Liquid-cooled server

Abstract

The present application provides a liquid-cooled server, comprising a case and a liquid cooling heat dissipation system. The case is provided with an accommodating cavity; a main control board, a central processing unit, a memory module and a graphics processing unit are provided in the accommodating cavity; and the central processing unit, the memory module and the graphics processing unit are electrically connected to the main control board; at least part of the liquid cooling heat dissipation system is located in the accommodating cavity; the liquid cooling heat dissipation system comprises a liquid cooling link; a cooling liquid in the liquid cooling link flows through a liquid cooling plate at the central processing unit, a liquid cooling assembly at the memory module, and the graphics processing unit to remove heat from the central processing unit by means of the liquid cooling plate, heat from the memory module by means of the liquid cooling assembly, and heat from the graphics processing unit by means of the cooling liquid flowing through the graphics processing unit. The present application solves the problem in the related art that the liquid cooling heat dissipation mode of liquid-cooled servers is complex.

Description

Liquid-cooled servers

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to a Chinese patent application filed with the Patent Office of China on January 31, 2024, with application number 202410136851.2 and entitled “LIQUID COOLED SERVER,” the entire contents of which are incorporated herein by reference. Technical Field

[0003] The present application relates to the technical field of electronic equipment, and in particular to a liquid-cooled server. Background Art

[0004] As society's digital transformation continues to deepen, computing power has become the core force supporting and driving the development of the digital economy, playing a vital role in promoting scientific and technological progress and social governance. At the same time, artificial intelligence technologies, exemplified by large-scale model training, are driving AI (artificial intelligence) toward greater generality. Consequently, servers, the electronic devices that provide computing power, are also evolving toward higher density and greater flexibility.

[0005] Traditional AI servers, equipped with eight GPU (Graphics Processing Unit) cards integrated on the same motherboard, are expensive, consume a lot of power, and often require a 4U or 6U height for the entire machine. However, for scenarios with smaller workloads that only require four or even two GPU cards, this loses flexibility in the number of GPUs and wastes deployment space in the IDC (Internet Data Center) rack. To address this situation, PCIe (Peripheral Component Interconnect Express) versions of GPUs have emerged. Each GPU card has an independent structure and does not need to be integrated on the same motherboard. It can be configured individually according to needs, so it can be used with general-purpose CPU (Central Processing Unit) servers. This not only meets the maximum flexibility in the number of GPUs, but also allows the entire server to be reduced to 1U or 2U. In addition, the requirements for the IDC rack are not high.

[0006] In related technologies, when PCIe GPUs are used with general-purpose CPU servers, the GPUs are typically cooled by air, relying on high-speed fans to dissipate heat. This generates considerable noise, adversely affecting both computer room operators and the surrounding environment. Furthermore, the increased fan speed consumes more power, increasing the data center's power usage effectiveness (PUE), hindering the green development of data centers focused on energy conservation and consumption reduction.

[0007] At present, liquid cooling technology is gradually being adopted in some CPU servers or 8-card GPU servers. That is, relying on coolant as a heat transfer medium, heat is brought out of the server through components such as cold plates in contact with the server's heat-generating components, providing a better operating temperature for high-power heat-generating components such as CPUs and GPUs. This can effectively improve the computing efficiency and stability of the server, meet the requirements of energy saving and noise reduction, and help increase the server density of a single cabinet, thereby improving the computing efficiency of the data center.

[0008] However, liquid cooling is difficult to apply and promote on PCIe version of GPU. The main reason is that the structure of each PCIe version of GPU card is relatively independent, and the relative position of installation in the server is not fixed. In addition, each GPU card needs to be replaced separately during operation and maintenance. The above conditions determine that each PCIe version of GPU card needs to be equipped with a separate cold plate. As a result, each PCIe version of liquid-cooled GPU card needs to have an independent supply and return liquid connector. Unlike the traditional 8-card GPU server chips are on the same horizontal plane and the relative position is fixed, the cold plate can be designed as a whole. Only one pair of supply and return liquid connectors is required. These quick connectors added to the PCIe version of liquid-cooled GPU card need to be plugged into the other end of the connector corresponding to the model to connect to the liquid link, and a suitable cooling liquid flow rate needs to be provided to meet the cooling requirements of each GPU card.

