Server and cabinet

Abstract

The invention discloses a server and a cabinet. The server comprises a case, a main liquid feeding pipe connected with a cooling liquid inlet, a main liquid return pipe connected with a cooling liquid outlet, a display card liquid cooling module, a mainboard liquid cooling module and a rear card liquid cooling module, wherein the display card liquid cooling module, the mainboard liquid cooling module and the rear card liquid cooling module are arranged in the case. A first panel of the case is provided with a cooling liquid inlet and a cooling liquid outlet, and the graphics card liquid cooling module can be inserted into and pulled out of the case through an opening of a second panel of the case. The display card liquid cooling module, the mainboard liquid cooling module and the rear card liquid cooling module can be connected in parallel to a main liquid feeding pipe and a main liquid return pipe, a liquid cooling inlet of the display card liquid cooling module is inserted into the liquid feeding end of the main liquid feeding pipe in a blind mode, and a liquid cooling outlet of the display card liquid cooling module is inserted into the liquid return end of the main liquid return pipe in a blind mode. According to the technical scheme, the server can be conveniently maintained, and normal operation of the server is guaranteed.

Description

Technical Field

[0001] This application relates to the technical field of servers, and particularly to a server and a cabinet. Background Art

[0002] At present, key components within a server can adopt a liquid cooling method for temperature reduction so that the server can operate normally. However, the liquid cooling pipelines within the server are complex, resulting in relatively high maintenance difficulty for the server. Summary of the Invention

[0003] The objective of the embodiments of this application is to provide a server and a cabinet to ensure the normal operation of the server while facilitating its maintenance.

[0004] In a first aspect, the embodiments of this application provide a server, including: a chassis, a main liquid supply pipe, a main liquid return pipe, and a graphics card liquid cooling module, a motherboard liquid cooling module, and a rear card liquid cooling module disposed within the chassis. The chassis has opposite first and second panels. The first panel has a coolant inlet and a coolant outlet. The graphics card liquid cooling module can be pluggable into and removable from the chassis through an opening in the second panel.

[0005] The main liquid supply pipe is connected to the coolant inlet and has a first liquid supply end, a second liquid supply end, and a third liquid supply end. The first liquid supply end is disposed close to the second panel. The liquid cooling inlet of the graphics card liquid cooling module is blindly plugged into the first liquid supply end. The liquid cooling inlet of the motherboard liquid cooling module is connected to the second liquid supply end. The liquid cooling inlet of the rear card liquid cooling module is connected to the third liquid supply end.

[0006] The main liquid return pipe is connected to the coolant outlet and has a first liquid return end, a second liquid return end, and a third liquid return end. The first liquid return end is disposed close to the second panel. The liquid cooling outlet of the graphics card liquid cooling module is blindly plugged into the first liquid return end. The liquid cooling outlet of the motherboard liquid cooling module is connected to the second liquid return end. The liquid cooling outlet of the rear card liquid cooling module is connected to the third liquid return end.

[0007] In the case of adopting the above technical solution, the coolant inlet of the first panel is connected to the main liquid supply pipe. The liquid cooling inlet of the graphics card liquid cooling module is blindly plugged into the first liquid supply end of the main liquid supply pipe. The liquid cooling inlet of the motherboard liquid cooling module is connected to the second liquid supply end of the main liquid supply pipe. The liquid cooling inlet of the rear card liquid cooling module is connected to the third liquid supply end. The first coolant outlet of the first panel is connected to the main liquid return pipe. The liquid cooling outlet of the graphics card liquid cooling module is blindly plugged into the liquid return end of the first main liquid return pipe. The liquid cooling outlet of the motherboard liquid cooling module is connected to the second liquid return end of the first main liquid return pipe. The liquid cooling outlet of the rear card liquid cooling module is connected to the third liquid return end of the first main liquid return pipe.

[0008] Therefore, the liquid cooling medium can be introduced into the main liquid supply pipe through the coolant inlet, so that the main liquid supply pipe can deliver the liquid cooling medium to the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module respectively, in order to use the liquid cooling medium to cool the graphics card module, the motherboard module, and the rear card module in the server. The liquid cooling medium that has absorbed heat in the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module can flow back to the main liquid return pipe and be sent out through the coolant outlet.

[0009] It can be seen that in the technical solution disclosed in the embodiment of the present application, the liquid cooling medium can cool the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module in a parallel manner. This cooling method can not only ensure that the cooling capacity of the liquid cooling medium entering the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module is as high as possible, so as to optimize the heat dissipation of the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module and reduce the power usage efficiency, but also reduce the pipeline pressure drop between the main liquid supply pipe and the main liquid return pipe, ensure the liquid cooling medium delivery efficiency, reduce pipeline losses, and lower the operation cost. And by using the liquid cooling method to dissipate heat from the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module, it is not necessary to set a fan in the server, thereby reducing the noise generated during the operation of the server.

[0010] Moreover, the graphics card liquid cooling module can be pluggable into the chassis through the opening of the second panel, so that the liquid cooling inlet of the graphics card liquid cooling module is blindly plugged into the first liquid supply end of the main liquid supply pipe, and the liquid cooling outlet of the graphics card liquid cooling module is blindly plugged into the first liquid return end. When the graphics card liquid cooling module needs to be independently maintained, the graphics card liquid cooling module can be pulled out through the opening of the second panel, so that the liquid cooling inlet of the graphics card liquid cooling module is quickly separated from the first liquid supply end of the main liquid supply pipe, and the liquid cooling outlet of the graphics card liquid cooling module is quickly separated from the first liquid return end of the main liquid return pipe. When the graphics card liquid cooling module needs to be installed into the chassis, the graphics card liquid cooling module can be inserted into the chassis through the opening of the second panel, and under the action of external force, the liquid cooling interface of the graphics card liquid cooling module can be blindly plugged into the first liquid supply end of the main liquid supply pipe, and the liquid cooling outlet of the graphics card liquid cooling module can be blindly plugged into the first liquid return end of the main liquid return pipe. It can be seen that the server in the embodiment of the present application can realize the independent pluggable maintenance of the graphics card liquid cooling module.

[0011] In a possible implementation, the server further includes a first liquid distributor and a second liquid distributor disposed inside the chassis. The total liquid supply pipe includes a first liquid supply pipe and a second liquid supply pipe, and the total liquid return pipe includes a first liquid return pipe and a second liquid return pipe. One end of the first liquid supply pipe is connected to the coolant inlet, and the other end is respectively connected to the first liquid supply end and the second liquid supply pipe through the first liquid distributor. The second liquid supply end and the third liquid supply end are connected to the second liquid supply pipe through the second liquid distributor. One end of the first liquid return pipe is connected to the coolant outlet, and the other end is respectively connected to the first liquid return end and the second liquid return pipe through the first liquid distributor. The second liquid return end and the third liquid return end are connected to the second liquid return pipe through the second liquid distributor.

[0012] The first liquid distributor in the embodiment of the present application can be used as a primary liquid distribution node to divide the liquid cooling medium conveyed by the first liquid supply pipe into two paths of primary liquid cooling media. One path of the primary liquid cooling medium enters the graphics card liquid cooling module through the first liquid supply end to dissipate heat from the graphics card liquid cooling module. The second liquid distributor can be used as a secondary liquid distribution node to divide the other path of the primary liquid cooling medium into two paths of secondary liquid cooling media through the second liquid distributor. One path of the secondary liquid cooling medium enters the motherboard liquid cooling module through the second liquid supply end to dissipate heat from the motherboard liquid cooling module, and the other path of the secondary liquid cooling medium enters the rear card liquid cooling module through the third liquid supply end to dissipate heat from the rear card liquid cooling module. It can be seen that the first liquid distributor can adjust the flow rate of the liquid cooling medium entering the graphics card liquid cooling module, as well as the flow rate of the liquid cooling medium entering the motherboard liquid cooling module and the rear card liquid cooling module, so as to meet the heat dissipation requirements of the graphics card liquid cooling module, the motherboard liquid cooling module and the rear card liquid cooling module as much as possible, and reduce the system pressure drop between the total liquid supply pipe and the total liquid return pipe to improve the operating efficiency of the server.

[0013] In a possible implementation, the first liquid supply pipe and the first liquid return pipe are arranged side by side, the second liquid supply pipe and the second liquid return pipe are arranged side by side, and the first liquid supply pipe and the second liquid supply pipe are stacked in the thickness direction of the chassis. At this time, the first liquid return pipe and the second liquid return pipe are also stacked in the thickness direction of the chassis. This stacked arrangement can reduce the occupation of the internal space of the chassis by the total liquid supply pipe and the total liquid return pipe, so as to improve the effective utilization rate and integration degree of the space inside the chassis by the graphics card liquid cooling module, the motherboard liquid cooling module and the rear card liquid cooling module.

[0014] In a possible implementation, the liquid cooling inlet of the graphics card liquid cooling module and the first liquid distributor are both close to the second panel. This can ensure that both the first liquid supply end and the first liquid return end are close to the second panel. On the other hand, when the graphics card liquid cooling module is inserted into the chassis through the opening of the second panel, the liquid cooling inlet of the graphics card liquid cooling module can be accurately blindly inserted into the first liquid supply end, and the liquid cooling outlet of the graphics card liquid cooling module can be accurately blindly inserted into the first liquid return end.

[0015] In a possible implementation, the rear card liquid cooling module is arranged close to the first panel, and the second liquid distributor is arranged close to the rear card liquid cooling module. At this time, the liquid cooling inlet of the rear card liquid cooling module and the third liquid supply end located in the second liquid distributor can be connected by a shorter pipeline. Similarly, the liquid cooling outlet of the rear card liquid cooling module and the third liquid return end located in the second liquid distributor can be connected by a shorter pipeline.

[0016] In a possible implementation, the graphics card liquid cooling module includes: a first liquid distribution device, a GPU chip cold plate, and a peripheral cold plate. The GPU chip cold plate and the peripheral cold plate are located on the side of the first liquid distribution device close to the first panel, and the first liquid distribution device has a first liquid cooling inlet and a first liquid cooling outlet.

[0017] The first liquid cooling inlet is blindly inserted into the first liquid supply end, and the first liquid cooling outlet is blindly inserted into the first liquid return end. The liquid cooling inlets of the GPU chip cold plate and the peripheral cold plate are both connected to the liquid cooling outlet of the first liquid distribution device, and the liquid cooling outlets of the GPU chip cold plate and the peripheral cold plate are both connected to the liquid cooling inlet of the first liquid distribution device. At this time, the GPU chip cold plate and the peripheral cold plate can be connected in parallel in the first liquid distribution device, so that the liquid cooling medium sent out from the first liquid supply end of the total liquid supply pipe can enter the GPU chip cold plate and the peripheral cold plate in parallel, so that the liquid cooling medium can exert its liquid cooling capacity as much as possible in the GPU chip cold plate and the peripheral cold plate, and ensure the heat dissipation requirements of the GPU chip cold plate and the peripheral cold plate.

[0018] Moreover, by introducing the liquid cooling medium into the GPU chip cold plate and the peripheral cold plate in parallel, the pressure drop at the liquid cooling inlet and the liquid cooling outlet of the first liquid distribution device can be fully reduced, the liquid cooling medium delivery efficiency can be improved, the pipeline loss can be reduced, and the operation cost can be lowered.

[0019] In a possible implementation, the peripheral cold plate includes a first peripheral component cold plate and a second peripheral component cold plate. The first liquid distribution device, the first peripheral component cold plate, and the GPU chip cold plate are distributed along the distribution direction from the second panel to the first panel, and the GPU chip cold plate is located in the area surrounded by the first peripheral component cold plate and the second peripheral component cold plate. At this time, the GPU chip cold plate does not need to occupy additional space in the chassis along the distribution direction from the second panel to the first panel, but is located in the space occupied by the peripheral cold plate in the chassis along the distribution direction from the second panel to the first panel. Therefore, the arrangement of the peripheral cold plate and the GPU chip cold plate can effectively reduce the space occupied by the graphics card liquid cooling module in the chassis and improve the integration of the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module.

