Rack server system, direct chip cooling system and rack server cooling system

Abstract

A rack server system is disclosed, directly to a chip cooling system and a rack server cooling system. The rack server system includes a coolant distribution unit (CDU), a server housing containing a plurality of servers, a manifold coupled between the CDU and the servers, the manifold configured to distribute coolant flowing from the CDU to the servers, and a plurality of server line filters coupled between the manifold and the plurality of servers.

Description

[0001] Cross-references to related applications

[0002] This application claims the benefit of U.S. Provisional Application No. 63 / 647,551, filed May 14, 2024. The entire disclosure of the above-cited application is incorporated herein by reference. Technical Field

[0003] This disclosure relates to systems and methods for cooling computing systems, such as server computing systems like rack servers. This disclosure also relates to server-level filtration for secondary fluid networks that provide direct-to-chip cooling. Background Technology

[0004] Servers (e.g., server computing systems such as rack servers) typically include one or more types of cooling systems. Example cooling systems include air cooling systems, direct-to-chip (e.g., liquid-based) cooling systems, and combinations thereof. Cooling systems may also include filtration systems / devices, such as air filters, filter dryer assemblies, etc. Summary of the Invention

[0005] The rack server system includes a coolant distribution unit (CDU), a server housing containing multiple servers, a manifold coupled between the CDU and the servers, and multiple server line filters coupled between the manifold and the multiple servers, the manifold being configured to distribute coolant flowing from the CDU to the servers.

[0006] Among other features, each server line filter has a diameter of less than 50 mm and a length of less than 150 mm.

[0007] Among other features, the system also includes a system-level filtration component coupled between the manifold and the CDU.

[0008] Among other features, the system-level filtration component is the filter dryer component.

[0009] Among other features, the system also includes multiple server-level filtering components. Each server-level filtering component includes a corresponding server line filter among multiple server line filters and a quick-disconnect coupler disposed between the corresponding server line filter and the manifold.

[0010] Among other features, each server-level filtering component includes a bypass valve arranged in parallel with the corresponding server line filter.

[0011] Other areas of application of this disclosure will become apparent from the detailed description, claims, and accompanying drawings. The detailed description and specific examples are intended for illustrative purposes only and are not intended to limit the scope of this disclosure. Attached Figure Description

[0012] This disclosure will be more fully understood in light of the detailed description and accompanying drawings, in which:

[0013] Figure 1A , Figure 1B and Figure 1C An example rack server system according to this disclosure is shown, which includes multiple servers and a coolant distribution unit;

[0014] Figure 2 A functional block diagram of an example rack server system according to this disclosure is shown;

[0015] Figure 3 An example server-level filtering system based on this disclosure is shown; and

[0016] Figure 4 An example of quickly disconnecting the coupler is shown according to this disclosure.

[0017] In the accompanying drawings, reference numerals may be used repeatedly to identify similar and / or identical elements. Detailed Implementation

[0018] Servers (e.g., server computing systems such as rack servers) typically include one or more types of cooling systems. Example cooling systems include air cooling systems, direct-to-chip (e.g., liquid-based) cooling systems, and combinations thereof. Cooling systems may also include filtration systems / devices, such as air filters, filter dryer assemblies, etc.

[0019] Direct-to-chip cooling systems demand higher cleanliness (i.e., compared to air-cooled systems) and therefore require high-performance, flexible filtration systems. For example, contaminants as small as 50 microns can threaten the essential functionality of a direct-to-chip cooling system. If contaminants are introduced between the CDU and the server, system-level filtration (e.g., within the cooling or coolant distribution unit or the CDU) may fail to protect all components / equipment of the direct-to-chip cooling system (e.g., the evaporator). Additional risks of clogging (e.g., cold plate) are introduced when new equipment enters the secondary fluid network (IT equipment, instruments such as flow meters or pressure sensors, field modifications to piping, etc.). As an example, the liquid distribution components between the filter-dryer assembly and the cold plate evaporator are manufactured / assembled in situ. During commissioning, any residual particles must flow through the cold plate evaporator before being captured at the filter-dryer assembly. Therefore, larger particles may be captured at the cold plate evaporator and potentially clog some flow channels.

[0020] With the rapid development of direct-to-chip technologies (e.g., to accommodate complex processing technologies such as artificial intelligence and machine learning (AI / ML) data clusters), protecting server health and improving the reliability of cooling systems are crucial.