[0009] Summary of the Invention

[0010] The main purpose of the present application is to provide a liquid-cooled server to solve the problem that the liquid cooling heat dissipation method of the liquid-cooled server in the related art is relatively complicated.

[0011] In order to achieve the above-mentioned objectives, the present application provides a liquid-cooled server, including a chassis and a liquid cooling and heat dissipation system, wherein the chassis has a accommodating cavity, in which a main control board, a central processing unit, a memory stick, and a graphics processing unit are arranged, and the central processing unit, the memory stick, and the graphics processing unit are all electrically connected to the main control board; at least part of the liquid cooling and heat dissipation system is located in the accommodating cavity, and the liquid cooling and heat dissipation system includes a liquid cooling link, and the cooling liquid in the liquid cooling link flows through the liquid cooling plate at the central processing unit, the liquid cooling component at the memory stick, and the graphics processing unit to remove the heat generated by the central processing unit through the liquid cooling plate, and remove the heat generated by the memory stick through the liquid cooling component, and remove the heat generated by the graphics processing unit through the cooling liquid flowing through the graphics processing unit.

[0012] In one embodiment, there are multiple liquid cooling plates, which are arranged in series end to end on a liquid cooling link so that the multiple liquid cooling plates can perform heat exchange with one or more central processing units; and / or, there are multiple liquid cooling components, which are arranged in series end to end on a liquid cooling link so that the multiple liquid cooling components can perform heat exchange with one or more memory sticks; and / or, there are multiple graphics processing units, which are arranged in parallel on the liquid cooling link.

[0013] In one embodiment, the liquid cooling heat dissipation system also includes a conversion device, which is arranged on the liquid cooling link and located downstream of the liquid cooling component, and the conversion device has a first flow channel and a second flow channel; wherein, the first flow channel has a main input interface and multiple branch output interfaces, the main input interface is connected to the outlet of the liquid cooling component so that the coolant in the liquid cooling component is collected in the first flow channel, and the multiple branch output interfaces are respectively connected to the first liquid inlets corresponding to the multiple graphics processing units; the second flow channel has a main output interface and multiple branch input interfaces, the main output interface is connected to the external coolant recovery device, and the multiple branch input interfaces are respectively connected to the first liquid discharge ports corresponding to the multiple graphics processing units.

[0014] In one embodiment, the accommodating cavity has a first accommodating area and a second accommodating area, the second accommodating area is located above the first accommodating area, the central processing unit, memory bar and part of the liquid cooling link are located in the first accommodating area, and the graphics processing unit and at least part of the remaining liquid cooling link are located in the second accommodating area.

[0015] In one embodiment, the liquid cooling system further includes a first leakage detection line, which is located in the first accommodation area and is laid along an extension path of the liquid cooling link located in the first accommodation area to detect whether there is a coolant leak in the liquid cooling link located in the first accommodation area.

[0016] In one embodiment, the liquid cooling system further includes a liquid collection tray and a second leakage detection line, wherein the liquid collection tray is located between the first accommodating area and the second accommodating area, and at least the liquid collection tray can receive coolant leaked from the connection between the liquid cooling link and the graphics processing unit located in the second accommodating area; the second leakage detection line is located in the liquid collection tray to detect whether there is coolant in the liquid collection tray.

[0017] In one embodiment, the bottom receiving surface of the liquid collection tray is inclined, and the bottom receiving surface corresponding to the connection is the lowest surface, and at least part of the second liquid leakage detection line is laid on the lowest surface.

[0018] In one embodiment, the first leakage detection line and the second leakage detection line are plugged into each other through a male detection line connector and a female detection line connector, wherein one end of the first leakage detection line away from the second leakage detection line is connected to a control module signal to transmit the detected leakage signal to the control module; or, one end of the second leakage detection line away from the first leakage detection line is connected to a control module signal to transmit the detected leakage signal to the control module.

[0019] In one embodiment, the liquid cooling component includes a liquid separation structure, a liquid collection structure and multiple sub-cold plates, wherein the liquid separation structure has a first flow chamber and a second liquid inlet connected to the first flow chamber, and multiple first liquid separation ports; the liquid collection structure has a second flow chamber and a second liquid discharge port connected to the second flow chamber, and multiple second liquid separation ports; multiple sub-cold plates are arranged in parallel between the liquid separation structure and the liquid collection structure, and the number of sub-cold plates, the number of first liquid separation ports, and the number of second liquid separation ports are equal, so that the two ends of each sub-cold plate are respectively connected to a corresponding first liquid separation port and a second liquid separation port.