[0020] In a possible implementation, the first peripheral component cold plate includes a first microchannel cold plate assembly provided on the first switching chip. The second peripheral component cold plate includes a first inserted tube cold plate provided on the first GPU power supply, a second inserted tube cold plate provided on the peripheral circuit, and a third inserted tube cold plate provided on the second GPU power supply. At this time, the first switching chip, the first GPU power supply, the peripheral circuit, and the second GPU power supply can all dissipate heat through liquid cooling. Moreover, compared with the power of the peripheral circuit, the power of the first GPU power supply, and the power of the second GPU power supply, the power of the first switching chip is relatively high. By setting the cold plate type corresponding to the first switching chip as a microchannel cold plate with relatively high heat dissipation capacity, the heat dissipation effect of the first switching chip can be ensured. The first CPU power supply, the second CPU power supply, and the peripheral circuit can select inserted tube cold plates with relatively weak liquid cooling capabilities for heat dissipation, which can effectively reduce the hardware cost.

[0021] In a possible implementation, the number of GPU chip cold plates is multiple, and the multiple GPU chip cold plates are connected in parallel to the first liquid distribution device, and multiple first peripheral component cold plates are connected in series with the first liquid distribution device, and multiple second peripheral cold plates are connected in series with the first liquid distribution device.

[0022] When multiple GPU chip cold plates are connected in parallel to the first liquid distribution device, the cooling capacity of the liquid cooling medium entering each GPU chip cold plate is relatively high, and each GPU chip in each GPU chip cold plate can be fully cooled, thereby improving the heat dissipation capacity of the GPU chips in each GPU chip cold plate and ensuring the stable operation of the server.

[0023] In a possible implementation, the liquid cooling inlet of the graphics card liquid cooling module is arranged close to the second panel, the rear card liquid cooling module is arranged close to the first panel, and the chassis has an upper end plate and a lower end plate distributed along the thickness direction of the chassis. The motherboard liquid cooling module is located on the surface of the graphics card liquid cooling module close to the upper end plate. This can reduce the unnecessary space occupation of the motherboard liquid cooling module and the graphics card liquid cooling module in the direction from the second panel to the first panel, and improve the integration degree of the motherboard liquid cooling module, the graphics card liquid cooling module, and the rear card liquid cooling module.

[0024] In a possible implementation, the main board liquid cooling module includes a hard disk cold plate, a main board peripheral cold plate, and a CPU chip cold plate. The hard disk cold plate, the main board peripheral cold plate, and the CPU chip cold plate are detachably connected in series between the second liquid supply end and the second liquid return end in sequence. Since the power consumption of the hard disk, the main board peripheral components, and the CPU chip increases in sequence, after the liquid cooling medium starts to dissipate heat from the hard disk, there is still enough cooling capacity left to dissipate heat from the subsequent main board peripheral components and the CPU chip, thereby reducing the impact of the cascading temperature rise on the cooling capacity of the liquid cooling medium. Moreover, the main board liquid cooling module adopts a series connection method to implement the liquid cooling solution, which can also ensure that the liquid cooling medium flow rates of the hard disk cold plate, the main board peripheral cold plate, and the CPU chip cold plate are the same, so as to more accurately control the liquid cooling medium flow rates of each hard disk cold plate, the main board peripheral cold plate, and the CPU chip cold plate.

[0025] In a possible implementation, partial areas of the hard disk cold plate, the main board peripheral cold plate, and the CPU chip cold plate are arranged on the surface of the graphics card liquid cooling module close to the upper end plate, and the hard disk cold plate, the hard disk cold plate, the main board peripheral cold plate, and the CPU chip cold plate are distributed along the distribution direction from the second panel to the first panel. This can ensure that the main board liquid cooling module can dissipate heat from the main board normally.

[0026] In a possible implementation, the main board peripheral cold plate includes a single-board chip cold plate provided for a single-board chip and a fourth embedded pipe cold plate provided for the CPU power supply. The single-board chip cold plate and the fourth embedded pipe cold plate are connected in series between the hard disk cold plate and the CPU chip cold plate in sequence. Compared with the energy consumption of the single-board chip, the energy consumption of the CPU power supply is relatively high. Connecting the single-board chip cold plate and the fourth embedded pipe cold plate in series between the hard disk cold plate and the CPU chip cold plate can ensure that there is still more cooling capacity for the CPU power supply to dissipate heat after the liquid cooling medium dissipates heat from the single-board chip, thereby further reducing the impact of the cascading temperature rise on the cooling capacity of the liquid cooling medium. Moreover, connecting the single-board chip cold plate and the fourth embedded pipe cold plate in series can make the flow rates through the single-board chip cold plate and the fourth embedded pipe cold plate the same, so as to accurately control the flow rates of the single-board chip cold plate and the fourth embedded pipe cold plate.

[0027] In a possible implementation, the rear card liquid cooling module includes a switch chip cold plate, a second liquid distribution device, and multiple rear card cold plates. The second liquid distribution device has a second liquid cooling inlet and a second liquid cooling outlet.

[0028] The third liquid supply end is connected to the liquid cooling inlet of the switch chip cold plate, the liquid cooling outlet of the switch chip cold plate is connected to the second liquid cooling inlet, multiple rear card cold plates are connected in parallel to the second liquid distribution device, and the second liquid cooling outlet is connected to the third liquid return end. Adopting this connection method, multiple rear card cold plates can dissipate heat from each rear card sufficiently, ensuring the heat dissipation effect of the rear cards in each rear card cold plate.

[0029] In a possible implementation, the second liquid distribution device is located above the cold plate of the switching chip, and the second liquid distribution device and multiple rear card cold plates are distributed along the distribution direction from the second panel to the first panel. At this time, the second liquid distribution device does not need to additionally occupy the area in the chassis along the distribution direction from the second panel to the first panel, thereby reducing the space of the rear card liquid cooling module in the chassis along the distribution direction from the second panel to the first panel, and improving the integration of the graphics card liquid cooling module, the motherboard liquid cooling module and the rear card liquid cooling module.

[0030] In a possible implementation manner, the multiple rear card cold plates include a first rear card cold plate and a second rear card cold plate arranged on the top of the first rear card cold plate. The plate surface of the first rear card cold plate is perpendicular to the upper end plate, and the plate surface of the second rear card cold plate is parallel to the upper end plate. At this time, the liquid cooling inlet of the first rear card cold plate can be blindly inserted into the third liquid cooling outlet of the second liquid distribution device, and the liquid cooling outlet of the first rear card cold plate can be blindly inserted into the third liquid cooling inlet of the second liquid distribution device.

[0031] In a second aspect, the present application further provides a cabinet, including a cabinet body and a server arranged in the cabinet body. The server is the server described in the first aspect or any possible implementation manner in the first aspect of the present application.

[0032] For the beneficial effects of the second aspect of the embodiments of the present application, refer to the beneficial effects of the air guiding device described in the first aspect or any possible implementation manner in the first aspect, and details are not described herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 FIG. shows a schematic structural diagram of an example of the cabinet according to the embodiments of the present application;

[0034] Figure 2 FIG. shows a schematic basic structural diagram of the server according to the embodiments of the present application;

[0035] Figure 3A FIG. shows a schematic structural diagram of an example of the server from a certain perspective according to the embodiments of the present application;

[0036] Figure 3B FIG. shows a schematic structural diagram of an example of the server from another perspective according to the embodiments of the present application;

[0037] Figure 4 FIG. shows a schematic assembly diagram of an example of the graphics card liquid cooling module, the motherboard liquid cooling module and the rear card liquid cooling module according to the embodiments of the present application;

[0038] Figure 5A FIG. shows a schematic connection structure diagram of an example of the total liquid supply pipe and the total liquid return pipe according to the embodiments of the present application;

[0039] Figure 5BShows a schematic diagram of the basic structure of the liquid cooling pipeline connection in the server according to an embodiment of the present application;

[0040] Figure 6A Shows a schematic diagram of an exemplary structure of the graphics card liquid cooling module according to an embodiment of the present application;

[0041] Figure 6B Shows a schematic diagram of the internal liquid cooling pipeline connection of the graphics card liquid cooling module according to an embodiment of the present application;

[0042] Figure 6C Shows a schematic diagram of an exemplary structure of the first liquid distribution device according to an embodiment of the present application;

[0043] Figure 6D Shows a schematic diagram of an exemplary structure of the first peripheral component cold plate according to an embodiment of the present application;

[0044] Figure 6E Shows a schematic diagram of an exemplary structure of the second peripheral component cold plate according to an embodiment of the present application;

[0045] Figure 6F Shows a schematic diagram of an exemplary structure of the GPU chip cold plate unit according to an embodiment of the present application;

[0046] Figure 7A Shows a schematic diagram of an exemplary structure of the motherboard liquid cooling module according to an embodiment of the present application;

[0047] Figure 7B Shows a schematic diagram of the internal liquid cooling pipeline connection of the motherboard liquid cooling module according to an embodiment of the present application;

[0048] Figure 7C Shows a schematic diagram of an exemplary structure of the hard disk cold plate according to an embodiment of the present application;

[0049] Figure 7D Shows a schematic diagram of an exemplary structure of the motherboard peripheral cold plate according to an embodiment of the present application;

[0050] Figure 7E Shows a schematic diagram of an exemplary structure of the CPU chip cold plate according to an embodiment of the present application;

[0051] Figure 8A Shows a schematic diagram of an exemplary structure of the rear card liquid cooling module according to an embodiment of the present application;

[0052] Figure 8B Shows a schematic diagram of the internal liquid cooling pipeline connection of the rear card liquid cooling module according to an embodiment of the present application;

[0053] Figure 8C Shows a schematic diagram of an exemplary structure of the switch chip cold plate according to an embodiment of the present application;

[0054] Figure 8D Shows a schematic structural diagram of an example of the second liquid separation device according to an embodiment of the present application;

[0055] Figure 8E Shows a schematic connection diagram of an example between the cold plate of the switching chip and the second liquid separation device according to an embodiment of the present application;

[0056] Figure 8F Shows a simplified connection diagram of the post - card liquid cooling module according to an embodiment of the present application. Detailed implementation manners

[0057] Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Instead, these embodiments are provided to more thoroughly and completely understand the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not used to limit the protection scope of the present application.

[0058] It should be understood that the various steps recorded in the method embodiments of the present application can be executed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this regard.

[0059] The term "including" and its variations used herein are open - ended, that is, "including but not limited to". The term "based on" is "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts such as "first" and "second" mentioned in the present application are only used to distinguish different devices, units or elements, and are not used to limit the order of the functions performed by these devices, units or elements or their interdependent relationships.

[0060] It should be noted that the modifications of "one" and "multiple" mentioned in the embodiments of the present application are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0061] The embodiments of the present application provide a cabinet, which may include a server cabinet, a network cabinet or a hybrid cabinet. But not limited thereto. Figure 1 Shows a schematic structural diagram of an example of the cabinet according to an embodiment of the present application. As Figure 1 shown, the cabinet 100 according to the embodiment of the present application may include a cabinet body 101 and a server 102 provided in the cabinet body 101.

[0062] In some alternative embodiments, as Figure 1 shown, in the cabinet 100 of the embodiments of the present application, other electronic devices may also be included, such as switches, adapters, etc. For example, a server cabinet may include computing devices such as servers, and a network cabinet may include network devices such as switches, fiber optic adapters or fiber optic distribution frames. A hybrid cabinet may include different types of servers. For example, a hybrid cabinet houses servers and switches.

[0063] In some alternative embodiments, as Figure 1 shown, the number of the servers 102 may be one or multiple, and the types of the multiple servers 102 may be the same or different. Taking a server cabinet as an example, the multiple servers 102 may be stacked along the height direction of the cabinet body 101. For example, a plurality of brackets may be arranged along the height direction of the inner wall of the cabinet body 101, and each bracket may support the corresponding server 102. It should be understood that Figure 1 only seven servers 102 are shown, but in practice, there may be fewer than seven servers 102, or there may be more than seven servers 102.