[0021] The systems and methods according to this disclosure are configured to implement one or more server-level filtration techniques. As used herein, "server-level" refers to a filter structure implemented downstream of a system-level filter structure (e.g., downstream of the filter dryer assembly of a CDU, between the CDU and a corresponding individual server, etc.). In contrast, "system-level" refers to a filter structure implemented at the CDU (e.g., a filter dryer such as a CDU configured to provide filtration for multiple servers). The addition of server-level filtration increases redundancy and thus increases confidence in the cleanliness of the refrigerant entering each server, accelerator, etc. For example, filtration implemented immediately at the inlet of each server reduces the likelihood of damage to components such as accelerators (e.g., processing accelerators). In some examples, the implementation of server-level filtration according to the principles of this disclosure can reduce the overall size of the CDU, for example, by eliminating the need for system-level filter filtration. In other examples, server-level filtration is implemented in addition to system-level filtration (e.g., in addition to the filter dryer assembly of the CDU).

[0022] In some examples, server-level filtering according to this disclosure includes a moisture capture medium.

[0023] In some examples, the systems and methods according to this disclosure are configured to monitor pressure drop across server-level filter components to measure the ratio of dirtiness to cleanliness and provide indications of a predictive schedule for filter replacement based on usage.

[0024] In some examples, server-level filtering systems and methods include one or more online filter components, couplers, and / or hose assemblies that can be quickly connected / disconnected, as described in more detail below.

[0025] Figure 1AAn example rack server system 100 is shown, comprising a plurality of servers 104 (shown in dashed lines) housed within a rack housing 108. A coolant distribution unit (CDU) 112 (which may be housed within or disposed outside the rack housing 108) is configured to control the flow of coolant to and from the servers 104 (e.g., via a corresponding coolant supply line or conduit via a manifold 116). Coolant (e.g., liquid refrigerant) flows from the manifold 116 into the servers 104 to absorb heat from a corresponding cooling plate or cold plate and returns from the servers 104 to the CDU 112. For example, the CDU 112 includes a heat exchanger configured to transfer / remove heat absorbed by the coolant.

[0026] The filter dryer assembly 120 is arranged between CDU 112 and server 104 (i.e., upstream of server 104). Figure 1A In the example shown, the filter dryer assembly 120 is arranged outside the CDU 112. Figure 1B In the example shown, multiple (e.g., two) filter dryer assemblies 120 are arranged within CDU 112. The filter dryer assemblies 120 are configured to filter particulate matter and other impurities or contaminants from the coolant, and to remove moisture (e.g., water) from the coolant. For example, the filter dryer assembly 120 includes a filter media configured to remove contaminants and a desiccant material configured to remove moisture. However, contaminants introduced downstream of the filter dryer assembly 120 (e.g., contaminants associated with components, pipes, etc., of manifold 116 that may be introduced after installation / assembly) may be included in the coolant initially flowing to server 104.

[0027] In some embodiments described in more detail below, the filtration portion of the filter dryer assembly 120 may be implemented as one or more in-line filter assemblies for a respective liquid supply line to each of the servers 104, the liquid supply line may include a bypass line to enable operation without interruption during filter replacement.

[0028] Figure 1C Another example implementation of a rack server system 100 populated with servers 104 is shown.

[0029] Figure 2A functional block diagram of an example rack server system 200 according to this disclosure is shown. Coolant flows from CDU 208 (e.g., via manifold 212) and is distributed to multiple servers 204. In some examples, as shown, coolant flows from CDU 208 through a system-level filtration device, such as a filter dryer assembly 216. The filter dryer assembly 216 may be located externally to (as shown) or internally to CDU 208, or in some examples, the filtration function may be implemented at the server level.

[0030] System 200 includes one or more server-level filtering devices, such as server filters or components (e.g., server line filters) 220. As shown, filter components 220 are arranged within a respective server 204. In other examples, one or more filter components 220 may be arranged externally to server 204 (e.g., at a corresponding inlet of server 204, between manifold 212 and server 204). As shown, each of servers 204 has a corresponding filter component 220. In other examples, a filter dryer component 220 may be coupled to two or more of servers 204.

[0031] Therefore, server filter assembly 220 provides filtration to remove contaminants introduced downstream of CDU 208 (e.g., contaminants present before the assembly of CDU 208 and downstream components of filter dryer assembly 216, such as contaminants within manifold 212 and related components).

[0032] Figure 3 An example server-level filtration assembly or system 300 according to this disclosure is shown. System / component 300 includes a filter assembly (server line filter) 304 coupled between a coolant supply line 308 (e.g., manifold 312) and an inlet supply hose 316 of server 320. As shown in this example, various components of system 300 (e.g., filter assembly 304) are located outside server 320 (e.g., outside server housing or enclosure 324). In other examples, one or more components, such as filter assembly 304, are located inside server 320.

