TWI877054B - Liquid cooling server - Google Patents

Abstract

A liquid cooling server includes: a casing, a GPU, a CPU and a liquid cooling module. The casing includes several to-be-cooled areas and a pipe bracket, and one to-be-cooled area is provided with at least one GPU and/or at least one CPU. The liquid cooling module includes a liquid distributor and several liquid cooling structures. The liquid distributor is fixed on the casing, and each liquid cooling structure is connected to the liquid distributor to form a liquid cooling loop, and coolant flows in the liquid cooling loop. The pipe bracket is used for a liquid inlet pipe and a liquid outlet pipe to be placed thereon. The liquid inlet pipe and the liquid outlet pipe are connected to the liquid distributor. The liquid cooling structures correspond to the to-be-cooled areas one by one, and the number of liquid cooling plates included in the liquid cooling structure is equal to the sum of the number of GPU and the number of CPU in the corresponding to-be-cooled area. One liquid cooling plate is in direct contact with one GPU or one CPU, and the liquid cooling plate performs heat exchange with the GPU or CPU through the coolant flowing therethrough. The liquid cooling server in the present invention has good heat dissipation efficiency, and the product structure is simple and easy to implement.

Description

Liquid Cooling Server

The present invention relates to the field of server technology, and in particular to a liquid cooling server.

With the booming development of graphics processing unit (GPU) computing, actual business has higher and higher computing power requirements for the underlying hardware infrastructure. The improvement of server performance directly leads to an increase in server power consumption. At the same time, the process upgrade brings not only a leap in computing power, but also a significant increase in overall power consumption and heat generation. When the number of servers in a single cabinet remains unchanged, the significant increase in power consumption of the entire cabinet brings great challenges to the heat exchange of the data center.

In order to cope with the above challenges, server cooling technology also needs to be innovated. Traditional air cooling has gradually reached its limit. Facing highly integrated, high power consumption, and high heat generation servers, liquid cooling technology can remove the heat generated inside the server through liquid circulation, with the advantages of high heat dissipation efficiency, low noise, and high energy efficiency.

However, due to the limited space in the server chassis and unreasonable layout inside the chassis, the design of liquid cooling is difficult and it is difficult to achieve the expected cooling effect.

The present invention provides a liquid cooling server, which has good heat dissipation efficiency, simple product structure and is easy to implement.

According to one aspect of the present invention, a liquid cooling server is provided, comprising: a chassis, a graphics processing unit (GPU), a central processing unit (CPU) and a liquid cooling module; wherein:

The chassis includes a plurality of cooling zones and water pipe brackets, and at least one GPU and/or at least one CPU is arranged in one cooling zone;

The liquid cooling module includes a liquid distributor and a plurality of liquid cooling structures; the liquid distributor is fixed on the chassis, and each liquid cooling structure is connected to the liquid distributor to form a liquid cooling circuit, and the cooling liquid flows in the liquid cooling circuit; the water pipe bracket is arranged on a side of the liquid distributor away from the liquid cooling structure, and is used to place a liquid inlet pipe and a liquid outlet pipe, and the liquid inlet pipe and the liquid outlet pipe are respectively connected to the liquid distributor;

The liquid cooling structure corresponds to the cooling zone one by one. The number of liquid cooling plates included in the liquid cooling structure is equal to the sum of the number of GPUs and CPUs in the corresponding cooling zone. One liquid cooling plate is in direct contact with one GPU or one CPU. The liquid cooling plate realizes heat exchange with the GPU or CPU through the cooling liquid flowing through the inside.

Optionally, a plurality of areas to be cooled are arranged along a first direction in the chassis; the liquid distributor extends along the first direction, and the liquid distributor and the liquid cooling structure are arranged along a second direction;

The second direction is perpendicular to the first direction.

Optionally, the number of CPUs is M, and the M CPUs are arranged in sequence along the first direction in the chassis; the number of GPUs is M×N, and the M×N GPUs are arranged in an array in the chassis;

Each cooling area is provided with N GPUs and a CPU arranged in sequence along the second direction;

Wherein, M and N are both integers greater than or equal to 2.