[0064] In some alternative embodiments, as Figure 1 shown, the cabinet 100 of the embodiments of the present application may further include a power supply system ( Figure 1 not shown in the figure). The power supply system includes a power conversion system, a cabinet management system, and a power supply busbar 103. The power conversion system can convert the alternating current or high-voltage direct current provided by the computer room into low-voltage direct current. The cabinet management system manages the power conversion system and the servers 102, and conducts data communication between the power conversion system and the servers 102. The power supply busbar supplies the direct current output by the power conversion system to the servers 102 at different heights, so as to supply power to each server 102 and enable each server 102 to work.

[0065] In some alternative embodiments, the cabinet 100 of the embodiments of the present application may further include a cabinet cooling unit ( Figure 1 not shown in the figure) provided in the cabinet body 101. The cabinet cooling unit may include a cooling fan and a heat dissipation window opened on the cabinet body 101 to cool the internal space of the cabinet.

[0066] In some alternative embodiments, as Figure 1 shown, the servers 102 accommodated in the cabinet 100 of the embodiments of the present application may be liquid-cooled servers, which have liquid-cooled cavities. The cabinet 100 may further include a liquid cooling system 104. The liquid cooling medium used by the liquid cooling system 104 may include water, alcohol-based solutions, fluorocarbon-based working fluids, mineral oils, silicone oils, etc., but is not limited thereto. The liquid cooling medium may be specifically selected according to the actual situation.

[0067] The liquid cooling system 104 includes an inlet pipeline 1041 and an outlet pipeline 1042 disposed within the cabinet 101. Both the inlet pipeline 1041 and the outlet pipeline 1042 communicate with the liquid cooling chamber of the server 102. Optionally, as Figure 1 shown, the inlet pipeline 1041 may include a first main pipeline 1041A and a plurality of first branch pipelines 1041B. The first main pipeline 1041A is connected to the liquid cooling chamber of the corresponding server 102 through each first branch pipeline 1041B. The outlet pipeline 1042 includes a second main pipeline 1042A and a plurality of second branch pipelines 1042B. The liquid cooling chamber of each server 102 can communicate with the second main pipeline 1042A through the corresponding second branch pipeline 1042B.

[0068] As Figure 1 shown, the liquid cooling medium can be sent into the cold plates of the respective servers 102 through the first main pipeline 1041A and the respective first branch pipelines 1041B to cool the respective servers 102 using the liquid cooling medium. After the liquid cooling medium absorbs heat in the cold plates, it can be sent back to the second main pipeline 1042A through the second branch pipeline 1042B. The second main pipeline 1042A can send the liquid cooling medium that has absorbed heat into an external cooling device for cooling, and then send it into the cold plates of the respective servers 102 through the first main pipeline 1041A and the respective first branch pipelines 1041B.

[0069] An embodiment of the present application provides a server that can achieve independent pluggable maintenance of a graphics card liquid cooling module. Figure 2 shows a schematic diagram of a basic structure of the server according to an embodiment of the present application, Figure 3A shows an exemplary structure diagram of the server according to an embodiment of the present application from one perspective, Figure 3B shows an exemplary structure diagram of the server according to an embodiment of the present application from another perspective. As Figure 2 、 Figure 3A and Figure 3B shown, the server 200 according to an embodiment of the present application may include: a chassis 201, a total liquid supply pipe S, a total liquid return pipe H, and a liquid cooling module disposed within the chassis 201. The liquid cooling module can be divided into three, namely a graphics card liquid cooling module 202, a motherboard liquid cooling module 203, and a rear card liquid cooling module 204.

[0070] As Figure 2 、 Figure 3A and Figure 3B shown, the chassis 201 has opposite first panel 201B and second panel 201F. The first panel 201B has a coolant inlet K1 and a coolant outlet K2. The coolant inlet K1 can be used as the liquid cooling chamber inlet of the server 200, and it is connected to Figure 1The first branch 1041B therein can be detachably connected in a manual insertion or blind insertion manner. The coolant outlet K2 can be used as the cooling chamber outlet of the server 200 and is connected to Figure 1 The second branch 1042B therein can be detachably connected in a manual insertion or blind insertion manner.

[0071] As Figure 2 , Figure 3A and Figure 3B shown, the total liquid supply pipe S is connected to the coolant inlet K1, and the total liquid return pipe H is connected to the coolant outlet K2. Optionally, the first panel 201B can be the rear panel of the chassis 201, and the second panel 201F can be the front panel of the chassis 201. However, it can also be that the first panel 201B is the front panel and the second panel 201F is the rear panel. Taking Figure 3A and Figure 3B shown, where the first panel 201B is the rear panel and the second panel 201F is the front panel as an example, the layout of the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 in the chassis 201 will be described in conjunction with the accompanying drawings.

[0072] Figure 4 shows an exemplary assembly schematic diagram of the graphics card liquid cooling module, the motherboard liquid cooling module, and the rear card liquid cooling module of the embodiments of the present application. As Figure 2 , Figure 3A and Figure 3B and Figure 4 shown, the graphics card liquid cooling module 202 is disposed close to the front panel (i.e., the second panel 201F), and the rear card liquid cooling module 204 is disposed close to the rear panel (i.e., the first panel 201F). At this time, the chassis 201 further has an upper end plate 201U and a lower end plate 201D distributed along the thickness direction of the chassis 201 ( Figure 4 the z direction in ), and the motherboard liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U. It should be understood that Figure 4 the coolant outlet K2 and the coolant inlet K1 in are only exemplary distributions, and in practice, the layout can also be combined with the actual environment.

[0073] As Figure 2 , Figure 3A and Figure 3B and Figure 4 shown, when the motherboard liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U, the unnecessary space occupation of the motherboard liquid cooling module 203 and the graphics card liquid cooling module 202 in the direction from the second panel 201F to the first panel 201B ( Figure 4 the x direction in ) can be reduced, and the integration degree of the motherboard liquid cooling module 203, the graphics card liquid cooling module 202, and the rear card liquid cooling module 204 can be improved.

[0074] Figure 5A shows a schematic diagram of an exemplary connection structure of the main liquid supply pipe and the main liquid return pipe according to an embodiment of the present application. As Figure 2 , Figure 4 and Figure 5A shown, the main liquid supply pipe S has a first liquid supply end G11, a second liquid supply end G12, and a third liquid supply end G13. The main liquid return pipe H is connected to the coolant outlet K2, and the main liquid return pipe H has a first liquid return end G21, a second liquid return end G22, and a third liquid return end G23. At this time, the liquid cooling inlet 202A of the graphics card liquid cooling module is connected to the first liquid supply end G11 of the main liquid supply pipe S, the liquid cooling outlet 202B of the graphics card liquid cooling module is connected to the first liquid return end G21, the liquid cooling inlet 203A of the motherboard liquid cooling module is connected to the second liquid supply end G12, the liquid cooling outlet 203B of the motherboard liquid cooling module is connected to the second liquid return end G22, the liquid cooling inlet 204A of the rear card liquid cooling module is connected to the third liquid supply end G13, and the liquid cooling outlet 204B of the rear card liquid cooling module is connected to the third liquid return end G23.

[0075] It can be seen that as Figure 2 , Figure 4 and Figure 5A shown, in the server 200 disclosed in the embodiment of the present application, the liquid cooling medium can cool the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 in a parallel manner. This cooling method can not only ensure that the cooling capacity of the liquid cooling medium entering the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 is as high as possible, so as to optimize the heat dissipation of the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 and reduce the power usage efficiency, but also reduce the pipeline pressure drop between the main liquid supply pipe S and the main liquid return pipe H, ensure the liquid cooling medium transportation efficiency, reduce pipeline losses, and reduce the operating cost. At the same time, by cooling the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 through the liquid cooling method, it is not necessary to set a fan in the server 200, thereby reducing the noise generated during the operation of the server 200. It has been proved by experiments that the heat dissipation power of the server 200 can reach the 10KW level product, and the power usage efficiency of the server 200 can be reduced to below 1.05; compared with the server using air-liquid mixing, its noise can be reduced by more than 30 dBA.

[0076] In an alternative manner, as Figure 2 , Figure 4 and Figure 5AAs shown, the liquid cooling inlet 202A of the graphics card liquid cooling module is detachably connected to the first liquid supply end G11 of the main liquid supply pipe S, the liquid cooling outlet 202B of the graphics card liquid cooling module is detachably connected to the first liquid return end G21, the liquid cooling inlet 203A of the motherboard liquid cooling module is detachably connected to the second liquid supply end G12, the liquid cooling outlet 203B of the motherboard liquid cooling module is detachably connected to the second liquid return end G22, the liquid cooling inlet 204A of the rear card liquid cooling module is detachably connected to the third liquid supply end G13, and the liquid cooling outlet 204B of the rear card liquid cooling module is detachably connected to the third liquid return end G23.

[0077] In the embodiments of the present application, as Figure 2 , Figure 4 and Figure 5A shown, the detachable connection between different components can be achieved by means of hand insertion or blind insertion, so that the graphics card liquid cooling module 202, the motherboard liquid cooling module 203 and the rear card liquid cooling module 204 can be independently maintained. For example, the liquid cooling inlet 202A of the graphics card liquid cooling module is blindly inserted into the first liquid supply end G11 of the main liquid supply pipe S, and the liquid cooling outlet 202B of the graphics card liquid cooling module is blindly inserted into the first liquid return end G21; the liquid cooling inlet 203A of the motherboard liquid cooling module is detachably connected to the second liquid supply end G12 of the main liquid supply pipe S by means of hand insertion, and the liquid cooling outlet 203B of the motherboard liquid cooling module is detachably connected to the second liquid return end G22 of the main liquid supply pipe S by means of hand insertion; the liquid cooling inlet 204A of the rear card liquid cooling module is detachably connected to the third liquid supply end G13 of the main liquid supply pipe S by means of hand insertion, and the liquid cooling outlet 204B of the rear card liquid cooling module is detachably connected to the third liquid return end G23 of the main liquid supply pipe S by means of hand insertion.

[0078] Optionally, as Figure 2 , Figure 3B , Figure 4 and Figure 5A shown, when the second panel 201F has an opening K, the graphics card liquid cooling module 202 can be pluggable into the chassis 201 through the opening K of the second panel 201F. Therefore, the graphics card liquid cooling module 202 is arranged close to the second panel 201F. At this time, the first liquid supply end G11 is arranged close to the second panel 201F, and the first liquid return end G21 is arranged close to the second panel 201F, so that the liquid cooling outlet 202B of the graphics card liquid cooling module can be conveniently blindly inserted into the first liquid return end G21, and the liquid cooling outlet 202B of the graphics card liquid cooling module can be conveniently blindly inserted into the first liquid return end G21.

[0079] As Figure 2 , Figure 3B , Figure 4 and Figure 5AAs shown, when the second panel 201F has an opening K, the graphics card liquid cooling module 202 can be pluggable into the chassis 201 through the opening K of the second panel 201F. And the liquid cooling inlet 202A of the graphics card liquid cooling module is blindly plugged into the first liquid supply end G11 of the main liquid supply pipe S, and the liquid cooling outlet 202B of the graphics card liquid cooling module is blindly plugged into the first liquid return end G21. Therefore, when it is necessary to independently maintain the graphics card liquid cooling module 202, the graphics card liquid cooling module 202 can be pulled out of the chassis 201 through the opening K of the second panel 201F, so that the liquid cooling inlet 202A of the graphics card liquid cooling module is quickly separated from the first liquid supply end G11 of the main liquid supply pipe S, and the liquid cooling outlet 202B of the graphics card liquid cooling module is quickly separated from the first liquid return end G21 of the main liquid return pipe H; when it is necessary to install the graphics card liquid cooling module 202 into the chassis 201, the graphics card liquid cooling module 202 can be inserted into the chassis 201 through the opening K of the second panel 201F, and under the action of an external force, the liquid cooling inlet 202A of the graphics card liquid cooling module can be blindly plugged into the first liquid supply end G11 of the main liquid supply pipe S, and the liquid cooling outlet 202B of the graphics card liquid cooling module can be blindly plugged into the first liquid return end G21 of the main liquid return pipe H. It can be seen that regardless of the complexity level of the liquid cooling pipeline of the graphics card liquid cooling module 202, independent pluggable maintenance can be achieved through the opening K of the second panel 201F.

