Immersion cooling system with various coolants

Abstract

The present application discloses an immersion cooling system with multiple cooling liquids. An apparatus is described. The apparatus includes: a chamber containing one or more electronic devices, a first liquid, and a second liquid. The electronic device is immersed in the second liquid. The first liquid has a density less than the density of the second liquid, so that the first liquid floats above the second liquid. The first liquid is used to return second liquid molecules received from the second liquid to the second liquid. The chamber includes a first fluidic channel, which is used to discharge the first liquid from the chamber when the second liquid is in the chamber.

Description

Background Art

[0001] With the advent of high-performance centralized computing (such as cloud computing, artificial intelligence, big data computing, etc.), the computing demands on the underlying electronic devices have caused the electronic devices to generate a lot of heat. As a result, engineers have focused on improving methods to remove heat from electronic devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0002] Figure 1 An immersion cooling system is shown;

[0003] Figure 2 An improved immersion cooling system is shown;

[0004] Figure 3 A more detailed embodiment of an improved immersion cooling system is shown;

[0005] Figure 4 A more detailed embodiment of an improved two-phase immersion cooling system is shown;

[0006] Figure 5 A data center is shown. DETAILED DESCRIPTION

[0007] Figure 1 Depicts the immersion cooling system. Figure 1 As observed in FIG. 1 , a plurality of electronic devices 101 are immersed in a dielectric liquid 102 that electrically isolates the exposed electrical nodes of the electronic devices 101 from their corresponding electronic devices ( Figure 1 A side view of an electronic device 101 oriented vertically within a liquid 102 is depicted. The electronic device generates heat when in operation, which is transferred to the liquid 102. The liquid 102 has a higher heat transfer coefficient than air, which enables heat to be removed from the electrical components more efficiently than would otherwise be achieved in an air-cooled environment. A filtration system 109 is coupled to the immersion bath chamber 103 to remove contaminants from the liquid 102.

[0008] The immersion bath chamber 103 is also fluidly coupled to a coolant distribution unit (CDU) 104 that includes a pump 105 and a heat exchanger 106. During continued operation of the electronic device 101, the temperature of the liquid will increase due to the heat it receives from the operating electronic device 101. The pump 105 draws the heated liquid 102 from the immersion bath chamber 103 to the heat exchanger 106. The heat exchanger 106 transfers heat from the heated fluid to a secondary liquid within a secondary cooling loop 107, which is fluidly coupled to a cooling tower and / or refrigeration unit 108. The removal of heat from the liquid 102 by the heat exchanger 106 reduces the temperature of the liquid, which is then returned to the chamber 103 as cooled liquid.

[0009] In a high computing environment such as a data center, respective CDUs of the multiple immersion bath chambers are coupled to a secondary loop 107, and cooling towers and / or refrigeration units 108 remove heat generated by the electronics within the multiple immersion bath chambers from the data center.

[0010] Preferred properties of the immersion bath liquid 102 (in addition to high specific heat and high electrical insulation) include a high boiling point and a corresponding low vapor pressure (the pressure exerted on the environment 110 above the liquid 102 by gas-phase molecules of the liquid 102 attempting to escape the liquid 102). However, a liquid that satisfies these criteria sufficiently still exhibits some persistent vapor pressure that results in a persistent loss of liquid material escaping the bath 103 and into the environment 110 in the vapor phase.

[0011] Unfortunately, the immersion bath chamber 103 is not hermetically sealed (not "airtight"). Here, several cables ( Figure 1 The wiring (not shown) passes through the sidewalls of the chamber 103 and / or the holes in the lid 111 of the chamber. Therefore, combined with the aforementioned continuous evaporation of the liquid 102, a large amount of liquid 102 material escapes the chamber 103 (e.g., through the aforementioned chamber holes) during the extended operation time of the operating electronic device 101.

[0012] Significant loss of immersion bath material is also exacerbated any time the cover 111 is opened (eg, for inserting / removing an electronic circuit board, and / or connecting / removing cables to / from an electronic circuit board).

[0013] The continuous loss of liquid material from the system requires regular replenishment of the liquid 102. Since the liquid 102 itself can be expensive, the continuous loss and replenishment of large amounts of liquid material results in higher costs for the operation of the immersion cooling system. In addition, some of the more efficient heat transfer immersion cooling liquids are not environmentally friendly (e.g., have high global warming potential (GWP) and / or include per- and polyfluoroalkyl (PFAS) chemicals). Thus, the continuous loss of liquid material from the system may not only increase the monetary cost of the system, but may also cause environmental problems.