Optionally, according to the extension direction of the second direction, the liquid outlet of the previous liquid cooling plate is connected to the liquid inlet of the next liquid cooling plate, the liquid outlet of the liquid distributor is connected to the liquid inlet of the first liquid cooling plate, and the liquid inlet of the liquid distributor is connected to the liquid outlet of the last liquid cooling plate.

Optionally, the value of M is 4 and the value of N is 2.

Optionally, it also includes: a liquid leakage detection device;

The liquid leakage detection device is wound around the connecting pipe joints and connecting pipes between the liquid distributor and the liquid cooling plate, and between the two liquid cooling plates, and is used to detect whether the liquid cooling module is leaking.

Optionally, when assembling a liquid-cooled server, all liquid cooling plates are pre-connected according to the arrangement of all GPUs and all CPUs, and then clamped by an assembly tool and placed in the chassis for fixing.

Optionally, the assembly tool includes a tool bracket and a plurality of clamping structures;

The size of the tooling bracket matches the total size of all areas to be cooled;

One end of the clamping structure is slidably fixed on the tooling bracket; a handle is provided on the liquid cooling plate, and the clamping structure is used to clamp or loosen the handle.

Optionally, the clamping structure includes a fixed portion, a non-movable clamping plate, a movable clamping plate and a driving structure;

The fixing part is slidably connected to the tooling bracket;

One end of the inactive clamping plate is fixed on the fixed part, and the other end of the inactive clamping plate is provided with a first limit block; one end of the active clamping plate is fixed on the fixed part, and the other end of the active clamping plate is provided with a second limit block;

The driving structure is connected to the movable clamping plate; the movable clamping plate is displaced under the driving of the driving structure, thereby clamping or loosening the handle through the first limit block and the second limit block.

Optionally, when all pre-connected liquid cooling plates are clamped by the assembly tool, the number of clamping structures is equal to the sum of the number of all GPUs and the number of all CPUs; or,

When using the assembly tool to clamp all pre-connected liquid cooling plates, the number of clamping structures is equal to the number of all GPUs plus two.

The technical solution of the embodiment of the present invention is designed through the structure of the liquid cooling server, which includes a chassis, a GPU, a central processing unit (CPU) and a liquid cooling module. The chassis includes a plurality of cooling areas and a water pipe bracket, and at least one GPU and/or at least one CPU is arranged in one cooling area; the liquid cooling module includes a liquid distributor and a plurality of liquid cooling structures; the liquid distributor is fixed on the chassis, and each liquid cooling structure is connected to the liquid distributor to form a liquid cooling circuit, and the cooling liquid flows in the liquid cooling circuit; the water pipe bracket is arranged on the side of the liquid distributor away from the liquid cooling structure, and is used to place The liquid inlet pipe and the liquid outlet pipe are connected to the liquid distributor respectively; the liquid cooling structure corresponds to the cooling area one by one, the number of liquid cooling plates included in the liquid cooling structure is equal to the sum of the number of GPUs and CPUs in the corresponding cooling area, one liquid cooling plate is in direct contact with one GPU or one CPU, and the liquid cooling plate realizes heat exchange with the GPU or CPU through the cooling liquid flowing inside. In this way, liquid cooling heat dissipation of the GPU and CPU can be realized through the liquid cooling plate. In addition, since all liquid cooling plates are pre-connected according to the arrangement of all GPUs and all CPUs, and then clamped and placed in the chassis by the assembly fixture, the connection between the liquid cooling plates can be ensured while ensuring that there is no displacement between the liquid cooling plates and the corresponding GPU/CPU, thereby maximizing the contact area between the liquid cooling plates and the corresponding GPU/CPU and ensuring heat dissipation efficiency. The assembly efficiency of liquid cooling servers can also be improved by clamping the liquid cooling plates with the assembly fixture. Finally, this product has a simple structure and is easy to implement. Even if an existing air-cooled chassis is used, it only needs to be simply modified on the existing air-cooled chassis, so it has strong promotional potential.

It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor to limit the scope of the present invention. Other features of the present invention will become easy to understand through the following description.

In order to enable people in the technical field to better understand the solution of the present invention, the technical solution in the embodiment of the present invention will be clearly and completely described below in combination with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by people with ordinary knowledge in the field without creative labor should belong to the scope of protection of the present invention.