[0080] In a possible implementation manner, Figure 5B A basic structural schematic diagram of the liquid cooling pipeline connection in the server according to the embodiment of the present application is shown. As Figure 2 、 Figure 4 、 Figure 5A and Figure 5BAs shown, the server 200 of the embodiment of the present application further includes a first liquid distributor 205A and a second liquid distributor 205B provided inside the chassis 201. The main liquid supply pipe S includes a first liquid supply pipe S1 and a second liquid supply pipe S2, such that one end of the first liquid supply pipe S1 is connected to the coolant inlet K1, and the other end is connected to the first liquid supply end G11 and the second liquid supply pipe S2 respectively through the first liquid distributor 205A. The second liquid supply end G12 and the third liquid supply end G13 are connected to the second liquid supply pipe S2 through the second liquid distributor 205B. At this time, the first liquid distributor 205A is a primary liquid distribution node, which divides the liquid cooling medium in the first liquid supply pipe S1 into two paths of primary liquid cooling media. The first path of primary liquid cooling medium enters the graphics card liquid cooling module 202 through the first liquid supply end G11 to dissipate heat from the graphics card liquid cooling module 202. The second path of primary liquid cooling medium enters the second liquid distributor 205B through the second liquid supply pipe S2. The second liquid distributor 205B can be used as a secondary liquid distribution node to divide the second path of primary liquid cooling medium into two paths of secondary liquid cooling media. The first path of secondary liquid cooling can enter the motherboard liquid cooling module 203 through the second liquid supply end G12 to dissipate heat from the motherboard liquid cooling module 203, and the second path of secondary branch liquid cooling can enter the rear card liquid cooling module 204 through the third liquid supply end G13 to dissipate heat from the rear card liquid cooling module 204.

[0081] As Figure 2 , Figure 4 , Figure 5A and Figure 5B shown, the main liquid return pipe H includes a first liquid return pipe H1 and a second liquid return pipe H2, such that one end of the first liquid return pipe H1 is connected to the coolant outlet K2, and the other end is connected to the first liquid return end G21 and the second liquid return pipe H2 respectively through the first liquid distributor 205A. The second liquid return end G22 and the third liquid supply end G13 are connected to the second liquid return pipe H2 through the second liquid distributor 205B. At this time, the liquid cooling media flowing out from the liquid cooling outlet 203B of the motherboard liquid cooling module and the liquid cooling outlet 204B of the rear card liquid cooling module can be converged into a branch liquid return medium through the second liquid distributor 205B and enter the first liquid distributor 205A through the second liquid return pipe H2. The liquid cooling medium flowing out from the liquid cooling outlet 202B of the graphics card liquid cooling module can also enter the first liquid distributor 205A. Therefore, the liquid cooling media flowing out from the liquid cooling outlet 202B of the graphics card liquid cooling module, the liquid cooling outlet 203B of the motherboard liquid cooling module, and the liquid cooling outlet 204B of the rear card liquid cooling module can finally converge at the first liquid distributor 205A to form the liquid return medium of the server 200, and finally be sent to the coolant outlet K2 through the first liquid return pipe H1, so that the liquid return medium of the server 200 is discharged from the server 200 through the coolant outlet K2.

[0082] It can be seen that as Figure 2 , Figure 4 , Figure 5A and Figure 5BAs shown in the figure, in the server 200 of the embodiment of the present application, the first liquid distributor 205A and the second liquid distributor 205B are used to perform two-stage liquid distribution on the liquid cooling medium. The graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 are connected in parallel between the total liquid supply pipe S and the total liquid return pipe H, so as to meet the heat dissipation requirements of the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 as much as possible, and reduce the system pressure drop between the total liquid supply pipe S and the total liquid return pipe H, so as to improve the operating efficiency of the server 200.

[0083] Optionally, as Figure 2 , Figure 4 , Figure 5A and Figure 5B shown, the flow rate and temperature of the liquid cooling medium entering the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 can be adjusted so that the pressure drops of the liquid cooling media of the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204 are equal.

[0084] For example, as Figure 2 , Figure 4 , Figure 5A and Figure 5B shown, a larger-sized coolant inlet K1 and coolant outlet K2 can be adopted. The first liquid supply pipe S1 and the first liquid return pipe H1 can be large-inner-diameter pipelines, and the first liquid return pipe H1 and the second liquid return pipe H2 can be small-inner-diameter pipelines, so as to realize the large-flow liquid cooling medium entering and leaving the server 200, thereby reducing the system pressure drop.

[0085] Optionally, as Figure 5B shown, the server 200 of the embodiment of the present application may further include a liquid inlet switch 206A and a liquid outlet switch 206B. The liquid inlet switch 206A can be arranged on the first liquid supply pipe S1, and the liquid outlet switch 206B can be arranged on the first liquid return pipe H1. In this way, in case of emergencies such as local liquid leakage, the liquid inlet switch 206A and the liquid outlet switch 206B can be closed.

[0086] Optionally, as Figure 2 , Figure 3A and Figure 3B shown, when the graphics card liquid cooling module 202 is close to the second panel 201F and the rear card liquid cooling module 204 is close to the first panel 201B, the liquid cooling inlet 202A of the graphics card liquid cooling module and the first liquid distributor 205A are both arranged close to the second panel 201F. And as Figure 2 , Figure 4 , Figure 5A and Figure 5BAs shown, the other end of the first liquid delivery pipe S1 is connected to the first liquid delivery end G11 through the first liquid distributor 205A, and the other end of the first liquid return pipe H1 is respectively connected to the first liquid return end G21 through the first liquid distributor 205A. Therefore, when both the liquid cooling inlet 202A of the graphics card liquid cooling module and the first liquid distributor 205A are close to the second panel 201F, on the one hand, it can ensure that both the first liquid delivery end G11 and the first liquid return end G21 are arranged close to the second panel 201F. On the other hand, during the process of inserting the graphics card liquid cooling module 202 into the chassis 201 through the opening K of the second panel 201F, the liquid cooling inlet 202A of the graphics card liquid cooling module can be accurately blindly inserted into the first liquid delivery end G11, and the liquid cooling outlet 202B of the graphics card liquid cooling module can be accurately blindly inserted into the first liquid return end G21.

[0087] For example, as Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 5A shown, when the graphics card liquid cooling module 202 needs to be installed into the chassis 201, it can be ensured that when the graphics card liquid cooling module 202 is basically inserted into the chassis 201 through the opening K of the second panel 201F, the liquid cooling inlet 202A of the graphics card liquid cooling module can start to be blindly inserted into the first liquid delivery end G11, and the liquid cooling outlet 202B of the graphics card liquid cooling module can start to be blindly inserted into the first liquid return end G21 until the liquid cooling inlet 202A of the graphics card liquid cooling module is successfully blindly inserted into the first liquid delivery end G11, and the liquid cooling outlet 202B of the graphics card liquid cooling module is successfully blindly inserted into the first liquid return end G21.

[0088] Optionally, as Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 5A shown, the rear card liquid cooling module 204 is arranged close to the first panel 201B, and the second liquid distributor 205B is arranged close to the rear card liquid cooling module 204. Since the third liquid delivery end G13 is connected to the second liquid delivery pipe S2 through the second liquid distributor 205B, and the third liquid return end G23 is connected to the second liquid return pipe H2 through the second liquid distributor 205B, both the third liquid delivery end G13 and the third liquid return end G23 are close to the first panel 201B. Therefore, a shorter pipeline can be used to connect the liquid cooling inlet 204A of the rear card liquid cooling module and the third liquid delivery end G13. Similarly, a shorter pipeline can be used to connect the liquid cooling outlet 204B of the rear card liquid cooling module and the third liquid return end G23. It can be seen that when the rear card liquid cooling module 204 is arranged close to the first panel 201B and the second liquid distributor 205B is arranged close to the rear card liquid cooling module 204, the connection between the rear card liquid cooling module 204 and the second liquid distributor 205B can be realized more conveniently.

[0089] Optionally, as Figure 2 ,Figure 4 and Figure 5A As shown, the first liquid delivery pipe S1 and the first liquid return pipe H1 are arranged side by side, and the second liquid delivery pipe S2 and the second liquid return pipe H2 are arranged side by side. The first liquid delivery pipe S1 and the second liquid delivery pipe S2 are arranged in the thickness direction of the chassis 201 (such as Figure 5A At this time, the first liquid return pipe H1 and the second liquid return pipe H2 are also stacked in the thickness direction of the chassis 201. This stacking arrangement can reduce the internal space occupied by the main liquid supply pipe S and the main liquid return pipe H in the chassis 201, so as to improve the effective utilization rate and integration of the graphics card liquid cooling module 202, the motherboard liquid cooling module 203 and the rear card liquid cooling module 204 in the chassis 201.

[0090] Optional, such as Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 5A As shown, the server 200 of the embodiment of the present application further includes an isolation plate (not shown in the figure) disposed in the chassis 201, the isolation plate may be close to the second panel 201F, the first liquid distributor 205A is located between the isolation plate and the lower end plate 201D of the chassis 201, and the second liquid distributor 205B is disposed on the surface of the isolation plate facing the upper end plate 201U of the chassis 201. At this time, the second liquid delivery pipe S2 may pass through the isolation plate from below the isolation plate and be connected to the second liquid distributor 205B above the isolation plate, and the second liquid return pipe H2 may pass through the isolation plate from above the isolation plate and be connected to the first liquid distributor 205A below the isolation plate.

[0091] like Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 5A As shown, when the motherboard liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U, the graphics card liquid cooling module 202 can be located between the isolation plate and the lower end plate 201D of the chassis 201, the motherboard liquid cooling module 203 can be located on the surface of the isolation plate facing the upper end plate 201U of the chassis 201, and the rear card liquid cooling module 204 can be located between the end of the isolation plate away from the second panel 201F and the first panel 201B.

[0092] In one possible implementation, Figure 6A FIG. 1 shows an exemplary structural diagram of a graphics card liquid cooling module according to an embodiment of the present application. Figure 2 and Figure 6AAs shown, the graphics card liquid cooling module 202 includes: a first liquid distribution device 2021, a peripheral cold plate 2022, and a GPU chip cold plate 2023. The GPU chip cold plate 2023 and the peripheral cold plate 2022 are located on the side of the first liquid distribution device 2021 close to the first panel 201B. At this time, when the graphics card liquid cooling module 202 is blindly inserted into the chassis 201 through the opening K of the second panel 201F, the first liquid distribution device 2021 is exposed through the opening K of the second panel 201F.

[0093] Optionally, as Figure 2 , Figure 3A , Figure 3B and Figure 6A shown, to facilitate the insertion and removal of the graphics card liquid cooling module 202, the server 200 further includes an insertion / removal assistance structure 207, which is provided on the side of the first liquid distribution device 2021 facing away from the first panel 201B. When inserting or removing the graphics card liquid cooling module 202, the graphics card liquid cooling module 202 can be removed from the chassis 201 through the insertion / removal assistance structure 207 or the graphics card liquid cooling module 202 can be inserted into the chassis 201 through the opening K of the second panel 201F through the insertion / removal assistance structure 207.

[0094] For example, as Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 6A shown, the insertion / removal assistance structure 207 may include a fixing frame 2071 and a fixing rod 2072 provided on the fixing frame 2071. The fixing frame 2071 is provided on the side of the first liquid distribution device 2021 facing away from the first panel 201B, and the fixing rod 2072 can be fixed to the second panel 201F of the chassis 201 to increase the stability of the graphics card liquid cooling module 202.