[0014] As in Figure 2As observed in , the solution is to introduce a "cap layer" 213 of liquid (upper liquid) within the chamber 203 along with the nominal dielectric immersion cooling liquid 202 (lower liquid). The upper liquid 213 should: 1) be composed of molecules that (e.g., strongly) tend to bind to each other rather than to molecules of the lower liquid 202; 2) have a lower density than the density of the lower liquid 202; and 3) be environmentally friendly (e.g., be PFAS-free so that it exhibits a low GWP). It is worth noting that the first two properties above will make the upper liquid 213 immiscible or nearly immiscible with the lower liquid 202 (they do not mix or mix very little) in the case where the upper liquid 213 floats on top of the lower liquid 202.

[0015] During operation, the upper liquid 213 substantially prevents lower liquid 202 molecules from reaching the environment 210. Here, based on nominal evaporation when the temperature of a liquid is below its boiling point, collisions of lower liquid molecules 202 at the interface of the lower liquid 202 and the upper liquid 213 may have sufficient energy to escape the lower liquid 202 and enter the upper liquid 213. However, the relative immiscibility between the lower liquid 202 molecules and the upper liquid 213 molecules, and / or the higher density of the lower liquid 202 molecules as compared to the upper liquid 213 molecules, will cause any such high-energy escaping lower liquid 202 molecules that reach the upper liquid 213 to be "pushed" from the upper liquid 213 back into the lower liquid 202, thereby preventing the lower liquid molecules from reaching the environment 210.

[0016] Thus, the above method allows for the use of a lower liquid 202 that exhibits a higher heat removal capacity, contains PFAS, and / or has a greater GWP risk than other more commonly used immersion bath coolants, for example because other more commonly used immersion bath coolants do not contain PFAS and / or have a lower GWP risk. Using a lower liquid 202 that has a higher heat removal capacity than more commonly used immersion cooling fluids improves the heat removal efficiency of the overall system.

[0017] In various embodiments, at least for liquids with lower density (lightweight), the upper liquid 213 also has a high boiling point and exhibits a low vapor pressure, such that the upper liquid 213 remains in the liquid phase during operation and pushes any lower liquid 202 molecules that escape the lower liquid 202 back into the lower liquid 202. Alternatively or in combination, in various embodiments, the upper liquid 213 does not contain PFAS and / or exhibits a very low GWP risk (at least a much lower GWP risk than the lower liquid 202), such that any evaporation of the upper liquid 213 causes only minimal GWP risk, if any.

[0018] For ease of explanation, Figure 2 Not shown Figure 1The filtration system 109 and the inlet / outlet for cold / hot fluids entering / exiting the system are also not shown as in Figure 1 The external CDU and cooling tower / refrigeration unit are observed in the Figure 2 chamber integration to achieve a complete working system.

[0019] Figure 3 An embodiment of an immersion chamber system designed to contain two different liquids 313, 302 as described above is shown. As described immediately below, Figure 3 The particular immersion chamber 303 in is designed to: 1) accommodate the removal of upper liquid 313 while retaining lower liquid 302 within the chamber 303 (e.g., to serve electronic devices 301 within the lower liquid 302); 2) accommodate thermal expansion of the lower liquid 302; and, 3) accommodate the flow 321 of heated lower liquid 302 from the chamber 303 toward the CDU and the return of cooled lower liquid 302 back into the chamber 303.

[0020] Regarding 1) above, if Figure 3 As observed in FIG. 3 , the upper liquid 313 has its own dedicated reservoir 331 , pump 332 and corresponding fluidic channels so that the upper liquid 313 can be discharged from the chamber 303 while maintaining the lower liquid 302 in the chamber.

[0021] It is desirable to drain the upper liquid 313 from the chamber 303 before servicing the electronic device 301 in the lower liquid 302 because the upper liquid 313 may have a higher viscosity and / or a much lower vapor pressure, so it may take longer for the liquid to drip away from the server before the board is allowed to be serviced. In contrast, the lower liquid has a higher vapor pressure (e.g., >0.8 kPa at 25° C.) and will evaporate quickly, leaving little or no visible residue on the electronic device 301.

[0022] To remove the upper liquid 313, the valve 333 is opened, which allows the upper liquid to drain into the reservoir 331. In various embodiments, the volume of the reservoir 331 is greater than the maximum volume of the upper liquid 313, so that the upper liquid 313 can be completely drained from the chamber 303 without overflowing the reservoir 331. The fluidic passage between the upper liquid 313 and the reservoir 331 also includes a baffle 334 within the chamber 303, which physically separates the upper liquid 313 from the lower liquid 302. The opening in the chamber wall through which the upper liquid flows when being drained is placed toward the bottom of the fluidic passage formed by the baffle 334 within the chamber 303, so that the upper liquid 313 is naturally drained from the chamber 303 due to gravity.