It should be noted that the terms "first", "second", etc. in the specification and patent scope of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatus.

Embodiment 1

FIG1 is a three-dimensional assembly diagram of a liquid-cooled server provided by the first embodiment of the present invention. As shown in FIG1 , the liquid-cooled server includes: a chassis 10, a GPU 20, a CPU 30 and a liquid cooling module 40. The chassis 10 generally includes a bottom shell and a top shell, and the bottom shell and the top shell are matched to form a receiving cavity, and various structures of the liquid-cooled server are arranged in the receiving cavity. In order to facilitate the description of the internal structure of the liquid-cooled server, the top shell is omitted in FIG1 .

Optionally, the liquid cooling server may further include other functional modules and a fan module (not shown in FIG. 1 ). Other functional modules may be understood as modules that need to be cooled except for the GPU 20 and the CPU 30, such as a power module. The fan module cools other functional modules through a fan. The fan module and the liquid cooling module 40 work together to cool the liquid cooling server, realizing the dual cooling technology of liquid cooling and air cooling, and further improving the cooling efficiency of the liquid cooling server.

Fig. 2 is a schematic diagram of the distribution of a cooling zone in a chassis provided by the first embodiment of the present invention. As shown in Fig. 2, the chassis 10 includes a plurality of cooling zones 11, and at least one GPU 20 and/or at least one CPU 30 is arranged in one cooling zone 11.

For a cooling zone 11, at least one GPU 20 and/or at least one CPU 30 is arranged in the cooling zone 11, which may include any one of the following three situations:

Case 1) There is only one GPU 20 disposed in the cooling area 11;

Case 2) There is only one CPU 30 in the cooling area 11;

Case 3) At least one GPU 20 and at least one CPU 30 are arranged in the cooling area 11.

At the same time, since the chassis 10 includes a plurality of cooling zones 11, the number of GPUs 20 and the number of CPUs 30 installed in different cooling zones 11 may be the same or different. When the number of GPUs 20 and the number of CPUs 30 installed in different cooling zones 11 are the same, the arrangement of GPUs 20 and CPUs 30 in the liquid cooling server is more regular, which is convenient for centralized deployment; when the number of GPUs 20 and the number of CPUs 30 installed in different cooling zones 11 are different, it is more suitable for flexible deployment of GPUs 20 and CPUs 30.

In the present invention, the cooling zone 11 can be reasonably divided according to the specific internal design of the liquid cooling server and the heat dissipation requirements. FIG. 2 is drawn based on the example that the number of GPUs 20 and the number of CPUs 30 arranged in each cooling zone 11 are the same, and one cooling zone 11 is provided with 2 GPUs 20 and 1 CPU 30.

FIG3 is a schematic diagram of a top view of the structure of a liquid cooling module corresponding to the cooling zone shown in FIG2 provided in the first embodiment of the present invention. As shown in FIG3, the liquid cooling module 40 includes a liquid distributor 41 and a plurality of liquid cooling structures 42. The liquid distributor 41 is fixed on the chassis 10. For example, the liquid distributor 41 can be fixed on the chassis 10 by a spring screw. Each liquid cooling structure 42 is connected to the liquid distributor 41 to form a liquid cooling circuit, and the cooling liquid flows in the liquid cooling circuit. Exemplarily, the liquid cooling structure 42 is connected to the liquid distributor 41 by a hose, that is, the liquid cooling plate 420 and the liquid distributor 41, and the liquid cooling plate 420 and the liquid cooling plate 420 are all connected by a hose. In this way, the difficulty of manufacturing can be reduced, and the problem that the hard pipe connection is easy to deform and damage is solved.

The liquid cooling structure 42 corresponds to the cooling zone 11 one by one, and the number of liquid cooling plates 420 included in the liquid cooling structure 42 is equal to the sum of the number of GPUs and CPUs in the corresponding cooling zone 11, thereby ensuring that each GPU 20/CPU 30 is provided with a liquid cooling plate 420. Continuing to refer to FIG. 2 and FIG. 3, since two GPUs 20 and one CPU 30 are provided in one cooling zone 11 shown in FIG. 2, one liquid cooling structure 42 in FIG. 3 includes three liquid cooling plates 420.