[0095] In one example, as Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 6A shown, the insertion / removal assistance structure 207 further includes a jack Q opened on the surface of the fixing frame 2071 facing away from the first panel 201B. When independent maintenance of the graphics card liquid cooling module 202 is required, the fixing rod 2072 can be separated from the second panel 201F of the chassis 201, and then the hand can be directly inserted into the jack Q of the fixing frame 2071 and the fixing frame 2071 can be pulled along the direction away from the first panel 201B, so that the fixing frame 2071 drives the graphics card liquid cooling module 202 to be pulled out of the chassis 201 through the opening K of the second panel 201F. When the graphics card liquid cooling module 202 needs to be installed into the chassis 201, the hand can be inserted into the jack Q, and then the graphics card liquid cooling module 202 can be pushed into the chassis 201 through the opening K of the second panel 201F.

[0096] As Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A and Figure 6A shown, the first liquid separation device 2021 has a first liquid cooling inlet and a first liquid cooling outlet. The first liquid cooling inlet can be blindly inserted into the liquid cooling inlet 202A of the graphics card liquid cooling module at the first liquid supply end G11, and the first liquid cooling outlet can be blindly inserted into the liquid cooling outlet 202B of the graphics card liquid cooling module at the first liquid return end G21. For example, when the server 200 includes the first liquid distributor 205A, the first liquid cooling inlet is blindly inserted into the first liquid supply end G11 located on the first liquid distributor 205A, and the first liquid cooling inlet is blindly inserted into the first liquid return end G21 located on the first liquid distributor 205A. At this time, the graphics card liquid cooling module 202 can be detached from the first liquid supply end G11 and the first liquid return end G21 at any time, so as to achieve independent maintenance.

[0097] Figure 6B shows a schematic diagram of the internal liquid cooling pipeline connection of the graphics card liquid cooling module according to an embodiment of the present application. Among them, Figure 6B the GPU chip cold plate 2023 and the peripheral cold plate 2022 shown in Figure 2 both show the objects they cool, and do not show other structures. As Figure 4 , Figure 6A and Figure 6B shown, the liquid cooling inlets of the GPU chip cold plate 2023 and the peripheral cold plate 2022 are both connected to the liquid cooling outlet of the first liquid separation device 2021, and the liquid cooling outlets of the GPU chip cold plate 2023 and the peripheral cold plate 2022 are both connected to the liquid cooling inlet of the first liquid separation device 2021. At this time, the GPU chip cold plate 2023 and the peripheral cold plate 2022 can be connected in parallel in the first liquid separation device 2021, so that the liquid cooling medium sent out from the first liquid supply end G11 of the total liquid supply pipe S can enter the GPU chip cold plate 2023 and the peripheral cold plate 2022 in parallel, so that the liquid cooling medium can exert the liquid cooling capacity as much as possible in the GPU chip cold plate 2023 and the peripheral cold plate 2022, and ensure the heat dissipation requirements of the GPU chip cold plate 2023 and the peripheral cold plate 2022.

[0098] Moreover, as Figure 6A and Figure 6B shown, by adopting a parallel method to introduce the liquid cooling medium into the GPU chip cold plate 2023 and the peripheral cold plate 2022, the pressure drop at the liquid cooling inlet of the first liquid separation device 2021 and the liquid cooling outlet of the first liquid separation device 2021 can be fully reduced. For example, at the designed flow rate, the pressure drop of the graphics card liquid cooling module 202 can be made as small as possible, so as to improve the liquid cooling medium transportation efficiency, reduce pipeline losses, and reduce the operation cost.

[0099] In an alternative manner, Figure 6C shows a schematic structural diagram of an example of the first liquid separation device according to an embodiment of the present application. As Figure 2 , Figure 4 and Figure 5A , Figure 6A - Figure 6C shown, the first liquid separation device 2021 according to an embodiment of the present application has a first liquid cooling channel 2021A and a second liquid cooling channel 2021B, and the first liquid cooling inlet and the first liquid cooling outlet are located at the same end of the first liquid cooling channel 2021A and the second liquid cooling channel 2021B.

[0100] For example, as Figure 2 , Figure 3A and Figure 3B shown, the chassis 201 further has a third panel 201R and a fourth panel 201L, and the distribution direction from the third panel 201R to the fourth panel 201L can refer to the y direction in Figure 2 , Figure 3A , Figure 3B and Figure 6A . For example, taking Figure 3A and Figure 3B as an example, when the first panel 201B is the rear panel and the second panel 201F is the front panel, the third panel 201R can be the right panel and the fourth panel 201L can be the left panel.

[0101] As Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A as well as Figure 6A - Figure 6C shown, when the first liquid separator 205A is close to the third panel 201R, both the first liquid cooling inlet and the first liquid cooling outlet are located at the end of the first liquid cooling channel 2021A and the second liquid cooling channel 2021B close to the third panel 201R, so that the first liquid cooling inlet can be normally blindly inserted into the first liquid supply end G11, and the first liquid cooling outlet can be normally blindly inserted into the first liquid return end G21.

[0102] Optionally, as Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A , Figure 6A and Figure 6C shown, the first liquid cooling channel 2021A and the second liquid cooling channel 2021B are distributed along the distribution direction from the lower end plate 201D to the upper end plate 201U (the z direction in Figure 6A and Figure 6C ), correspondingly, the first liquid cooling inlet and the first liquid cooling outlet are distributed along the distribution direction from the lower end plate 201D to the upper end plate 201U (the z direction in Figure 6A and Figure 6CThe distribution in the z - direction). Since the first liquid - cooling inlet is blindly inserted into the first liquid - supply end G11 and the first liquid - cooling outlet is blindly inserted into the first liquid - return end G21, the first liquid - supply end G11 and the first liquid - return end G21 are distributed along the distribution direction from the lower end - plate 201D to the upper end - plate 201U. In this way, not only can the space occupied by the first liquid - distribution device 2021 be reduced, but also it can be ensured that both the GPU chip cold - plate 2023 and the peripheral cold - plate 2022 can be connected to the first liquid - cooling channel 2021A and the second liquid - cooling channel 2021B.

[0103] Optionally, as Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A , Figure 6A and Figure 6C shown, both the first liquid - cooling channel 2021A and the second liquid - cooling channel 2021B extend along the distribution direction from the third panel 201R to the fourth panel 201L (refer to the y - direction in Figure 6A and Figure 6C ). At this time, the GPU chip cold - plate 2023 and the peripheral cold - plate 2022 can be distributed along the distribution direction from the third panel 201R to the fourth panel 201L. For example, the GPU chip cold - plate 2023 can be close to the third panel 201R and the peripheral cold - plate 2022 can be close to the fourth panel 201L; or, the GPU chip cold - plate 2023 can be close to the fourth panel 201L and the peripheral cold - plate 2022 can be close to the third panel 201R. In this case, the space occupied by the graphics - card liquid - cooling module 202 along the distribution direction from the second panel 201F to the first panel 201B (such as the x - direction in Figure 6A and Figure 6C ) can be reduced, and the integration degree of the graphics - card liquid - cooling module 202 can be improved.

[0104] As Figure 6A and Figure 6C shown, the liquid - cooling outlet of the first liquid - cooling channel 2021A is detachably connected to the liquid - cooling inlet of the GPU chip cold - plate 2023, and the liquid - cooling outlet of the GPU chip cold - plate 2023 is detachably connected to the liquid - cooling inlet of the second liquid - cooling channel 2021B. The liquid - cooling outlets of the first liquid - cooling channel 2021A and the liquid - cooling inlets of the second liquid - cooling channel 2021B can both adopt hand - inserted connectors, so that the liquid - cooling outlet of the second liquid - cooling channel 2021B can be detachably connected to the liquid - cooling inlet of the peripheral cold - plate 2022 by a hand - inserted method, and the liquid - cooling inlet of the peripheral cold - plate 2022 can be detachably connected to the second liquid - cooling inlet of the second liquid - cooling channel 2021B by a hand - inserted method.

[0105] Optionally, as Figure 2 , Figure 3A , Figure 3B , Figure 4 ,Figure 5A , Figure 6A and Figure 6C As shown in Figure 5A , Figure 6A , and Figure 6C , the liquid cooling outlet of the first liquid cooling channel 2021A can be located on the surface of the first liquid cooling channel 2021A close to the first panel 201B, and the liquid cooling inlet of the second liquid cooling channel 2021B can be located on the surface of the second liquid cooling inlet close to the first panel 201B. Both the GPU chip cold plate 2023 and the peripheral cold plate 2022 are located on one side of the first liquid distribution device 2021 close to the first panel 201B. Therefore, both the GPU chip cold plate 2023 and the peripheral cold plate 2022 can be conveniently connected between the liquid cooling outlet of the first liquid cooling channel 2021A and the liquid cooling inlet of the second liquid cooling channel 2021B.

[0106] In an alternative embodiment, as shown in Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A , Figure 6A , Figure 6C , the peripheral cold plate 2022 includes a first peripheral component cold plate 2022A and a second peripheral component cold plate 2022B. The first liquid distribution device 2021, the first peripheral component cold plate 2022A, and the GPU chip cold plate 2023 are distributed along the distribution direction from the second panel 201F to the first panel 201B, and the GPU chip cold plate 2023 is located within the area enclosed by the first peripheral component cold plate 2022A and the second peripheral component cold plate 2022B. At this time, the GPU chip cold plate 2023 does not need to occupy additional space in the chassis 201 along the distribution direction from the second panel 201F to the first panel 201B, but is located within the space occupied by the peripheral cold plate 2022 along the distribution direction from the second panel 201F to the first panel 201B (refer to the x direction in Figure 2 ) in the chassis 201. Therefore, the arrangement of the peripheral cold plate 2022 and the GPU chip cold plate 2023 can effectively reduce the space occupied by the graphics card liquid cooling module 202 in the chassis 201 and improve the integration of the graphics card liquid cooling module 202, the motherboard liquid cooling module 203, and the rear card liquid cooling module 204.

[0107] In one example, Figure 6D shows a schematic structural diagram of an example of the first peripheral component cold plate of the embodiment of the present application. As shown in Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A , Figure 6A - Figure 6D , Figure 2 , Figure 3A , Figure 3B , Figure 4 , Figure 5A , Figure 6A - Figure 6D ​​​​​​​​​​​​​​​​As shown, the number of the first peripheral component cold plates 2022A is multiple, and the multiple first peripheral component cold plates 2022A can be connected in series with the first liquid distribution device 2021. For example, the first peripheral component cold plate 2022A can include a first microchannel cold plate provided on the first switching chip. The first switching chip can be located on one side of the first microchannel cold plate close to the lower end plate 201D.

[0108] In one example, Figure 6E Fig. shows a schematic structural diagram of an example of the second peripheral component cold plate according to an embodiment of the present application. As Figure 6A 、 Figure 6B and Figure 6E shown, the number of the second peripheral component cold plates 2022B is multiple, and the multiple second peripheral component cold plates 2022B are connected in series with the first liquid distribution device 2021. For example, the second peripheral component cold plate 2022B includes a first embedded tube cold plate 2022B1 provided on the first GPU power supply, a second embedded tube cold plate 2022B2 provided on the peripheral circuit, and a third embedded tube cold plate 2022B3 provided on the second GPU power supply. At this time, the first switching chip, the first GPU power supply, the peripheral circuit, and the second GPU power supply can all dissipate heat through liquid cooling.

[0109] Compared with the energy consumption of the first switching chip, the energy consumption of the first CPU power supply, the second CPU power supply, and the peripheral circuit is relatively low. Therefore, the first switching chip can select the first microchannel cold plate with relatively strong liquid cooling ability for heat dissipation, while the first CPU power supply, the second CPU power supply, and the peripheral circuit can select the embedded tube cold plate with relatively weak liquid cooling ability for heat dissipation, which can effectively reduce the hardware cost.

[0110] For example, when the peripheral circuit includes a first retimer chip, an FPGA chip, and a second retimer chip, the first power supply, the first retimer chip, the FPGA chip, the second retimer chip, and the second power supply are connected in series to the first liquid distribution device. Among them, the Retimer chip is an integrated circuit chip used to recover and re-time the high-speed signals of the GPU. The FPGA chip is a field programmable gate array chip.