[0020] In one embodiment, the middle portion of each sub-cold plate is concave, and both ends are bent.

[0021] In one embodiment, the liquid cooling assembly further includes a heat conducting plate, which is disposed at the middle recessed portion of the sub-cold plate, and the thickness of the heat conducting plate is equal to that of the sub-cold plate.

[0022] In one embodiment, the heat conducting plate has a first avoidance gap, and the first avoidance gap is used to avoid the chip protruding from the memory module.

[0023] By applying the technical solution of the present application, a liquid-cooled server with a liquid cooling system is provided. The cooling liquid in the liquid cooling link of the liquid cooling system flows through the liquid cooling plate at the central processing unit, the liquid cooling assembly at the memory stick, and the graphics processing unit, thereby removing the heat generated by the central processing unit through the liquid cooling plate, removing the heat generated by the memory stick through the liquid cooling assembly, and removing the heat generated by the graphics processing unit by the cooling liquid flowing through the graphics processing unit, thereby ensuring that the temperatures of the central processing unit, the memory stick, and the graphics processing unit are all within the normal operating range, thereby ensuring the working reliability of the central processing unit, the memory stick, and the graphics processing unit.

[0024] In addition, by adding liquid cooling components at the memory modules and graphics processing units to the liquid cooling chain, the proportion of liquid cooling in the total power consumption of liquid-cooled servers is further increased. That is, liquid cooling accounts for more than 90% of the total power consumption of the server, which is conducive to further reducing the data center's PUE (Power Usage Effectiveness), which refers to an indicator for evaluating the energy efficiency of a data center. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings that constitute part of this application are used to provide a further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation on this application. In the drawings:

[0026] FIG1 shows an internal schematic diagram of a chassis of a liquid-cooled server according to an optional embodiment of the present application;

[0027] FIG2 shows a schematic diagram of a portion of the structure of the chassis in FIG1 ;

[0028] FIG3 shows a schematic diagram of the exploded structure of the chassis in FIG2 ;

[0029] FIG4 shows a schematic structural diagram of FIG2 omitting the graphics processing unit and the liquid accumulation tray;

[0030] FIG5 is a schematic structural diagram showing a liquid cooling link connected to the graphics processing unit and the conversion device omitted in FIG4 ;

[0031] FIG6 shows a schematic structural diagram of FIG5 omitting the conversion device;

[0032] FIG7 is a schematic diagram showing the flow direction of the cooling liquid in the liquid cooling link in FIG6 ;

[0033] FIG8 is a schematic structural diagram showing a configuration in which the graphics processing unit is omitted in FIG2 ;

[0034] FIG9 is a schematic diagram showing a portion of the structure of the conversion device in FIG8 , in which the cover is in an open state;

[0035] FIG10 shows a schematic structural diagram of the liquid cooling assembly in FIG6 ;

[0036] FIG11 shows a schematic diagram of the exploded structure of the liquid cooling assembly in FIG10 ;

[0037] FIG12 shows a schematic structural diagram of a memory bar of a liquid-cooled server according to an optional embodiment of the present application.

[0038] In particular, the above drawings include the following figure numbers: 10, chassis; 11, accommodating cavity; 12, memory bar; 121, memory chip; 13, graphics processing unit; 14, liquid cooling plate; 15, liquid cooling assembly; 151, liquid separation structure; 1511, second liquid inlet; 1512, first liquid separation port; 152, liquid collection structure; 1521, second liquid discharge port; 153, sub-cold plate; 154, heat conducting plate; 1541, first avoidance gap; 155, thermal pad; 1551, second avoidance gap; 20, liquid cooling link; 21, liquid inlet interface; 22, liquid outlet interface; 30, conversion device; 31, first flow channel; 311, main line input interface; 312, branch line output interface; 32, second flow channel; 321, main line output interface; 322, branch line input interface; 33, cover plate; 40. First liquid leakage detection line; 41. Female connector of detection line; 42. Male connector of end; 50. Liquid accumulation tray; 60. Second liquid leakage detection line; 61. Male connector of detection line; 70. Bracket structure; 100. Positioning hole; 200. Positioning guide pin; 300. Pipe joint. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and is in no way intended to limit the present application and its application or use. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0040] In order to solve the problem of complex liquid cooling heat dissipation method of liquid-cooled servers in related technologies, the present application provides a liquid-cooled server.