[0033] Filter assembly 304 has a reduced size compared to typical filter assemblies. For example, filter assembly 304 may have a diameter of less than 50 mm, and in some examples less than or equal to 25 mm. The length of filter assembly 304 may be less than 150 mm, and in some examples less than or equal to 50 mm. Furthermore, filter assembly 304 may include a removable filter element / media configured to filter contaminants smaller than 50 micrometers. In some examples, the filter media is configured to filter contaminants smaller than 30 micrometers (e.g., 25 micrometers and larger). In some examples, the filter media is configured to filter contaminants as small as 20 micrometers. In various embodiments, the filter assembly may be composed of metal or metal frit and may include refrigerant-compatible gaskets (i.e., composed of materials such as PEEK, NYLON, PTFE, or neoprene). Filter assembly 304 is also configured to provide a flow rate of 1 to 4 liters per minute and has a pressure rating between 1.5 MPa and 2.0 MPa (e.g., 1.75 MPa).

[0034] In the example (as shown), the server-grade filter assembly 328 includes a filter assembly 304, one or more quick-disconnect connectors / couplers (such as couplers 332 and 334), a bypass valve (e.g., a two-way ball valve) 336, and corresponding supply line / hose portions 340 between the components. The bypass valve 336 can be opened and closed (e.g., manually) to facilitate cleaning / maintenance of the filter assembly 304 without interrupting the flow of coolant to the server 320. For example, the bypass valve 336 can be normally closed and selectively opened before disconnecting the filter assembly 304 for cleaning (e.g., to replace the filter media). In some examples, the quick-disconnect coupler 344 facilitates the complete removal of the filter assembly 304 from the system 300. Figure 4 An example server 320 is shown, which has one or more fast disconnect couplers 334 configured to couple the server 320 to a server-level filter component 328.

[0035] In some implementations, couplers 332, 334, and 344 can be used to provide quick and easy connection and disconnection of fluid or air lines. These fittings, also known as quick-connect or quick-release couplers, are used to replace mating connections that would otherwise require tools for assembly and disassembly. Such fittings may include a spring-loaded ball latch mechanism that automatically locks the two halves of the fitting together when they are pushed together. To release the connection, the sleeve is pulled back and the ball is unloaded. The advantage of quick disconnection is that it can be done quickly by hand. Another example utilizes a manual tightening to engage a spring-loaded seal, with a male / female threaded end that is finally tightened by a wrench.

[0036] In some examples, filter assembly 328 includes one or more pressure ports 346, 348. Pressure ports 346, 348 can be used to monitor coolant pressure and / or flow rate. For example, the pressure drop between ports 346, 348 can be used to determine whether the filter media needs cleaning or replacement. Furthermore, the pressure drop can be used to determine optimal filter media, minimize subcooling, optimize heat transfer, and contribute to stable coolant flow, etc. In some examples, system 300 can be configured to provide notification (e.g., alarms or other visual or audio indications) in response to a determination that cleaning or other maintenance is required.

[0037] Although the above description pertains to pumped two-phase (refrigerant) systems, the principles of this disclosure can also be implemented in pumped single-phase systems or other direct-to-chip cooling systems. In an example single-phase system, the coolant is water-based, such as a 25% propylene glycol solution. One or more filter assemblies can be arranged at the CDU, and the piping / piping from the CDU to the server is manufactured and installed in situ. The single-phase heat exchanger deployed at the CPU / GPU has a similar geometry to that used in two-phase systems, such as a 100-micron fin geometry. Startup / commissioning filtration in a single-phase system presents the same challenges as in a pumped two-phase system (i.e., contaminants introduced from the piping / piping downstream of the filter assembly). Therefore, a server-grade filter assembly according to this disclosure can be implemented in a single-phase system.

[0038] The foregoing description is illustrative in nature and is in no way intended to limit the scope of this disclosure, its application, or its uses. The broad teachings of this disclosure can be implemented in various forms. Therefore, while this disclosure includes specific examples, its true scope should not be so limited, as other modifications will become apparent upon examination of the drawings, specification, and appended claims. It should be understood that one or more steps within the method may be performed in a different order (or simultaneously) without altering the principles of this disclosure. Furthermore, although each embodiment is described above as having certain features, any one or more of those features described with respect to any embodiment of this disclosure may be implemented in any other embodiment and / or combined with features of any other embodiment, even if such combination is not explicitly described. In other words, the described embodiments are not mutually exclusive, but substitution of one or more embodiments for each other remains within the scope of this disclosure.

[0039] Spatial and functional relationships between components (e.g., between modules, circuit elements, semiconductor layers, etc.) are described using various terms, including “connection,” “joint,” “coupled,” “adjacent,” “closely adjacent,” “on top,” “above,” “below,” and “set.” Unless explicitly described as “direct,” when describing the relationship between a first component and a second component in the above disclosure, the relationship can be a direct relationship where no other intermediate components exist between the first and second components, or it can be an indirect relationship where one or more intermediate components exist between the first and second components (spatially or functionally). As used herein, at least one of the phrases A, B, and C should be interpreted as meaning the use of a non-exclusive OR logic (A or B or C) and should not be interpreted as meaning “at least one of A, at least one of B, and at least one of C.”