[0111] In the embodiment of the present application, it is possible to Figure 6E The material of the embedded tubes on the first embedded tube cold plate 2022B1, the second embedded tube cold plate 2022B2, and the third embedded tube cold plate 2022B3 can be selected as materials such as copper embedded tubes, and specific selection is made according to the actual situation. Moreover, the bending path of the embedded tubes can be adjusted to achieve uniform heat dissipation and flow resistance control for the first power supply, the first retimer chip, the FPGA chip, the second retimer chip, and the second power supply.

[0112] Optionally, as Figure 6A - Figure 6EAs shown, the number of GPU chip cold plates 2023 is multiple, and each CPU chip cold plate is disposed on the CPU chip, and the CPU chip can be located on the side of the CPU chip cold plate close to the lower end plate 201D. The multiple GPU chip cold plates 2023 are connected in parallel to the first liquid distribution device 2021. For example, when the GPU chip cold plates 2023 are divided into four GPU chip cold plate units 2023, each GPU chip cold plate unit 2023 is connected in parallel to the first liquid distribution device 2021. When the first liquid distribution device 2021 has a first liquid cooling channel 2021A and a second liquid cooling channel 2021B, the number of the first liquid supply connectors of the first liquid cooling channel 2021A and the number of the second liquid cooling inlets of the second liquid cooling channel 2021B can be designed to be six (but can also be more, such as 9 paths), so that the liquid cooling medium entering the first liquid cooling channel 2021A can be divided into six paths of liquid cooling media by the six first liquid supply connectors.

[0113] Take Figure 6A - Figure 6E as an example. The first path of liquid cooling medium can exchange heat with four first switching chips in sequence, and finally enter the second liquid cooling channel 2021B through the second liquid cooling inlet. The second path of liquid cooling medium can dissipate heat from the first power supply, the first retimer chip, the FPGA chip, the second retimer chip and the second power supply in sequence, and then enter the second liquid cooling channel 2021B. The remaining four paths of liquid cooling media respectively enter the four GPU chip units corresponding to the four GPU chip cold plate units 2023 of the corresponding GPU chip cold plates 2023 to dissipate heat, and then enter the second liquid cooling channel 2021B.

[0114] In the embodiment of the present application, as Figure 6A - Figure 6E shown, the first liquid supply connectors of the first liquid cooling channel 2021A and the second liquid cooling inlets of the second liquid cooling channel 2021B can be adjusted and can be presented in the form of connectors, and the caliber and length of the connectors can be adjusted to design the liquid cooling medium flow rate and pressure drop entering the first peripheral component cold plate 2022A, the second peripheral component cold plate 2022B and each GPU chip cold plate 2023, so as to ensure the heat dissipation effect of the first peripheral component cold plate 2022A, the second peripheral component cold plate 2022B and each GPU chip cold plate 2023.

[0115] Figure 6F shows a schematic structural diagram of an example of the GPU chip cold plate unit in the embodiment of the present application. As Figure 2 、 Figure 3A 、 Figure 3B 、 Figure 4 、 Figure 5A 、 Figure 6A - Figure 6FAs shown, the GPU chip cold plate 2023 unit includes a second microchannel cold plate 2023A, a third microchannel cold plate 2023B, a first liquid cooling pipe 2023Y1, and a second liquid cooling pipe 2023Y2. The second microchannel cold plate 2023A and the third microchannel cold plate 2023B are arranged on the GPU chip. When the chassis 201 has an upper end plate 201U and a lower end plate 201D, both the first liquid cooling pipe 2023Y1 and the second liquid cooling pipe 2023Y2 are arranged on the surface of the second microchannel cold plate 2023A close to the upper end plate 201U and the surface of the second microchannel cold plate 2023A close to the upper end plate 201U.

[0116] As Figure 6A - Figure 6F shown, the liquid cooling outlets of the first liquid cooling channels 2021A are detachably connected to the inlets of the first liquid cooling pipe 2023Y1 and the second microchannel cold plate 2023A respectively. The outlet of the first liquid cooling pipe 2023Y1 is also connected to the liquid cooling inlet of the third microchannel cold plate 2023B. The liquid cooling outlet of the third microchannel cold plate 2023B is connected to the inlet of the second liquid cooling pipe 2023Y2. The outlets of the second liquid cooling pipe 2023Y2 and the second microchannel cold plate 2023A are both detachably connected to the liquid cooling inlets of the second liquid cooling channels 2021B.

[0117] It can be seen that, as Figure 6A - Figure 6F shown, for each GPU chip cold plate 2023 unit, the included second microchannel cold plate 2023A and third microchannel cold plate 2023B are connected in parallel between the first liquid cooling channel 2021A and the second liquid cooling channel 2021B. Moreover, by means of the first liquid cooling pipe 2023Y1 and the second liquid cooling pipe 2023Y2 arranged on the second microchannel cold plate 2023A and the third microchannel cold plate 2023B, the heat dissipation effect of the second microchannel cold plate 2023A and the third microchannel cold plate 2023B on the GPU chip can be further increased.

[0118] Optionally, as Figure 6A - Figure 6F shown, the GPU chip cold plate 2023 unit may further include a first three-way valve T1 and a second three-way valve T2. Both the first three-way valve T1 and the second three-way valve T2 are arranged on the surface of the third microchannel cold plate 2023B close to the upper end plate 201U. The liquid cooling outlet of the third microchannel cold plate 2023B is detachably connected to the inlet of the first three-way valve T1. The first outlet of the first three-way valve T1 is connected to the inlet of the first liquid cooling pipe 2023Y1. The second outlet of the first three-way valve T1 is connected to the inlet of the third microchannel cold plate 2023B.

[0119] In a possible implementation, as Figure 2 , Figure 3A , Figure 3B , Figure 4 and Figure 5AAs shown, the liquid cooling inlet 202A of the graphics card liquid cooling module is arranged close to the second panel 201F, the rear card liquid cooling module 204 is arranged close to the first panel 201B, and the chassis 201 has an upper end plate 201U and a lower end plate 201D distributed along the thickness direction of the chassis 201. The motherboard liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U. This can reduce the unnecessary space occupation of the motherboard liquid cooling module 203 and the graphics card liquid cooling module 202 in the direction from the second panel 201F to the first panel 201B, and improve the integration of the motherboard liquid cooling module 203, the graphics card liquid cooling module 202 and the rear card liquid cooling module 204.

[0120] Figure 7A FIG. 4 shows a schematic structural diagram of an example of the motherboard liquid cooling module according to an embodiment of the present application. Figure 7B FIG. 5 shows a schematic diagram of the internal liquid cooling pipeline connection of the motherboard liquid cooling module according to an embodiment of the present application. As Figure 2 、 Figure 4 、 Figure 5A 、 Figure 7A and Figure 7B shown, the motherboard liquid cooling module 203 includes a hard disk cold plate 2031, a motherboard peripheral cold plate 2032 and a CPU chip cold plate 2033. The hard disk cold plate 2031, the motherboard peripheral cold plate 2032 and the CPU chip cold plate 2033 are sequentially detachably connected in series between the second liquid supply end G12 and the second liquid return end G22.

[0121] In the embodiment of the present application, as Figure 2 、 Figure 4 、 Figure 5A 、 Figure 7A and Figure 7B shown, the liquid cooling medium sent out by the second liquid supply end G12 can sequentially cool the hard disk, motherboard peripheral components and CPU chip through the liquid cooling inlet 203A of the motherboard liquid cooling module, and then enter the second liquid return end G22 through the liquid cooling inlet 203A of the motherboard liquid cooling module. And the power consumption of the hard disk, motherboard peripheral components and CPU chip increases in sequence. Therefore, after the liquid cooling medium starts to dissipate heat from the hard disk, there is still enough cooling capacity left to dissipate heat from the subsequent motherboard peripheral components and CPU chip, thereby reducing the influence of the cascading temperature rise on the cooling capacity of the liquid cooling medium.

[0122] As Figure 7A and Figure 7B shown, when the hard disk cold plate 2031, the motherboard peripheral cold plate 2032 and the CPU chip cold plate 2033 are connected in series, the flow rate of the liquid cooling medium flowing through the hard disk cold plate 2031, the motherboard peripheral cold plate 2032 and the CPU chip cold plate 2033 is the same, so as to more accurately control the flow rate of the liquid cooling medium of each of the hard disk cold plate 2031, the motherboard peripheral cold plate 2032 and the CPU chip cold plate 2033.

[0123] In an alternative manner, as Figure 2 , Figure 4 , Figure 5A , Figure 7A and Figure 7B shown, when the server 200 further includes a first liquid distributor 205A and a second liquid distributor 205B, the liquid cooling inlet of the hard disk cold plate 2031 can be connected to the liquid cooling inlet 203A of the main board liquid cooling module, and the liquid cooling inlet 203A of the main board liquid cooling module can be detachably connected to the second liquid delivery end G12 located at the second liquid distributor 205B by means of quick connection such as hand insertion. The liquid cooling outlet of the CPU chip cold plate 2033 can be connected to the liquid cooling outlet 203B of the main board liquid cooling module, and it can be detachably connected to the second liquid return end G22 located at the second liquid distributor 205B by means of quick connection such as hand insertion.

[0124] In an alternative manner, as Figure 2 , Figure 4 , Figure 5A , Figure 7A and Figure 7B shown, when the main board liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U, the hard disk cold plate 2031, the hard disk cold plate 2031, the main board peripheral cold plate 2032 and the CPU chip cold plate 2033 are distributed along the distribution direction from the second panel 201F to the first panel 201B ( Figure 7A the x direction in Figure 7A ), and partial areas of the hard disk cold plate 2031, the main board peripheral cold plate 2032 and the CPU chip cold plate 2033 are arranged on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U, so that the main board liquid cooling module 203 can dissipate heat from the main board normally.

[0125] Figure 7C shows a schematic structural diagram of an example of the hard disk cold plate according to an embodiment of the present application. As Figure 2 - , ​ shown, the hard disk cold plate 2031 can be a multi-layer structure such as a sandwich structure, and there is a gap between adjacent layers for placing disks. In this case, the liquid cooling medium entering the disk cold plate can perform full-wrap heat dissipation on the hard disk while reducing the pressure drop. The hard disk cold plate 2031 can be located on one side of the graphics card liquid cooling module 202 close to the upper end plate 201U.

[0126] ​ shows a schematic structural diagram of an example of the main board peripheral cold plate according to an embodiment of the present application. As ​As shown in the figure, the cold plate 2032 on the periphery of the main board includes a cold plate 2032A for the single-board chip (i.e., the low-power single-board chip) and a fourth embedded tube cold plate 2032B for the CPU power supply. For example, the fourth embedded tube cold plate 2032B can all adopt an embedded tube cold plate with a large inner diameter to effectively dissipate heat from the CPU power supply under low flow resistance.

[0127] As ​ shown in the figure, the cold plate 2032A for the single-board chip and the fourth embedded tube cold plate 2032B are successively connected in series between the hard disk cold plate 2031 and the CPU chip cold plate 2033. The cold plate 2032A for the single-board chip and the fourth embedded tube cold plate 2032B can be connected in series between the hard disk cold plate 2031 and the CPU chip cold plate 2033. In this way, the liquid cooling medium flowing out of the hard disk cold plate 2031 can successively enter the cold plate 2032A for the single-board chip and the fourth embedded tube cold plate 2032B to dissipate heat from the single-board chip and the CPU power supply in sequence.

[0128] For example, as ​ shown in the figure, the liquid cooling outlet of the hard disk cold plate 2031 is detachably connected to the liquid cooling inlet of the cold plate 2032A for the single-board chip, the liquid cooling outlet of the cold plate 2032A for the single-board chip is detachably connected to the liquid cooling inlet of the fourth embedded tube cold plate 2032B, and the liquid cooling outlet of the fourth embedded tube cold plate 2032B is detachably connected to the liquid cooling inlet of the CPU chip cold plate 2033. Here, the detachable connection can be a quick connection method such as hand insertion. This can facilitate the independent maintenance and assembly of the hard disk cold plate 2031, the cold plate 2032A for the single-board chip, and the fourth embedded tube cold plate 2032B.