[0023] After the upper liquid 313 has been drained from the chamber 303 and is in the reservoir 331, the valve 333 can be closed. When the time is right to reintroduce the upper liquid 313 back into the chamber 303, for example, after the electronic device 301 has been serviced, the valve 333 remains closed, the valve 334 is opened and the pump 332 is activated. This creates an open passage from the reservoir 331 through the pump 332 to the spigot (controlled by the valve 334) extending into the chamber 303. The upper liquid 313 is then pumped back into the chamber 303 from the reservoir 331 through the open fluidic passage. If the fluid level in the reservoir 331 drops below a certain level (which is detected using the fluid level sensor 351, which determines the amount of upper liquid 313 in the chamber 303), the pump 332 can then be turned off and the valve 334 can be closed.

[0024] For example, during nominal operation of the electronic device 301 and in the absence of any need to service the electronic device 301, the above design also allows regular filtering of the upper liquid 313. In this case, the valves 333 and 334 are opened, and the pump 332 is activated, so that the upper liquid 313 is regularly circulated from the chamber 303 to the reservoir 331, and the upper liquid 313 is pumped back into the chamber 303 from the reservoir 331 by the pump 332. If the rate at which the upper liquid 313 is discharged from the chamber 303 is equal to the rate at which the upper liquid 313 is pumped back into the chamber 303, the amount of upper liquid in the chamber 303 will remain constant. This condition can be established by controlling the respective sizes of the openings in the valves 333, 334, the pumping rate of the pump 332, or some combination of these. In practice, if valves 334 and 333 are open and pump 332 is operating, the pump flow rate should be kept below the rate at which fluid 313 can drain back into reservoir 331 to ensure that the fluid level within reservoir 331 remains above some minimum desired level. If for some reason the rate of draining is slower, reservoir 331 will begin to empty and the fluid level in reservoir 331 will drop. A fluid level sensor 351 within reservoir 331 can detect the drop to cause pump 332 to shut off.

[0025] One or more particle filters (not in Figure 3 ) may also be placed anywhere along the above-described fluidic channel of the upper liquid to remove contaminants / particles from the upper liquid 313 as it circulates through the fluidic channel.

[0026] If contaminants exist together with the upper liquid 313, those contaminants having a lower density than the density of the upper liquid 313 will float (be able to float) on the surface of the upper liquid 313 in the reservoir 331, while those contaminants having a higher density than the density of the upper liquid 313 will sink to the bottom of the reservoir 332.

[0027] Thus, additional filtration systems 352, 353 (similar to Figure 1 A filter system 109 of the embodiment of the present invention can be integrated with the reservoir 331 to specifically remove these types of contaminants. Here, for example, the upper filter system 352 can draw from the upper surface of the fluid in the reservoir 331 (which contains floating particles) and pump the drawn fluid through the filter before returning the filtered fluid to the reservoir 331 (alternatively, the floating particles can be siphoned from the upper surface of the fluid in the reservoir 331). Similarly, the lower filter system 353 can draw fluid from the bottom of the reservoir 331 (which contains particles with a density greater than that of the upper liquid 313), pump the fluid through the filter, and return the filtered fluid to the reservoir 331.

[0028] Additionally, when the upper liquid 313 is drained from the chamber 303, it is contemplated that additional contaminants having a density greater than that of the upper liquid 313 but less than that of the lower liquid 302 will float on the top surface of the lower liquid 302 (such particles accumulate at the interface between the two liquids 313, 302 when both liquids 313, 302 are within the chamber 303). These particles can be easily siphoned from the top surface of the lower liquid 302 after the upper liquid 313 has been drained from the chamber 303 and into the reservoir 331.

[0029] Another valve 335 is positioned toward the bottom of the reservoir 331, which is opened when the upper liquid 313 is to be flushed from the system (eg, and replaced with new / fresh upper liquid).

[0030] Another set of fluidic channels is integrated with chamber 303 to accommodate thermal expansion of lower liquid 302 , as well as to accommodate the flow of heated lower liquid 302 from chamber 321 (towards the CDU) and return cooled lower liquid (from the CDU) to chamber 322 .

[0031] Here, another baffle 336 forms a fluidic passage to an escape port 337, and when the lower liquid 302 is receiving little / no heat from the electronic device 301 and is therefore in a state of minimal thermal expansion, the escape port 337 is placed above the upper surface of the lower liquid 302. However, as the electronic device increasingly transfers heat into the lower liquid 302, the lower liquid 302 will responsively expand, including rising toward the escape port 337 within the fluidic passage formed by the baffle 336.