Optionally, the size of the liquid cooling plate 420 can match the GPU 20/CPU 30 that it directly contacts. For example, the size of the liquid cooling plate 420 is equal to the GPU 20/CPU 30 that it directly contacts, or the size of the liquid cooling plate 420 is slightly larger than the GPU 20/CPU 30 that it directly contacts. In this way, the liquid cooling plate 420 can be fully attached to the GPU 20/CPU 30, increasing the heat exchange area of the GPU 20/CPU 30, thereby improving the heat dissipation efficiency.

In one embodiment, referring to FIG. 1 , the chassis 10 further includes a water pipe bracket 12 . The water pipe bracket 12 is disposed on a side of the liquid distributor 41 away from the liquid cooling structure 42 , and is used to place a liquid inlet pipe and a liquid outlet pipe, which are respectively connected to the liquid distributor 41 .

The chassis 10 provided by the present invention can be modified based on a traditional air-cooled chassis. By replacing some accessories, the traditional air-cooled chassis can be directly modified, thereby controlling the cost of the chassis 10.

When the liquid-cooled server is assembled, all the liquid cooling plates 420 are pre-connected according to the arrangement of all the GPUs 20 and all the CPUs 30 (optionally, the pre-connection may also include the connection between the liquid cooling plates 420 and the liquid distributor 41), and then are clamped and placed in the chassis 10 for fixing using an assembly tool, so that a liquid cooling plate 420 is in direct contact with a GPU 20 or a CPU 30, and the liquid cooling plate 420 achieves heat exchange with the GPU 20 or the CPU 30 through the cooling liquid flowing through the inside. In this way, the connection relationship between the liquid cooling plates 420 can be ensured while ensuring that the liquid cooling plates 420 and the corresponding GPU 20/CPU 30 are completely aligned, thereby maximizing the contact area between the liquid cooling plates 420 and the corresponding GPU 20/CPU 30 and ensuring the heat dissipation efficiency. Using the assembly tool to clamp the liquid cooling plates 420 can also improve the assembly efficiency of the liquid-cooled server.

Optionally, the liquid cooling plate 420 and the GPU 20/CPU 30 to which it directly contacts can be fixed by screws to prevent the liquid cooling plate 420 from shifting after the liquid cooling server is assembled.

In one embodiment, referring to Figures 2 and 3, a plurality of cooling zones 11 are arranged along a first direction in the chassis 10; the liquid distributor 41 extends along the first direction, and the liquid distributor 41 and the liquid cooling structure 42 are arranged along a second direction that is perpendicular to the first direction.

The number of CPUs 30 is M, and the M CPUs 30 are arranged in sequence along the first direction in the chassis 10; the number of GPUs 20 is M×N, and the M×N GPUs 20 are arranged in an array in the chassis 10; each cooling area 11 is provided with N GPUs 20 and one CPU 30 arranged in sequence along the second direction, wherein M and N are both integers greater than or equal to 2. This is more in line with the conventional internal design of the server.

Preferably, the value of M is 4 and the value of N is 2.

In one embodiment, FIG4 is a cooling liquid flow diagram of the liquid cooling module shown in FIG3 provided in the first embodiment of the present invention. As shown in FIG4, according to the extension direction of the second direction, the liquid outlet of the first liquid cooling plate 420 is connected to the liquid inlet of the second liquid cooling plate 420, the liquid outlet of the liquid distributor 41 is connected to the liquid inlet of the first liquid cooling plate 420, and the liquid inlet of the liquid distributor 41 is connected to the liquid outlet of the last liquid cooling plate 420. In this way, the use of quick connectors can be reduced, and at the same time, the cooling liquid is ensured to pass through the GPU 20 with higher power consumption first, and then pass through the CPU 30, thereby improving the performance of the entire liquid cooling module 40.

In one embodiment, the liquid cooling server further includes: a liquid leakage detection device. The liquid leakage detection device is wound between the liquid distributor 41 and the liquid cooling plate 420, and on the connecting pipe joints and connecting pipes between the two liquid cooling plates 420, and is used to detect whether the liquid cooling module 40 is leaking.