[0129] Compared with the energy consumption of the single-board chip, the energy consumption of the CPU power supply is relatively high. Connecting the cold plate for the single-board chip and the fourth embedded tube cold plate in series between the hard disk cold plate and the CPU chip cold plate can ensure that after the liquid cooling medium dissipates heat from the single-board chip, there is still enough cooling capacity for the CPU power supply, thereby further reducing the impact of the cascading temperature rise on the cooling capacity of the liquid cooling medium. Moreover, connecting the cold plate for the single-board chip and the fourth embedded tube cold plate in series can make the flow rates through the cold plate for the single-board chip and the fourth embedded tube cold plate the same, thus accurately controlling the flow rates of the cold plate for the single-board chip and the fourth embedded tube cold plate.

[0130] Optionally, as ​ , ​ , ​ , ​ , ​ shown in the figure, when the main board liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U, the fourth embedded tube cold plate 2032B and the cold plate 2032A for the single-board chip are distributed along the direction from the third panel 201R to the fourth panel 201L (such as ​The distribution in the y direction). Moreover, a partial area of the third peripheral component can extend along the direction close to the fourth panel 201L to the graphics card liquid cooling module 202, so as to ensure that the single-board chip cold plate 2032A can normally dissipate heat from the single-board chip.

[0131] ​ Fig. shows a schematic structural diagram of an example of the CPU chip cold plate according to an embodiment of the present application. As ​ 、 ​ 、 ​ 、 ​ 、 ​ 、 ​ shown, the CPU chip cold plate 2033 includes a memory cold plate 2033A, a heat conduction structure 2033B for arranging the CPU memory chip, and a fourth microchannel cold plate 2033C for arranging the CPU chip. The memory cold plate 2033A and the fourth microchannel cold plate 2033C are connected in series between the motherboard peripheral cold plate 2032 and the third liquid return end G23. At this time, the liquid cooling outlet of the motherboard peripheral cold plate 2032 is detachably connected to the liquid cooling inlet of the memory cold plate 2033A, the liquid cooling outlet of the memory cold plate 2033A is detachably connected to the liquid cooling inlet of the fourth microchannel cold plate 2033C, and the liquid cooling outlet of the fourth microchannel cold plate 2033C is detachably connected to the third liquid return end G23. The detachable connection here can be a quick connection method such as hand insertion, so that the memory cold plate 2033A, the CPU chip cold plate 2033, and the motherboard peripheral cold plate 2032 can be independently maintained and assembled.

[0132] In the embodiment of the present application, as ​ shown, one end of the heat conduction structure 2033B can be in contact with the memory cold plate 2033A. At this time, the heat conduction structure 2033B can absorb the heat released by the CPU memory chip and conduct it into the memory cold plate 2033A. The memory cold plate 2033A can be a large-inner-diameter straight-through cold plate, which has a low flow resistance, thereby improving the heat dissipation effect on the CPU memory.

[0133] As ​ shown, after the liquid cooling medium flowing out of the motherboard peripheral cold plate 2032 passes through the memory cold plate 2033A, although the temperature of the liquid cooling medium increases, resulting in a decrease in the liquid cooling capacity of the liquid cooling medium, the liquid cooling capacity of the fourth microchannel cold plate 2033C is higher than that of the memory cold plate 2033A. Therefore, after the liquid cooling medium enters the fourth microchannel cold plate 2033C, it can still ensure the heat dissipation effect on the CPU chip, thereby reducing the influence of the cascaded temperature increase on the heat dissipation effect of the liquid cooling medium.

[0134] As ​ 、 ​ 、 ​ 、 ​ 、 ​As shown, the fourth microchannel cold plate 2033C and the memory cold plate 2033A are arranged along the distribution direction from the second panel 201F to the first panel 201B (such as the x-direction in ​ ), while the heat conduction structure 2033B and the fourth microchannel cold plate 2033C are arranged side by side along the distribution direction from the third panel 201R to the fourth panel 201L (such as the y-direction in ​ ). In ​ , the number of CPU chip cold plates 2033 is two, the number of heat conduction structures 2033B is three, and the CPU chip is located between two adjacent heat conduction structures 2033B.

[0135] Optionally, as shown in ​ , ​ , ​ , ​ , ​ , when the main board liquid cooling module 203 is located on the surface of the graphics card liquid cooling module 202 close to the upper end plate 201U, part of the area of the heat conduction structure 2033B and part of the area of the fourth microchannel cold plate 2033C are both arranged on the surface of the graphics card liquid cooling module 202 close to the upper surface, while the memory cold plate 2033A is located between the graphics card liquid cooling module 202 and the rear card liquid cooling module 204, and the memory cold plate 2033A is located on the side of the graphics card liquid cooling module 202 close to the first panel 201B.

[0136] As shown in ​ , ​ , ​ , ​ , ​ , when the heat conduction structure 2033B extends out of the graphics card liquid cooling module 202 along the direction close to the first panel 201B, the surface of the heat conduction structure 2033B close to the lower end plate 201D contacts the CPU memory on the main board, so as to normally conduct the heat dissipated by the CPU memory to the memory cold plate 2033A. When the fourth microchannel cold plate 2033C extends out of the graphics card liquid cooling module 202 along the direction close to the first panel 201B, the surface of the fourth microchannel cold plate 2033C close to the lower end plate 201D can contact the CPU chip on the main board to ensure normal heat dissipation of the CPU chip on the main board.

[0137] Optionally, as shown in ​ , ​ , ​ , ​ , ​As shown in the figure, the cold plate 2032 on the periphery of the main board includes the cold plate 2032A of the single-board chip provided on the single-board chip and the fourth embedded tube cold plate 2032B provided on the CPU power supply. When the CPU chip cold plate 2033 includes the memory cold plate 2033A, the heat conduction structure 2033B for setting the CPU memory chip, and the fourth microchannel cold plate 2033C for setting the CPU chip, the liquid cooling medium sent by the second liquid supply end G12 into the main board liquid cooling module 203 can dissipate heat from the hard disk, the single-board chip, the CPU power supply, the CPU memory, and the CPU chip in sequence.

[0138] In the embodiment of the present application, the power consumptions of the hard disk, the single-board chip, the CPU power supply, the CPU memory, and the CPU chip increase in sequence. Therefore, the cold plate on the periphery of the main board can adopt a heat dissipation method of first dissipating heat from low-power components and then dissipating heat from high-power components. In this heat dissipation method, after the liquid cooling medium dissipates heat from the low-power components, there is still enough cooling capacity to dissipate heat from the high-power components, thus ensuring the heat dissipation effect of the liquid cooling medium on the high-power components.

[0139] In a possible implementation manner, ​ shows a schematic structural diagram of an example of the rear card liquid cooling module in the embodiment of the present application. ​ shows a schematic diagram of the internal liquid cooling pipeline connection of the rear card liquid cooling module in the embodiment of the present application. As ​ 、 ​ 、 ​ 、 ​ and ​ shown, the rear card liquid cooling module 204 includes an exchange chip cold plate 2041, a second liquid distribution device 2042, and a plurality of rear card cold plates 2043. The second liquid distribution device 2042 has a second liquid cooling inlet and a second liquid cooling outlet. The third liquid supply end G13 is connected to the liquid cooling inlet of the exchange chip cold plate 2041. The liquid cooling outlet of the exchange chip cold plate 2041 is connected to the second liquid cooling inlet. The plurality of rear card cold plates 2043 are connected in parallel to the second liquid distribution device 2042, and the second liquid cooling outlet is connected to the third liquid return end G23.

[0140] As ​ 、 ​ 、 ​ 、 ​ and ​As shown, the liquid cooling medium of the third liquid supply end G23 can first enter the liquid cooling plate 2041 of the switching chip to cool the second switching chip, and then enter the second liquid distribution device 2042 from the second liquid cooling inlet, so that the second liquid distribution device 2042 divides the liquid cooling medium into multiple branches. The liquid cooling medium of each branch can enter the corresponding rear card liquid cooling plate 2043 to cool the rear card liquid cooling plate 2043. The liquid cooling medium flowing out of the rear card liquid cooling plate 2043 can return to the second liquid distribution device 2042 again, and finally the liquid cooling medium flows out of the second liquid distribution device 2042 from the second liquid cooling outlet. It can be seen that the rear card liquid cooling module 204 can implement the liquid cooling solution for the rear card in a series-parallel manner, making the liquid cooling plate 2041 of the switching chip in series with the second liquid distribution device 2042, while multiple rear card liquid cooling plates 2043 are in parallel, so as to improve the heat dissipation effect of the liquid cooling medium on multiple rear card liquid cooling plates 2043. It should be noted that the number of rear card liquid cooling plates 2043 can be set according to the actual situation. For example ​ 8 rear card liquid cooling plates 2043 are illustrated in

[0141] In an alternative embodiment, as shown in ​ 、 ​ 、 ​ 、 ​ and ​ the connection manners between the third liquid supply end G13 and the liquid cooling plate 2041 of the switching chip, between the liquid cooling plate 2041 of the switching chip and the second liquid distribution device 2042, and between the second liquid distribution device 2042 and each rear card liquid cooling plate 2043 are all quick connection manners such as hand insertion or blind insertion, which can ensure the independent maintenance and installation of the liquid cooling plate 2041 of the switching chip, the second liquid distribution device 2042 and multiple rear card liquid cooling plates 2043.

[0142] In an alternative embodiment, as shown in ​ 、 Figure 3A 、 Figure 3B 、 Figure 4 、 Figure 5A 、 Figure 8A and Figure 8B the second liquid distribution device 2042 can be arranged on the surface of the liquid cooling plate 2041 of the switching chip close to the upper end plate 201U, and the second liquid distribution device 2042 and multiple rear card liquid cooling plates 2043 are arranged along the distribution direction from the second panel 201F to the first panel 201B (for example Figure 8AThe distribution in the x - direction). At this time, the second liquid - separating device 2042 does not need to additionally occupy the area in the chassis 201 along the distribution direction from the second panel 201F to the first panel 201B, thereby reducing the space of the rear - card liquid - cooling module 204 in the chassis 201 along the distribution direction from the second panel 201F to the first panel 201B, and improving the integration degree of the graphics - card liquid - cooling module 202, the motherboard liquid - cooling module 203, and the rear - card liquid - cooling module 204.

[0143] Figure 8C shows a schematic structural diagram of an example of the cold plate of the switching chip according to an embodiment of the present application. As Figure 8C shown, the switching - chip cold plate 2041 may include a fifth - embedded - tube cold plate provided on the second switching chip. The second switching chip may be located in the fifth - embedded - tube cold plate, and the fifth - embedded - tube cold plate may be a large - flow - rate and low - flow - resistance embedded - tube cold plate. For example, the heat - dissipation effect and the flow resistance can be controlled by adjusting the tube diameter and the bending radius of the embedded tubes in the fifth - embedded - tube cold plate.

[0144] In one example, as Figure 2 、 Figure 3A 、 Figure 3B 、 Figure 5A and Figure 8C shown, along the distribution direction from the third panel 201R to the fourth panel 201L (such as the y - direction in Figure 8C ), the liquid - cooling inlet of the fifth - embedded - tube cold plate is located at one end of the fifth - embedded - tube cold plate close to the third panel 201R. As Figure 8C shown, the liquid - cooling outlet 2041A of the fifth - embedded - tube cold plate is located at one end of the fifth - embedded - tube cold plate close to the fourth panel 201L.

[0145] In one example, as Figure 2 ~、 Figure 3A 、 Figure 3B 、 Figure 5A and Figure 8C shown, the liquid - cooling inlet of the fifth - embedded - tube cold plate may be connected to the liquid - cooling outlet of the rear - card liquid - cooling module 204 for being hand - inserted into the third liquid - supply end G13, while the liquid - cooling outlet 2041A of the fifth - embedded - tube cold plate may be detachably connected to the second liquid - cooling inlet of the second liquid - separating device 2042.