[0032] The vertical position (height) of the escape port 337 establishes the maximum height to which the lower liquid 302 will rise due to thermal expansion within the chamber 303. Here, the narrower width of the fluidic passage formed by the baffle 336 will cause the expanding lower liquid 302 to rise to a higher level within the passage than within the chamber 303 (thermal expansion creates a greater upward force in a narrower passage).

[0033] Once the expanded lower liquid 302 reaches the escape port 337, the lower liquid 302 will begin to flow out of the channel and into the expansion sleeve 338, which in turn causes the height of the lower liquid 302 within the chamber 303 to remain approximately constant (any further volume increase of the lower liquid 302 due to thermal expansion is handled by the loss of an equal amount of lower liquid volume from the chamber 303). A bellows 339 (e.g., an elastic balloon) is mechanically coupled to the expansion sleeve 338 to accommodate conditions of extreme thermal expansion of the lower liquid 302, where the volume of fluid escaping from the chamber 303 exceeds the volume of the expansion sleeve 338.

[0034] It is worth noting that the lower liquid 302 maintains a constant height due to the flow of the expanded lower liquid 302 from the chamber 303, and the height of the baffle 334 can also be established for the fluidic channel system of the upper liquid. That is, the lower end of the escape port 337 only needs to extend to just above the designed maximum height of the lower liquid 302 in the main chamber 303. The baffle 336 extends below the designed maximum height of the lower liquid 302 to ensure that the upper liquid 313 is physically separated from the lower liquid 302 when the upper liquid 313 is being discharged from the chamber 303, and also to ensure that the upper liquid 313 cannot enter the expansion sleeve 338 and flow to the CDU 321. It is worth noting that, as shown in Figure 3 As observed in FIG, since the lower liquid 302 in the narrower fluidic channel formed by the baffle 336 will rise to a higher liquid level than the lower fluid 302 in the chamber 303, the height of the baffle 334 that forms the fluidic channel for the upper liquid 313 can be lower than the heights of the escape port 337 and the baffle 336.

[0035] In further embodiments, the height of the baffle 334 that forms the fluidic passage for the upper liquid 313 is adjustable so that, for example, during the discharge of the upper liquid 313 from the chamber 303, the height of the baffle 334 can be lowered so that (ideally) all of the upper liquid 313 is discharged from the chamber 303. In addition, if any of the lower liquid 302 passes over the baffle 334 and flows into the reservoir 331, the lower liquid 302 will sink to the bottom of the reservoir 331 due to the immiscibility of the two liquids 313, 302 with the lower liquid 302 having a greater density than the upper liquid 313. In addition, please note that any contaminant particles that are heavier than the upper liquid 313 and lighter than the lower liquid 302 will flow into the reservoir 331. When the pump 332 is activated, it will pump the upper fluid 313, the lower fluid 302 that flows into the reservoir 331, and pump the contaminants. A filter 354 positioned between the reservoir 331 and the pump 332 will capture these contaminant particles so that they can be removed from the system.

[0036] Although the above-described embodiments have emphasized that the electronic device 301 is retained within the lower liquid 302, a portion of the electronic device 301 may extend into the upper liquid 313 (for example, a backplane through which an input / output cable is coupled to a circuit board may be positioned within the upper liquid 313) because both the upper liquid 313 and the lower liquid 302 are dielectric fluids.

[0037] Figure 3 The system also includes a controller 340, which can perform / control any of the operations described above, such as opening / closing / adjusting any of the aforementioned valves 333, 334, 335 and / or controlling the pumping rate of the pump 332. In addition, the controller 340 can monitor conditions within the system to ensure that the system is operating correctly. For example, the controller 340 can monitor the quality of the above-mentioned filter to detect when filter replacement is appropriate. As another example, the controller 340 can monitor the fluid level in the reservoir 331 and determine when to shut off the pumping activity of the pump 332, etc., based on one or more sensors 351 within the reservoir 331. Moreover, a low fluid level in the expansion sleeve 338 (e.g., detected by a sensor 355 within the expansion sleeve volume) can be used by the controller 340 to cause a reduction in the CDU 321 pump flow rate and / or can be used as an indicator that the lower liquid 302 is running low, which causes the CDU to raise a sign that the lower liquid needs to be replenished.

[0038] The controller 340 may also measure the temperature of the lower liquid 302 (e.g., using one or more thermistors in the chamber) and / or may measure the power consumption of the electronics 301 (e.g., by monitoring the electronics workload and / or supply current draw), in response to which it may calculate an expected thermal expansion of the lower liquid 302, and then calculate an expected escape rate of the lower liquid 302 from the escape port 337 and / or a final level of the lower liquid within the expansion sleeve 338. The controller 340 may then compare the calculated expected escape rates and / or fluid levels to confirm that an expected amount of the lower liquid 302 is escaping from the chamber 303.