[0041] As shown in Figures 1 to 12, the liquid-cooled server includes a chassis 10, a liquid-cooling heat dissipation system and a liquid-cooling link 20, wherein the chassis 10 has a accommodating cavity 11, in which a main control board, a central processing unit, a memory bar 12, and a graphics processing unit 13 are arranged, and the central processing unit, the memory bar 12, and the graphics processing unit 13 are all electrically connected to the main control board; at least part of the liquid-cooling heat dissipation system is located in the accommodating cavity 11, and the liquid-cooling heat dissipation system includes a liquid-cooling link 20, and the coolant in the liquid-cooling link 20 flows through a liquid-cooling plate 14 at the central processing unit, a liquid-cooling assembly 15 at the memory bar 12, and the graphics processing unit 13, so as to remove the heat generated by the central processing unit through the liquid-cooling plate 14, remove the heat generated by the memory bar 12 through the liquid-cooling assembly 15, and remove the heat generated by the graphics processing unit 13 through the coolant flowing through the graphics processing unit 13.

[0042] The present application provides a liquid-cooled server with a liquid cooling system. The cooling liquid in the liquid cooling link 20 of the liquid cooling system flows through the liquid cooling plate 14 at the central processing unit, the liquid cooling assembly 15 at the memory bar 12, and the graphics processing unit 13, thereby removing the heat generated by the central processing unit through the liquid cooling plate 14, removing the heat generated by the memory bar 12 through the liquid cooling assembly 15, and removing the heat generated by the graphics processing unit 13 by the cooling liquid flowing through the graphics processing unit 13, thereby ensuring that the temperatures of the central processing unit, the memory bar 12, and the graphics processing unit 13 are all within the normal operating range, thereby ensuring the operating reliability of the central processing unit, the memory bar 12, and the graphics processing unit 13.

[0043] In addition, by adding the liquid cooling component 15 at the memory bar 12 and the graphics processing unit 13 to the liquid cooling link 20, the proportion of liquid cooling in the total power consumption of the liquid-cooled server is further increased. That is, liquid cooling accounts for more than 90% of the total power consumption of the server, which is conducive to further reducing the data center's PUE (Power Usage Effectiveness). PUE refers to an indicator for evaluating the energy efficiency of a data center.

[0044] It should be noted that, in this application, the above-mentioned central processing unit refers to the CPU chip, and the above-mentioned graphics processing unit 13 refers to the PCIe version of the GPU card. This application provides a liquid cooling system for a PCIe version of the GPU and a general-purpose CPU server. Based on the 2U general-purpose dual-core CPU server configuration, it can internally support liquid cooling of up to 4 PCIe version GPU cards, 32 memory sticks and two CPU chips.

[0045] Optionally, there are multiple liquid cooling plates 14, and the multiple liquid cooling plates 14 are arranged in series end to end on the liquid cooling link 20, so that the multiple liquid cooling plates 14 can perform heat exchange with one or more central processing units.

[0046] Optionally, there are multiple liquid cooling components 15 , and the multiple liquid cooling components 15 are arranged in series end to end on the liquid cooling link 20 to enable the multiple liquid cooling components 15 to perform heat exchange with one or more memory bars 12 .

[0047] Optionally, there are multiple graphics processing units 13 , and the multiple graphics processing units 13 are arranged in parallel on the liquid cooling link 20 .

[0048] Specifically, as shown in Figures 2, 3, 6 to 9, the liquid cooling system also includes a conversion device 30, which is arranged on the liquid cooling link 20 and downstream of the liquid cooling component 15. The conversion device 30 has a first flow channel 31 and a second flow channel 32; wherein, the first flow channel 31 has a main input interface 311 and multiple branch output interfaces 312, the main input interface 311 is connected to the outlet of the liquid cooling component 15, so that the coolant in the liquid cooling component 15 is collected in the first flow channel 31, and the multiple branch output interfaces 312 are respectively connected to the first liquid inlets corresponding to the multiple graphics processing units 13; the second flow channel 32 has a main output interface 321 and multiple branch input interfaces 322, the main output interface 321 is connected to the external coolant recovery device, and the multiple branch input interfaces 322 are respectively connected to the first liquid discharge ports corresponding to the multiple graphics processing units 13.