Specifically, the leakage detection device can be electrically connected to the mainboard of the liquid-cooled server to send the leakage detection result to the mainboard. When the leakage detection device detects that the liquid cooling module 40 is leaking, the mainboard will sound an alarm, so that the operation and maintenance personnel can handle it in time to prevent the leaked liquid from further damaging the liquid-cooled server.

Fig. 5 is a three-dimensional structural diagram of an assembly tool provided in Embodiment 1 of the present invention. As shown in Fig. 5 , the assembly tool 50 includes a tool support 51 and a plurality of clamping structures 52.

The size of the tooling bracket 51 matches the total size of all the areas to be cooled 11; one end of the clamping structure 52 is slidably fixed on the tooling bracket 51; a handle is provided on the liquid cooling plate 420, and the clamping structure 52 is used to clamp or loosen the handle.

Specifically, the clamping structure 52 includes a fixed portion 521, a non-active clamping plate 522, a movable clamping plate 523 and a driving structure 524. The fixed portion 521 is slidably connected to the tooling bracket 51; one end of the non-active clamping plate 522 is fixed to the fixed portion 521, and the other end of the non-active clamping plate 522 is provided with a first limit block; one end of the movable clamping plate 523 is fixed to the fixed portion 521, and the other end of the movable clamping plate 523 is provided with a second limit block; the driving structure 524 is connected to the movable clamping plate 523; the movable clamping plate 523 is displaced under the driving of the driving structure 524, thereby clamping or loosening the handle through the first limit block and the second limit block.

In one embodiment, when the assembly tool 50 is used to clamp all the pre-connected liquid cooling plates 420, the number of the clamping structures 52 is equal to the sum of the number of all the GPUs 20 and the number of all the CPUs 30. In this way, it can be ensured that each liquid cooling plate 420 can be clamped by the clamping structure 52, and the stability of all the pre-connected liquid cooling plates 420 is guaranteed; or, when the assembly tool 50 is used to clamp all the pre-connected liquid cooling plates 420, the number of the clamping structures 52 is equal to the number of all the GPUs 20 plus two. In this way, it can be mainly ensured that each GPU 20 can be clamped by the clamping structure 52, and the stability of all the pre-connected liquid cooling plates 420 is guaranteed. 30 is clamped by two clamping structures 52, so that the stability of all pre-connected liquid cooling plates 420 can be ensured while saving costs.

FIG6 is a schematic diagram of a method of clamping all pre-connected liquid cooling plates using an assembly tool according to the first embodiment of the present invention; FIG7 is a schematic diagram of a method of placing all pre-connected liquid cooling plates into a chassis using an assembly tool according to the first embodiment of the present invention. As shown in FIG6 and FIG7, the assembly tool 50 is independent of the liquid cooling server, and can clamp the flexible whole formed by all pre-connected liquid cooling plates and place it into the chassis for fixing, thereby greatly reducing the difficulty of assembling the liquid cooling server and the cost of making the jig. After assembly, the assembly tool 50 can also be quickly disassembled for reuse.

In one embodiment, the connecting hose of the liquid cooling plate 420 corresponding to the GPU 20 can be connected to the cabinet branch pipe (Manifold) using a quick connector, so that the liquid cooling server can be more convenient and quick when it is put on the shelf. Moreover, when operating and maintaining the GPU 20 alone, it is only necessary to unplug the corresponding water pipe without removing the entire machine from the shelf, thereby greatly improving the operation and maintenance efficiency.

In one embodiment, GPU 20 may be an NV SWITCH chip.