[0039] The controller 340 may also be coupled to sensors (not shown) in the chamber 303 that are positioned at respective heights along the walls of the chamber 303 to detect whether the lower liquid 302 drops below a minimum desired level for proper operation and / or rises above a maximum desired level for proper operation. If the minimum or maximum level is exceeded or about to be exceeded, the controller 340 may sound an alarm.

[0040] In various embodiments, the upper liquid 313 is composed of polyalphaolefins (PAO) (e.g., PAOx, where x is in the range of 2 to 8, including 2 and 8); refined mineral oil; gas-liquid fluids; isoalkanes; polyol esters; and silicone oils. Examples include Synfluid PAO2, PAO2.5, PAO3, PAO3.5, PAO4, PAO6, SpectraSyn fluid series, CompuZol, IM2015, Renolin FECC5 Synthetic, Castrol ON BOT 2151, Renolin FECC7, ENEOS fluids, BP DC-15, Shell 3447, DC Cooling BioLife 5, Palub ICE-5, and Dowsil ICL-1202. The upper liquid 313 can also meet the sustainability criteria of negligible PFAS and / or zero / low GWP. The upper liquid 313 should also exhibit minimal volatility (e.g., having a vapor pressure of less than 0.8 kPa at 25° C.). In various embodiments, the lower liquid 302 is composed of fluorochemical fluids (PFAS), perfluorocarbons, perfluoroolefins, perfluoropolyethers, fluoroketones, hydrofluorocarbons, hydrofluoroethers, and hydrofluoroolefins.

[0041] The lower liquid 302 may be composed of a PFAS fluid, a volatile fluid (e.g., exhibiting a vapor pressure greater than 0.8 kPa at 25°C), and / or may have excellent thermal and signal integrity capabilities, such as a dielectric constant ≤ 2.3 and a dielectric loss tangent ≤ 0.05 from 20 MHz to 20 GHz to 40 GHz in the range of 20°C to 70°C. Thus, the lower liquid 302 may be PFAS-containing and / or high GWP (such as ≥ 1). Some examples include Noah 3000A, 3M TM Fluorinert TM E-liquids (e.g., FC-40, FC-43), PFPE HT fluids (e.g., HT270) and 3M TM Novec TM Engineered fluids (e.g., Novec 7500).

[0042] Figure 4 Another similar chamber design is shown, which includes the above-mentioned Figure 3 Many of the features and operating principles of the described systems. Specifically, Figure 4 The system includes a lower density PFAS-free upper liquid 413 for preventing molecules from the higher density PFAS lower liquid 402 from reaching the environment 410. Therefore, a baffle 434, a reservoir 431, a pump 432, and corresponding (one or more) fluidic channels for the upper liquid 413 are included in Figure 4 in the system.

[0043] However, Figure 3 In contrast, the system uses a "single-phase" cooling method that cools primarily by convection. Figure 4 The chamber system of applies "two-phase" immersion cooling, where cooling from both boiling and convection is used to remove heat from the electronic device 401. Heat is removed from the system by circulating a lower liquid 402 in direct contact with the electronic device, whereby the liquid undergoes evaporation due to its lower boiling point to transfer heat away from the liquid.

[0044] In the case of two-phase cooling, when the electronic device 401 is emitting a moderate amount of heat, the lower liquid 402 cools the electronic device 401 by convection cooling, as described above for single-phase cooling. However, when the electronic device 401 emits a greater amount of heat, the heat absorbed by the lower liquid 402 causes the temperature of the lower liquid 402 to reach its boiling point, and the lower liquid 402 responsively boils. Bubbles of the vaporized lower liquid 402 rise above the lower liquid 402 and into the upper liquid 413.

[0045] The first condenser 451 is placed in the upper liquid 413, and if necessary, also in a small portion of the top layer of the lower liquid 402, so that if any vapor accumulates below the upper liquid 413 layer, it effectively cools the vapor back into droplets that fall back into the lower liquid 402. In addition, the first condenser 451 also cools the upper liquid 413, and any heat received by the upper liquid 413 from the vapor is removed from the system by the first condenser 451. More specifically, the heat is transferred to the cooled water within the condenser (not shown), which is responsively heated and then pumped out of the condenser and chamber (not shown), cooled again by an external heat exchanger (not shown), and returned to the condenser 451 as a cooled liquid (not shown). The process is then continuously repeated so that heat is continuously removed from the upper liquid 413.

