OC FORMULA|
|Graphics Card||EVGA GeForce RTX 2080 Ti Black Edition Gaming|
|Solid State Drive||Team Group L5 LITE 3D SSD|
|Power Supply||Enermax MaxTytan 80+ Titanium 1250W|
|Operating System||Windows 10 x64|
According to the installation guide, Bitspower suggests mounting the graphics card to the container with all 8 mounting screws. Over the years, we have found that there is an optimal way to mate a heavy container with a lightweight graphics card. Starting with a flat surface, set the LN2 vessel on its back with the mounting screws pointing up. Apply thermal paste to the GPU, then bring the card down and set it on top of the copper.
Carefully apply even pressure to the back of the card and tighten the nuts in an x-pattern. We observed that the optimal method to mount the card is to tighten only the inside four nuts first. Once those have been tightened to the desired level, then screw on the outer four nuts. Utilizing the back-plate and included washers should help support the card more securely and help spread out the tension evenly.
Due to the very large die size of Nividia’s RTX 2080 Ti, the mating contact is absolutely critical for the best results. The objective here is to see a uniform paste thickness after the copper has been mounted and removed. If the paste is not uniform in thickness, problems such as paste degradation can occur with liquid nitrogen temperatures.
In the pictures below, we did a test mount and the result was acceptable, but not ideal. As you can see, the paste is thickest on the right side compared to the left side. This is an operator error and doesn’t have anything to do with the hardware. This may seem like a small detail, but in the world of extreme overclocking, this could mean the difference between 1st place and 20th place.
Temperature is critical with LN2 overclocking. A high-quality thermal probe and accurate digital thermometer are required tools of the trade. Bitspower provided a temperature probe hole, highlighted by a red arrow in the picture below. Temperature probe placement and depth are two very big factors in determining the accuracy of the temperature readings. The temperature probe hole is only about 9 mm deep. This effectively puts the temperature read point about 17 mm away from the edge of the graphics card die and about 30 mm away from the center of the die. Early estimates are that this probe location is too far away from the center of the heat source to provide an accurate reading.
The weight of the copper container needs to be supported to prevent damage to the graphics card and PCIe slot. To support the weight and absorb moisture, we build a simple shelf of chamois cloth and blue shop towels.
With the LN2 vessel mounted, and temperature probe in place, it’s always a good idea to turn on the computer. Enter the operating system to verify that everything is working correctly before using liquid nitrogen. This is intended to be a quick proof-of-life test only. Running the graphics card passively, with no active cooling, should only be done for a few minutes or until the temperature reaches a max of 50 Celsius.
Extreme Overclocking with LN2
As we started pouring LN2 from positive 35 Celsius down to minus 39 Celsius, the temperature dropped relatively slowly, about 1 degree every 2.5 seconds. The reason for this is a phenomenon known as the Leidenfrost effect. To help the container cool faster, we employ a method known as “glazing”, where we build a small layer of frost on the inside of the container using a 1500 watt heat gun. The frost layer helps defeat the Leidenfrost effect and allows the LN2 to cool more effectively. After the initial glazing, the temperature dropped about 1 degrees calculus every 1.5 seconds.
Once the copper reaches about -100 °C we are ready to start running benchmarks. As seen in the pictures below, the container is in the optimal state to begin benchmarks. Running benchmarks too early–before the copper is properly glazed and in the optimal temperature range–can have adverse effects on the thermal paste and performance.
To start things off, we tested Port Royal with the 2080Ti running 1.30 V core. Starting the benchmark with a temperature of -120 °C the copper warmed up by about 2 degrees in the first second. Small additions of LN2 (about 2 ounces) were required to maintain -120 °C. Working our way up in voltage, and down in temperature, we observed the same responsiveness to load and LN2 consumption.
In our testing, we ended up at the lowest temperature possible, around -190.5 °C . This is known as the ‘full-pot’ temperature. The ideal LN2 temperature of -196 °C is not achievable when the graphics card is running, even in an idle state. In the video below, the benchmark starts after about 29 seconds. Even with a starting temperature of -190.4 °C, the copper warmed up to about -188 °C.
The temperature probe location does not allow the probe to provide accurate readings under heavy loads. If the probe hole was drilled deeper, or closer to the graphics die in general, then the temperature readings would be substantially different. We estimate that with better temperature probe placement the readings might show values of -180 °C under load instead of -188.5 °C. While the probe location wouldn’t change the performance of the LN2 vessel, it would give a better indication of the actual temperature.
In the picture below, you can see we tested our card with voltages as high as 1.43 V for the core. Generally speaking, increasing the core voltage will increase the heat output and thus increase the copper temperature. Therefore, it can take many test runs to find the optimal ratio of heat output, copper temperature, and core voltage.
With a load temperature of about -188.5 °C and a core voltage of 1.38 V, we were able to pass the Port Royal benchmark at 2520 MHz core and 2100 MHz memory. The overall result was a score of 11566, which landed us in the #15 spot globally for all graphics cards. While it may not seem like an outstanding result, keep in mind that this is simply a reference graphics card and not one designed for LN2, such as the Kingpin Edition. In fact, this might be the only reference 2080Ti running LN2 to make it on the leaderboard. You can find all of the Port Royal hall of fame records on the 3DMark website.
We learned that the reference 2080 Ti scores are substantially lower than aftermarket card for the same clock speeds. There are several factors which contribute to scoring. However, the biggest factor, in this case, is that this card was hitting a maximum consumption of a scant 400 watts. The power consumption was held at 400 W regardless of the fact that all the ‘observable’ power limitations were removed with physical mods. As mentioned earlier, some aftermarket PCB designs can consume nearly double that figure when running liquid nitrogen.
Bitspower has made a reputation for themselves for offering high-quality water cooling products. From the fit and finish to the selection of materials, their products are highly sought after, and with good reason. In the world of extreme overclocking and liquid nitrogen, however, they are an unknown quantity. Today, we are happy to report that our overall experience with the Bitspower GeForce RTX 20 Series VGA LN2 Container is very positive.
With a removable top extension piece, over 5lbs of copper, and an incredibly well-built mounting system, the VGA LN2 Container is a home run. They impressed us by going the extra mile and removing the nickel playing from the GPU mounting surface for the best possible thermal heat transfer. For our test condition, the overall responsiveness of the cooling system felt balanced and perfectly tuned to match the heat output of the Nvidia RTX 2080 Ti graphics card.
We thought they missed the mark with the placement of the thermal probe, though. The position and depth simply don’t provide enough real-time temperature measurement capabilities. In our full-pot test, the temperature probe reading was not a problem. However, If the test was conducted with warmer temperatures, such as -120 °C, then the thermal probe reading would certainly not be ideal.
With a retail price of just $220.50, the Bitspower VGA LN2 is the most cost-effective GPU cooling option currently available. The other competitors in this market are the EKWB EK-SF3D Inflection Point EVO for $285.95 and the Kingpin Cooling TEK-9 ICON 3.0 for $309.00. Regardless of price and market availability, we feel that Bitspower’s introduction into the extreme cooling category is a resounding success.
David Miller – mllrkllr88