The technical solution of the embodiment of the present invention is designed through the structure of the liquid cooling server, which includes a chassis, a GPU, a central processing unit (CPU) and a liquid cooling module. The chassis includes a plurality of cooling areas and a water pipe bracket, and at least one GPU and/or at least one CPU is arranged in one cooling area; the liquid cooling module includes a liquid distributor and a plurality of liquid cooling structures; the liquid distributor is fixed on the chassis, and each liquid cooling structure is connected to the liquid distributor to form a liquid cooling circuit, and the cooling liquid flows in the liquid cooling circuit; the water pipe bracket is arranged on the side of the liquid distributor away from the liquid cooling structure, and is used to place The liquid inlet pipe and the liquid outlet pipe are connected to the liquid distributor respectively; the liquid cooling structure corresponds to the cooling area one by one, the number of liquid cooling plates included in the liquid cooling structure is equal to the sum of the number of GPUs and CPUs in the corresponding cooling area, one liquid cooling plate is in direct contact with one GPU or one CPU, and the liquid cooling plate realizes heat exchange with the GPU or CPU through the cooling liquid flowing inside. In this way, liquid cooling heat dissipation of the GPU and CPU can be realized through the liquid cooling plate. In addition, since all liquid cooling plates are pre-connected according to the arrangement of all GPUs and all CPUs, and then clamped and placed in the chassis by the assembly fixture, the connection between the liquid cooling plates can be ensured while ensuring that there is no displacement between the liquid cooling plates and the corresponding GPU/CPU, thereby maximizing the contact area between the liquid cooling plates and the corresponding GPU/CPU and ensuring heat dissipation efficiency. The assembly efficiency of liquid cooling servers can also be improved by clamping the liquid cooling plates with the assembly fixture. Finally, this product has a simple structure and is easy to implement. Even if an existing air-cooled chassis is used, it only needs to be simply modified on the existing air-cooled chassis, so it has strong promotional potential.

Embodiment 2

The present invention also provides a server unit, comprising a cabinet and a plurality of liquid-cooled servers in the above-mentioned embodiments. The plurality of liquid-cooled servers are stacked in sequence in the cabinet.

The liquid cooling server comprises: a chassis, a graphics processing unit (GPU), a central processing unit (CPU) and a liquid cooling module; wherein the chassis comprises a plurality of cooling zones, each of which is provided with at least one GPU and/or at least one CPU; the liquid cooling module comprises a liquid distributor and a plurality of liquid cooling structures; the liquid distributor is fixed on the chassis, each liquid cooling structure is connected to the liquid distributor to form a liquid cooling circuit, and the cooling liquid flows in the liquid cooling circuit; the liquid cooling structure and the cooling zones are connected one by one. Correspondingly, the number of liquid cooling plates included in the liquid cooling structure is equal to the sum of the number of GPUs and CPUs in the corresponding cooling area; when the liquid cooling server is assembled, all liquid cooling plates are pre-connected according to the arrangement of all GPUs and all CPUs, and then clamped and placed in the chassis for fixing using assembly tools, so that one liquid cooling plate is in direct contact with one GPU or one CPU, and the liquid cooling plate achieves heat exchange with the GPU or CPU through the cooling liquid flowing through it.

The above specific implementation does not constitute a limitation on the protection scope of the present invention. It should be understood by those with ordinary knowledge in the field that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

10: Chassis 11: Cooling area 12: Water pipe bracket 20: GPU 30: CPU 40: Liquid cooling module 41: Liquid distributor 42: Liquid cooling structure 420: Liquid cooling plate 50: Assembly tooling 51: Tooling bracket 52: Clamping structure 521: Fixed part 522: Non-movable clamping plate 523: Movable clamping plate 524: Driving structure

In order to more clearly explain the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those with ordinary knowledge in this field, other drawings can be obtained based on these drawings without creative labor. Figure 1 is a three-dimensional assembly diagram of a liquid-cooled server provided in the first embodiment of the present invention; Figure 2 is a schematic diagram of the distribution of a cooling zone in a chassis provided in the first embodiment of the present invention; Figure 3 is a schematic diagram of a top view of the structure of a liquid cooling module corresponding to the cooling zone shown in Figure 2 provided in the first embodiment of the present invention; Figure 4 is a cooling liquid flow diagram of the liquid cooling module shown in Figure 3 provided in the first embodiment of the present invention; Figure 5 is a three-dimensional structural schematic diagram of an assembly tool provided in the first embodiment of the present invention; Figure 6 is a schematic diagram of clamping all pre-connected liquid cooling plates using an assembly tool provided in the first embodiment of the present invention; Figure 7 is a schematic diagram of placing all pre-connected liquid cooling plates into the chassis using an assembly tool provided in the first embodiment of the present invention.