[0040] In the accompanying drawings, as indicated by arrows, the direction of the arrows typically shows the flow of information (such as data or instructions) of interest to the illustration. For example, when components A and B exchange various types of information, but the information transmitted from component A to component B is relevant to the illustration, the arrow can point from component A to component B. This unidirectional arrow does not imply that no other information is transmitted from component B to component A. Furthermore, for information sent from component A to component B, component B may send a request for the information or an acknowledgment of receipt of the information to component A.

[0041] The apparatus and methods described in this application can be implemented, in part or in whole, by a dedicated computer created by configuring a general-purpose computer to perform one or more specific functions embodied in a computer program. The aforementioned function blocks, flowchart components, and other elements serve as software specifications that can be routinely compiled into a computer program by an experienced technician or programmer.

Claims

1. A rack server system, comprising: Coolant distribution unit; A server enclosure containing multiple servers; A manifold, coupled between the coolant distribution unit and the server, is configured to distribute coolant flowing from the coolant distribution unit to the server; as well as Multiple server line filters are coupled between the manifold and the multiple servers.

2. The rack server system according to claim 1, wherein, Each of the server line filters has a diameter of less than 50mm and a length of less than 150mm.

3. The rack server system of claim 1 further includes a system-level filtration assembly coupled between the manifold and the coolant distribution unit.

4. The rack server system according to claim 3, wherein, The system-level filtration component is a filter dryer component.

5. The rack server system according to claim 4, wherein, The plurality of server line filters are located downstream of the filter dryer assembly and between the filter dryer assembly and the plurality of servers.

6. The rack server system according to claim 1 further includes multiple server-level filtering components, wherein, Each of the server-level filtering components includes: The corresponding server line filter among the plurality of server line filters; and A quick-disconnect coupler is disposed between the respective server line filter and the manifold.

7. The rack server system according to claim 6, wherein, Each of the server-level filtering components includes a bypass valve arranged in parallel with the corresponding server line filter.

8. The rack server system according to claim 1, wherein, The multiple server line filters include moisture capture media.

9. The rack server system according to claim 1 further includes at least one pressure port.

10. The rack server system according to claim 9, wherein, The pressure port is configured to provide an indication of pressure drop.

11. The rack server system according to claim 1, further comprising multiple server-level filtering components, wherein, Each of the server-level filtering components includes: A corresponding server line filter, the corresponding server line filter being coupled between the inlet supply hose and the coolant supply line of the corresponding server among the plurality of servers; and A bypass valve is arranged in parallel with the corresponding server line filter between the coolant supply line and the inlet supply hose.

12. The rack server system according to claim 11, wherein, The corresponding server line filter is arranged on the outside of the server casing of the corresponding server.

13. A direct-to-chip cooling system, comprising the rack server system of claim 1.

14. A rack server cooling system that goes directly to the chip, comprising: Coolant distribution unit; A server enclosure containing multiple servers; A manifold coupled between the coolant distribution unit and the servers, the manifold being configured to distribute coolant flowing from the coolant distribution unit to the plurality of servers; as well as Multiple server-level filtering components are coupled between the manifold and the multiple servers. Each of the plurality of server-level filtering components includes: Server line filter, the server line filter being coupled between the coolant supply line of the manifold and the inlet supply hose of a respective server among the plurality of servers; and A bypass valve is arranged in parallel with the server line filter between the coolant supply line and the inlet supply hose.

15. The rack server cooling system according to claim 14, wherein, The server line filter is arranged on the outside of the server casing of the respective server.

16. The rack server cooling system of claim 14, further comprising a system-level filtration assembly coupled between the manifold and the coolant distribution unit, wherein, The plurality of server-level filtering components are located downstream of the system-level filtering component and between the system-level filtering component and the plurality of servers.

17. The rack server cooling system according to claim 14, wherein, Each of the plurality of server-level filtering components includes: A first quick-disconnect coupler is disposed between the coolant supply line and the server line filter; and A second quick-disconnect coupler is disposed between the server line filter and the inlet supply hose.

18. The rack server cooling system of claim 17, further comprising a third quick-disconnect coupler disposed between the second quick-disconnect coupler and the inlet supply hose.

19. The rack server cooling system of claim 14, further comprising a first pressure port disposed between the coolant supply line and the server line filter, and a second pressure port disposed between the server line filter and the inlet supply hose.

20. The rack server cooling system according to claim 14, wherein, The server line filter has a diameter of less than 50 mm and a length of less than 150 mm.

Images