[0046] A second condenser 452 (which heats the water outlet and cools the water return, also not shown), operating according to the same principles as described above, is also placed within the environment 410 above the upper liquid 413. Here, if any of the vapor from the lower liquid 402 is not condensed by the condenser 451 within the upper liquid 413 and rises into the environment 410, the vapor will be cooled by the second condenser 452 into droplets that fall back into the upper liquid 413 and then further sink into the lower liquid 402. The second condenser 452 may also condense any vapor from the upper liquid (if any) back into droplets that fall back into the upper liquid 413.

[0047] Chilled plant water may first be sent through condenser 452, and then the same chilled water may be sent through condenser 451 to exchange heat with steam from lower fluid 402. As the chilled water picks up heat within condenser 451, it becomes hot and is pumped out of condenser 451 and chamber 403 for subsequent cooling.

[0048] Importantly, condensing the vapor of the lower liquid back into liquid as provided by the first condenser 451 and the second condenser 452 greatly reduces the amount of lower liquid material that will escape the system. Thus, operational costs and GWP risks are controlled.

[0049] It is worth noting that since excess heat is removed from the system by the first condenser 451 and the second condenser 452, Figure 4 The chamber 403 includes neither a heated fluid outlet nor a cooled fluid inlet. Figure 4 The system does not include an expansion sleeve, associated fluidic channels, an inlet for heated lower liquid to be sent to an external CDU, and a corresponding return for cooled lower liquid to be returned to the chamber, such as Figure 3If additional cooling of the lower liquid 302 by an external CDU is included, jackets, fluidic channels, inlets and returns may be added, provided these components and the external heat exchanger system are adequately sealed to prevent any vapor within the fluid from escaping.

[0050] However, Figure 4 The system includes a bellows 439 that is mechanically coupled to the environment 410 to handle any thermal expansion of the liquids 413, 402. Here, to the extent that either or both of the upper liquid 413 and the lower liquid 402 expand as the temperature increases, the height of the upper liquid 413 will increase, the environment 410 within the chamber 403 will contract, and the bellows 439 will expand to accommodate the increased resulting ambient pressure. If one or both of the liquids 413, 402 relax (reduce) their expansion, the opposite process occurs. If the vaporization rate of the (one or more) liquids 413, 402 exceeds the condensation rate of the condensers 451, 452, the bellows 439 may also expand, which may also result in increased ambient 410 pressure. Similarly, if the condensation rate exceeds the vaporization rate, the bellows 439 may contract.

[0051] It is noteworthy that in various embodiments, even if the upper liquid 413 is actively cooled by the first condenser 451, the upper liquid 413 remains above the lower liquid 402 because the upper liquid 413 has a lower density than the lower liquid 402 (even at its cooler temperature). The upper liquid 413 can have a boiling point that is inherently higher than the boiling point of the lower liquid 402, so that when the upper liquid 413 does not boil, the lower liquid 402 naturally boils, (there is a temperature range where, if the two liquids 413, 402 have the same temperature within the range, the lower liquid 402 boils, but the upper liquid 413 does not boil). Alternatively, if the upper liquid 413 has a boiling point that is comparable to (or even lower than) the boiling point of the lower liquid 402, the cooling action of the first condenser 451 can prevent the upper liquid 413 from boiling.

[0052] Additionally, chamber 403 may include one or more mechanical agitation devices (eg, transducers) 461 that "shake" the interface between upper liquid 413 and lower liquid 402 to remove vapor that may be trapped at the interface between the two liquids.

[0053] The controller 440 may perform the above Figure 3In addition, the controller 440 can cause (one or more) stirring devices to stir the upper / lower liquid interface, control the corresponding fluid flow through the corresponding condenser, monitor the evaporation and condensation rates, etc.

[0054] In various embodiments, the upper liquid 413 is composed of polyalphaolefin (PAO) (e.g., PAOx, where x is in the range of 2 to 8, including 2 and 8), for example, having a pressure of 800 kg / m 3 -900kg / m 3 Examples include Shell S3X, Shell S5X, Shell S5 LV, Electrocool EC-110, EC-120, SmartCoolant SC-003, Renolin, FECC5 Synthetic, Compuzol IM 2015, Castrol ON BOT 2100-DC20, Renolin FECC7, Castrol ON, BOT 2151-DC15. In various embodiments, the lower liquid 302 is composed of a liquid having a density greater than that of the upper liquid (e.g., greater than 1400 kg / m 3 Examples include 3M Novec 649, FC-3284, Solvay HT-55, Novec 7200, Novec 7500, Opteon DF-50, and Noah 2100A.

[0055] Figure 5 A new, emerging data center environment is shown in which "infrastructure" tasks are migrated from the traditional general-purpose "host" CPU (where application software programs are executed) to an infrastructure processing unit (IPU) or a data processing unit (DPU), any / all of which are referred to as IPUs hereinafter.