10: Chassis

12: Water pipe bracket

20: GPU

30:CPU

40: Liquid cooling module

Claims (10)

A liquid cooling server, characterized in that it comprises: a chassis, a graphics processing unit (GPU), a central processing unit (CPU) and a liquid cooling module; wherein the chassis comprises a plurality of cooling zones and a water pipe bracket, and at least one GPU and/or at least one CPU is arranged in one cooling zone; the liquid cooling module comprises a liquid distributor and a plurality of liquid cooling structures; the liquid distributor is fixed on the chassis, and each of the liquid cooling structures is connected to the liquid distributor to form a liquid cooling circuit, and the cooling liquid flows in the liquid cooling circuit; the water pipe bracket The frame is arranged on a side of the liquid distributor far away from the liquid cooling structure, and is used to place a liquid inlet pipe and a liquid outlet pipe, and the liquid inlet pipe and the liquid outlet pipe are respectively connected with the liquid distributor; the liquid cooling structure corresponds to the area to be cooled one by one, and the number of liquid cooling plates included in the liquid cooling structure is equal to the sum of the number of GPUs and the number of CPUs in the corresponding area to be cooled, and one liquid cooling plate is in direct contact with one GPU or one CPU, and the liquid cooling plate realizes heat exchange with the GPU or the CPU through the coolant flowing through the inside. The liquid-cooled server as described in claim 1 is characterized in that a plurality of the areas to be cooled are arranged along a first direction in the chassis; the liquid distributor extends along the first direction, and the liquid distributor and the liquid cooling structure are arranged along a second direction; wherein the second direction is perpendicular to the first direction. The liquid-cooled server as described in claim 2 is characterized in that the number of the CPUs is M, and the M CPUs are arranged in sequence along a first direction in the chassis; the number of the GPUs is M×N, and the M×N GPUs are arranged in an array in the chassis; each of the cooling areas is provided with N GPUs and one CPU arranged in sequence along a second direction; wherein M and N are both integers greater than or equal to 2. The liquid-cooled server as described in claim 3 is characterized in that, along the extension direction of the second direction, the liquid outlet of the preceding liquid cooling plate is connected to the liquid inlet of the succeeding liquid cooling plate, the liquid outlet of the liquid distributor is connected to the liquid inlet of the first liquid cooling plate, and the liquid inlet of the liquid distributor is connected to the liquid outlet of the last liquid cooling plate. The liquid-cooled server as described in claim 3 is characterized in that the value of M is 4 and the value of N is 2. The liquid-cooled server as described in claim 4 is characterized in that it also includes: a liquid leakage detection device; the liquid leakage detection device is wrapped around the connecting pipe joints and connecting pipes between the liquid distributor and the liquid cooling plate, and between the two liquid cooling plates, and is used to detect whether the liquid cooling module is leaking. The liquid-cooled server as described in claim 1 is characterized in that when assembling the liquid-cooled server, all liquid cooling plates are pre-connected according to the arrangement of all GPUs and all CPUs, and then clamped by an assembly tool and placed in the chassis for fixation. The liquid-cooled server as described in claim 7 is characterized in that the assembly tool includes a tool bracket and multiple clamping structures; the size of the tool bracket matches the total size of all the areas to be cooled; one end of the clamping structure is slidably fixed on the tool bracket; a handle is provided on the liquid cooling plate, and the clamping structure is used to clamp or loosen the handle. The liquid-cooled server as described in claim 8 is characterized in that the clamping structure includes a fixed part, a non-active clamping plate, a movable clamping plate and a driving structure; the fixed part is slidably connected to the tooling bracket; one end of the non-active clamping plate is fixed to the fixed part, and the other end of the non-active clamping plate is provided with a first limit block; one end of the movable clamping plate is fixed to the fixed part, and the other end of the movable clamping plate is provided with a second limit block; the driving structure is connected to the movable clamping plate; the movable clamping plate is displaced under the drive of the driving structure, thereby clamping or loosening the handle through the first limit block and the second limit block. The liquid-cooled server as described in claim 8 is characterized in that, when the assembly tool is used to clamp all pre-connected liquid cooling plates, the number of the clamping structures is equal to the sum of the number of all GPUs and the number of all CPUs; or, when the assembly tool is used to clamp all pre-connected liquid cooling plates, the number of the clamping structures is equal to the number of all GPUs plus two.