[0146] Figure 8D shows a schematic structural diagram of an example of the second liquid - separating device according to an embodiment of the present application. As Figure 2 、 Figure 3A 、 Figure 3B and Figure 8D shown, the second liquid - separating device 2042 has a third liquid - cooling channel 2042A and a fourth liquid - cooling channel 2042B. Both the third liquid - cooling channel 2042A and the fourth liquid - cooling channel 2042B are along the distribution direction from the third panel 201R to the fourth panel 201L (such as Figure 8Din the y direction).

[0147] Figure 8E shows a schematic diagram of an exemplary connection between the cold plate of the switching chip and the second liquid distribution device according to an embodiment of the present application. As Figure 2 , Figure 3A , Figure 3B , 8C ~ Figure 8E shown, when the second liquid distribution device 2042 is provided on the surface of the cold plate 2041 of the switching chip close to the upper end plate 201U, the third liquid cooling channel 2042A and the fourth liquid cooling channel 2042B can be arranged side by side on the top of the cold plate 2041 of the switching chip. And the cold plate 2041 of the switching chip, the third liquid cooling channel 2042A and the fourth liquid cooling channel 2042B are all along the distribution direction from the third panel 201R to the fourth panel 201L (such as Figure 8D in the y direction). When the liquid cooling inlet J10 of the third liquid cooling channel is close to the fourth panel 201L, the liquid cooling inlet J10 of the third liquid cooling channel can be detachably connected to the liquid cooling outlet 2041A of the fifth inserted tube cold plate by a hand-inserting method as the second liquid cooling inlet of the second liquid distribution device.

[0148] Figure 8F shows a simplified connection schematic diagram of the post-card liquid cooling module according to an embodiment of the present application. As Figures 8A - 8F shown, a plurality of post-card cold plates 2043 can be divided into a first post-card cold plate 2043A and a second post-card cold plate 2043B provided on the top of the first post-card cold plate 2043A. It should be understood that the first post-card corresponding to the first post-card cold plate 2043A and the second post-card corresponding to the second post-card cold plate 2043B can be a sound card, a network card, etc., but are not limited thereto.

[0149] As Figures 8A - 8F shown, the plate surface of the first post-card cold plate 2043A is perpendicular to the upper end plate 201U, and the plate surface of the second post-card cold plate 2043B is parallel to the upper end plate 201U. At this time, the liquid cooling inlet of the first post-card cold plate 2043A can be blindly inserted into the third liquid cooling outlet of the second liquid distribution device 2042, and the liquid cooling outlet of the first post-card cold plate 2043A can be blindly inserted into the third liquid cooling inlet of the second liquid distribution device 2042.

[0150] Optionally, as Figures 8A - 8FAs shown, both the liquid cooling inlet J10 of the third liquid cooling channel and the liquid cooling outlet of the third liquid cooling channel 2042A can be provided on the surface of the third liquid cooling channel 2042A close to the upper end plate 201U. The liquid cooling inlet J10 of the third liquid cooling channel can be located at the end of the third liquid cooling channel 2042A. For example, the liquid cooling inlet J10 of the third liquid cooling channel can be close to one end of the fourth panel 201L or close to one end of the third panel 201R. The liquid cooling outlet of the third liquid cooling channel 2042A is equivalent to the third liquid cooling outlet of the second liquid distribution device 2042, and it can be detachably connected to the liquid cooling inlet of the rear card cooling plate 2043 in a quick connection manner. For example, the liquid cooling outlet of the third liquid cooling channel 2042A can be divided into two types. One is the blind plug liquid cooling outlet J11, which can be blindly plugged into the liquid cooling inlet of the first rear card cooling plate 2043A, and the other is the hand plug liquid cooling outlet J12, which can be hand plugged into the liquid cooling inlet of the second rear card cooling plate 2043B.

[0151] As Figures 8A - 8F shown, both the liquid cooling inlet of the fourth liquid cooling channel 2042B and the liquid cooling outlet J20 of the fourth liquid cooling channel can be provided on the surface of the fourth liquid cooling channel 2042B close to the upper end plate 201U. For example, the liquid cooling inlet of the fourth liquid cooling channel 2042B is equivalent to the third liquid cooling inlet of the second liquid distribution device 2042, and it can be divided into two types. One is the blind plug liquid cooling inlet J21, which can be blindly plugged into the liquid cooling outlet of the first rear card cooling plate 2043A, and the other is the hand plug liquid cooling inlet J22, which can be hand plugged into the liquid cooling outlet of the second rear card cooling plate 2043B.

[0152] As Figures 8A - 8F shown, the liquid cooling outlet J20 of the fourth liquid cooling channel, as the second liquid cooling outlet of the second liquid distribution device 2042, can be close to the third panel 201R or close to the fourth panel 201L. For example, when the liquid cooling outlet J20 of the fourth liquid cooling channel is close to the third panel 201R, the liquid cooling outlet of the first rear card cooling plate 2043A is blindly plugged into the blind plug liquid cooling inlet J21 of the fourth liquid cooling channel, and the liquid cooling inlet of the second rear card cooling plate 2043B is hand plugged into the hand plug liquid cooling inlet J22 of the fourth liquid cooling channel. The liquid cooling outlet J20 of the fourth liquid cooling channel can be connected to the liquid cooling outlet 204B of the rear card liquid cooling module.

[0153] From Figure 8FIt can be seen that the second liquid separation device 2042 can support two quick-connection methods: blind insertion and hand insertion, so as to achieve the quick connection between the second liquid separation device 2042 and the first rear-mounted card cold plate 2043A and the second rear-mounted card cold plate 2043B. In this way, each liquid cooling component in the rear-mounted card liquid cooling module 204 can be maintained independently. Moreover, by adjusting the inner diameters and lengths of the blind-inserted liquid cooling outlet J11, the blind-inserted liquid cooling inlet J21, the hand-inserted liquid cooling outlet J12, and the hand-inserted liquid cooling inlet J22, the pressure drop and flow rate of the first rear-mounted card cold plate 2043A and the second rear-mounted card cold plate 2043B can be adjusted.

[0154] Although the present application has been described in connection with specific features and their embodiments, it is obvious that various modifications and combinations can be made without departing from the spirit and scope of the present application. Accordingly, this specification and the drawings are merely exemplary illustrations of the present application defined by the appended claims, and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A server, characterized in that, Comprising: A chassis having opposite first and second panels, the first panel having a coolant inlet and a coolant outlet; A graphics card liquid cooling module, a motherboard liquid cooling module, and a rear card liquid cooling module disposed within the chassis, the graphics card liquid cooling module being pluggable into and out of the chassis through an opening in the second panel; A main liquid supply pipe connected to the coolant inlet, having a first liquid supply end, a second liquid supply end, and a third liquid supply end, the first liquid supply end being disposed near the second panel, the liquid cooling inlet of the graphics card liquid cooling module being blindly plugged into the first liquid supply end, the liquid cooling inlet of the motherboard liquid cooling module being connected to the second liquid supply end, and the liquid cooling inlet of the rear card liquid cooling module being connected to the third liquid supply end; And a main liquid return pipe connected to the coolant outlet, having a first liquid return end, a second liquid return end, and a third liquid return end, the first liquid return end being disposed near the second panel, the liquid cooling outlet of the graphics card liquid cooling module being blindly plugged into the first liquid return end, the liquid cooling outlet of the motherboard liquid cooling module being connected to the second liquid return end, and the liquid cooling outlet of the rear card liquid cooling module being connected to the third liquid return end.

2. The server according to claim 1, wherein The server further includes a first liquid distributor and a second liquid distributor disposed inside the chassis, the main liquid supply pipe includes a first liquid supply pipe and a second liquid supply pipe, and the main liquid return pipe includes a first liquid return pipe and a second liquid return pipe; One end of the first liquid supply pipe is connected to the coolant inlet, and the other end is respectively connected to the first liquid supply end and the second liquid supply pipe through the first liquid distributor, and the second liquid supply end and the third liquid supply end are connected to the second liquid supply pipe through the second liquid distributor; One end of the first liquid return pipe is connected to the coolant outlet, and the other end is respectively connected to the first liquid return end and the second liquid return pipe through the first liquid distributor, and the second liquid return end and the third liquid supply end are connected to the second liquid return pipe through the second liquid distributor.

3. The server according to claim 2, wherein The first liquid supply pipe and the first liquid return pipe are arranged side by side, the second liquid supply pipe and the second liquid return pipe are arranged side by side, and the first liquid supply pipe and the second liquid supply pipe are stacked in the thickness direction of the chassis.

4. The server according to claim 2, wherein The liquid cooling inlet of the graphics card liquid cooling module and the first liquid distributor are both disposed near the second panel, the rear card liquid cooling module is disposed near the first panel, and the second liquid distributor is disposed near the rear card liquid cooling module.

5. The server according to any one of claims 1 to 4, characterized in that The graphics card liquid cooling module includes: a first liquid distribution device, a GPU chip cold plate, and a peripheral cold plate, the GPU chip cold plate and the peripheral cold plate are located on a side of the first liquid distribution device close to the first panel, and the first liquid distribution device has a first liquid cooling inlet and a first liquid cooling outlet; The first liquid cooling inlet is blindly plugged into the first liquid supply end, the first liquid cooling outlet is blindly plugged into the first liquid return end, the liquid cooling inlets of the GPU chip cold plate and the peripheral cold plate are both connected to the liquid cooling outlet of the first liquid distribution device, and the liquid cooling outlets of the GPU chip cold plate and the peripheral cold plate are both connected to the liquid cooling inlet of the first liquid distribution device.

6. The server according to claim 5, wherein The peripheral cold plate includes a first peripheral component cold plate and a second peripheral component cold plate; The first liquid separation device, the first peripheral component cold plate, and the GPU chip cold plate are distributed along the distribution direction from the second panel to the first panel, and the GPU chip cold plate is located within the area enclosed by the first peripheral component cold plate and the second peripheral component cold plate.

7. The server according to claim 6, characterized in that, The first peripheral component cold plate includes a first microchannel cold plate assembly provided on the first switching chip; and / or The second peripheral component cold plate includes a first embedded tube cold plate provided on the first GPU power supply, a second embedded tube cold plate 2022A2 provided on the peripheral circuit, and a third embedded tube cold plate provided on the second GPU power supply.

8. The server according to claim 6, wherein The number of the GPU chip cold plates is multiple, and the multiple GPU chip cold plates are connected in parallel to the first liquid separation device. The multiple first peripheral component cold plates are connected in series to the first liquid separation device, and the multiple second peripheral cold plates are connected in series to the first liquid separation device.

9. The server according to any one of claims 1 to 8, characterized in that, The main board liquid cooling module includes a hard disk cold plate, a main board peripheral cold plate, and a CPU chip cold plate. The hard disk cold plate, the main board peripheral cold plate, and the CPU chip cold plate are detachably connected in series between the second liquid supply end and the second liquid return end in sequence.

10. The server according to claim 9, characterized in that, The main board peripheral cold plate includes a single board chip cold plate provided on the single board chip and a fourth embedded tube cold plate provided on the CPU power supply. The single board chip cold plate and the fourth embedded tube cold plate are connected in series between the hard disk cold plate and the CPU chip cold plate in sequence.

11. The server according to any one of claims 1 to 8, characterized in that, The rear card liquid cooling module includes a switching chip cold plate, a second liquid separation device, and multiple rear card cold plates. The second liquid separation device has a second liquid cooling inlet and a second liquid cooling outlet; The third liquid supply end is connected to the liquid cooling inlet of the switching chip cold plate. The liquid cooling outlet of the switching chip cold plate is connected to the second liquid cooling inlet. The multiple rear card cold plates are connected in parallel to the second liquid separation device, and the second liquid cooling outlet is connected to the third liquid return end.

12. A cabinet, characterized in that, It includes a cabinet body and a server provided in the cabinet body. The server is the server according to any one of claims 1 to 11.

Images