[0056] Network-based computer services (such as those provided by cloud services and / or large enterprise data centers) typically execute application software programs for remote clients. Here, the application software programs typically perform specific (e.g., "business") end functions (e.g., customer service, procurement, supply chain management, email, etc.). Remote clients call / use these applications through temporary network sessions / connections established between the client and the application by the data center. A recent trend is to strip the functionality of at least some of the applications into finer-grained atomic functions ("microservices"), which are called by client programs as needed. Microservices are typically designed to charge clients / customers based on their actual use (function call invocation) of the microservice application.

[0057] However, in order to support the functionality of a network session and / or application, certain underlying computationally intensive and / or traffic intensive functions ("infrastructure" functions) are performed.

[0058] Examples of infrastructure functionality include routing layer functionality (e.g., IP routing), transport layer protocol functionality (e.g., TCP), encryption / decryption for secure network connections, compression / decompression for smaller footprint data storage and / or network communications, virtual networking between clients and applications and / or between applications, packet processing, ingress / egress queuing of networking traffic between clients and applications and / or between applications, ingress / egress queuing of command / response traffic between applications and mass storage devices, error checking (including checksum calculations to ensure data integrity), distributed computing remote memory access functionality, and the like.

[0059] Traditionally, these infrastructure functions have been performed by CPU units "below" their end-function applications. However, the intensity of infrastructure functions has begun to impact the ability of the CPU to execute its end-function applications in a timely manner relative to the expectations of customers and / or to execute its end-functions in an energy-efficient manner relative to the expectations of data center operators.

[0060] Therefore, as in Figure 5 As observed in , infrastructure functions are being migrated to the infrastructure processing unit (IPU) 507. Figure 5 An exemplary data center environment 500 is depicted that integrates an IPU 507 to migrate infrastructure functions from a host CPU 501 , as described above.

[0061] As in Figure 5As observed in FIG. 5 , an exemplary data center environment 500 includes a pool of CPU units 501 that executes final function application software programs 505 that are typically invoked by remote call clients. The data center also includes a separate memory pool 502 and a large storage pool 503 to assist in executing applications. The CPU pool 501, the memory pool 502, and the large storage pool 503 are coupled accordingly via one or more networks 504.

[0062] It is worth noting that each pool 501, 502, 503 has an IPU 507_1, 507_2, 507_3 on its front end or network side. Here, each IPU 507 performs pre-configured infrastructure functions on the inbound (request) packets it receives from the network 504 before delivering the request to the final function of its corresponding pool (e.g., executing application software in the case of CPU pool 501, running memory in the case of memory pool 502, and storing in the case of mass storage pool 503).

[0063] When the final function sends certain communications into the network 504, the IPU 507 performs pre-configured infrastructure functions on the outbound communications before transmitting the outbound communications into the network 504. Communication 512 between the IPU 507_1 and the CPUs in the CPU pool 501 can occur over a network (e.g., multi-hop Ethernet) and / or a more direct channel (e.g., point-to-point link) such as Compute Express Link (CXL), Advanced Extensible Interface (AXI), Open Coherent Accelerator Processor Interface (OpenCAPI), Gen-Z, etc.

[0064] Depending on the implementation, one or more CPU pools 501, memory pools 502, mass storage pools 503, and networks 504 may exist within a single chassis, for example, as a traditional rack-mounted computing system (e.g., a server computer). In a decomposed computing system implementation, one or more CPU pools 501, memory pools 502, and mass storage pools 503 are separate rack-mountable units (e.g., rack-mountable CPU units, rack-mountable memory units (M), rack-mountable mass storage units (S)).

[0065] In various embodiments, the software platform on which the application 505 is executed includes a virtual machine monitor (VMM) or hypervisor that instantiates multiple virtual machines (VMs). The operating system (OS) instance is executed on the VM accordingly, and the application is executed on the OS instance. Alternatively or in combination, the container engine (e.g., the Kubernetes container engine) is executed on the OS instance accordingly. The container engine provides a virtualized OS instance, and the container is executed on the virtualized OS instance accordingly. The container provides an isolated execution environment for the application suite, which may include applications for microservices.

[0066] It is worth noting that the corresponding electronic boards / components of the above data center components can be Figure 2 , Figure 3 and Figure 4 Described teaching to cool.

[0067] Embodiments of the present invention may include each process as set forth above. These processes may be embodied in program code (e.g., machine executable instructions). When the program code is processed, a general or special processor is used to execute the process of the program code. Alternatively, these processes may be performed by a specific / customized hardware component comprising a hard-wired interconnected logic circuit system (e.g., application specific integrated circuit (application specific integrated circuit, ASIC) logic circuit system) or a programmable logic circuit system (e.g., field programmable gate array (field programmable gate array, FPGA) logic circuit system, programmable logic device (programmable logic device, PLD) logic circuit system) for performing the process, or performed by any combination of program code and logic circuit system.

[0068] Element of the present invention can also be provided as machine-readable storage medium for storing program code.Machine-readable medium can include but is not limited to floppy disk, optical disk, CD-ROM and magneto-optical disk, flash memory, ROM, RAM, EPROM, EEPROM, magnetic card or optical card, or other types of medium / machine-readable medium suitable for storing electronic instructions.

[0069] In the foregoing description, the present invention has been described with reference to specific exemplary embodiments of the present invention. However, it will be apparent that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the present invention as set forth in the appended claims. Therefore, the description and drawings should be regarded as illustrative rather than restrictive.

Claims

1. A device comprising: A chamber, the chamber containing one or more electronic devices, a first liquid and a second liquid, the electronic device being immersed in the second liquid, the first liquid having a density less than that of the second liquid so that the first liquid floats above the second liquid, the first liquid being used to return second liquid molecules received from the second liquid to the second liquid, the chamber comprising a first fluidic channel, the first fluidic channel being used to discharge the first liquid from the chamber when the second liquid is in the chamber.

2. The device according to claim 1, wherein: The chamber includes a second fluidic passage for allowing the second liquid to escape the chamber in response to thermal expansion of the second liquid.

3. The device as claimed in claim 2, wherein: The second fluidic channel is coupled to a third fluidic channel for directing a heated second liquid to the CDU.

4. The device according to any one of claims 1, 2 or 3, wherein: The chamber includes a valve coupled to the first fluidic passage to control the discharge of the first liquid from the chamber.

5. The device of any one of claims 1, 2 or 3, further comprising a reservoir coupled to the first fluidic channel, the reservoir being used to collect the first liquid after the first liquid is drained from the chamber.

6. The device according to any one of claims 1, 2 or 3, wherein: The second liquid is a two-phase cooled immersion liquid.

7. The apparatus of claim 6, further comprising a condenser disposed within the first liquid.

8. The device of claim 1, wherein: The first liquid exhibits a lower vapor pressure than the second liquid.

9. A method comprising: Performing the following operations in a chamber containing one or more electronic devices, a first liquid, and a second liquid, wherein the electronic device is immersed in the second liquid, and wherein the first liquid has a density less than a density of the second liquid such that the first liquid floats above the second liquid: the first liquid returns second liquid molecules received by the first liquid from the second liquid to the second liquid; and, While the second liquid is retained in the chamber, the first liquid is discharged from the chamber.

10. The method of claim 9, wherein: The discharging of the first liquid includes discharging the first liquid through a fluid channel in the chamber.

11. The method of claim 9, further comprising: The discharged first liquid is filtered.

12. The method of claim 10, further comprising: The drained and filtered first liquid is pumped back into the chamber.

13. The method of claim 9, further comprising: A portion of the second liquid that has escaped the chamber in response to thermal expansion of the second liquid is directed to the CDU.

14. The method of claim 9, further comprising: Heat is removed from the first liquid using a first condenser positioned within the first liquid.

15. A data center, comprising: network; a CPU pool coupled to the network; a memory pool coupled to the network; a storage pool coupled to the network; an accelerator pool coupled to the network; as well as A chamber, wherein the chamber contains an electronic device of any one of the following items: the CPU pool, the memory pool, the storage pool and the accelerator pool, the chamber includes a first liquid and a second liquid, the electronic device is immersed in the second liquid, the density of the first liquid is less than the density of the second liquid, so that the first liquid floats above the second liquid, the first liquid is used to condense the vapor from the second liquid, and the chamber includes a first fluid channel, and the first fluid channel is used to discharge the first liquid from the chamber when the second liquid is in the chamber.

16. The data center of claim 15, wherein: The chamber includes a second fluidic passage for allowing the second liquid to escape the chamber in response to thermal expansion of the second liquid.

17. The data center of claim 16, wherein: The second fluidic channel is coupled to a third fluidic channel for directing a heated second liquid to the CDU.

18. The data center of claim 15, wherein: The chamber includes a valve coupled to the first fluidic passage to control the discharge of the first liquid from the chamber.

19. The data center of any one of claims 15, 16 or 17, further comprising a reservoir coupled to the first fluidic channel, the reservoir for collecting the first liquid after the first liquid is drained from the chamber.

20. The data center of any one of claims 15, 16 or 17, wherein: The second liquid is a two-phase cooled immersion liquid.

Images