[0049] As shown in FIG9 , in order to clearly see the first flow channel 31 and the second flow channel 32 , the cover 33 is in an open state.

[0050] Specifically, the coolant flows through the four groups of liquid cooling components 15 in sequence and reaches the conversion device 30 through the liquid cooling link 20. The conversion device 30 is divided into two layers, the upper and lower layers, which can evenly divide the coolant flow into four routes to supply the graphics processing unit 13. Then, after passing through the four graphics processing units 13, the coolant re-enters the conversion device 30 to achieve the parallel connection of the four graphics processing units 13. In addition, since the branches are in a parallel structure, the design of the conversion device 30 can support flexible configuration of 2 to 4 graphics processing units 13, that is, when the number of graphics processing units 13 changes, the conversion device 30 can also evenly divide the main flow according to the actual number of graphics processing units 13 configured and supply it to each graphics processing unit 13, without affecting other components in the liquid cooling system.

[0051] It should be noted that, in the present application, it is taken into account that the cooling liquid in the liquid cooling link 20 flows through the liquid cooling plate 14 at the central processing unit, the liquid cooling component 15 at the memory stick 12, and the graphics processing unit 13. At the same time, since the liquid cooling heat dissipation system also includes a conversion device 30, the conversion device 30 is arranged on the liquid cooling link 20 and is located downstream of the liquid cooling component 15, so that the external cooling liquid supply source flows into the liquid cooling link 20 through the liquid inlet interface 21. The cooling liquid in the liquid cooling link 20 flows through the liquid cooling plate 14 at the central processing unit and the liquid cooling component 15 at the memory stick 12, and then converges to the conversion device 30 and is divided into 4 paths to supply 4 graphics processing units 13 respectively. Afterwards, the 4 graphics processing units 13 are converged to the conversion device 30 and flow to the liquid outlet interface 22 through the main output interface 321, and flow into the external cooling liquid recovery device through the liquid outlet interface 22. Here, for the subsequent For a clearer description, the section of piping between the liquid inlet port 21 and where the liquid flows out of the liquid cooling assembly 15 and converges with the conversion device 30, and the section of piping between the conversion device 30 and the liquid outlet port 22 through the main output port 321 is referred to as the first portion of the liquid cooling link 20. Furthermore, the section of piping where the liquid is split into four routes after the conversion device 30 and supplied to the four graphics processing units 13, respectively, and where the coolant in the four graphics processing units 13 converges with the conversion device 30, is referred to as the second portion of the liquid cooling link 20, hereinafter referred to as the GPU piping. The GPU piping includes a female quick-connect connector that connects to the male quick-connect connector on the graphics processing unit 13, thereby conducting the coolant and removing heat from the chips on each graphics processing unit 13 (i.e., GPU card) without any leakage. Liquid sealing can also be achieved during installation, maintenance, or when a branch circuit is not equipped with a graphics processing unit 13, that is, when the quick connector on the corresponding GPU piping is disconnected from the quick connector on the graphics processing unit 13. Finally, the liquid from the conversion device 30 flows to the liquid outlet interface 22 through the main output interface 321 , forming a complete liquid loop inside the liquid-cooled server.

[0052] As shown in Figures 1 to 3 and Figure 8, the accommodating cavity 11 has a first accommodating area and a second accommodating area. The second accommodating area is located above the first accommodating area. The central processing unit, memory bar 12 and part of the liquid cooling link 20 are located in the first accommodating area, and the graphics processing unit 13 and at least part of the remaining liquid cooling link 20 are located in the second accommodating area.

[0053] Specifically, in terms of spatial distribution, the liquid inlet interface 21 and the liquid outlet interface 22, the liquid cooling plate 14, the liquid cooling assembly 15 and the first part of the liquid cooling link 20 are arranged in the lower 1U space in the liquid-cooled server, and the conversion device 30, the graphics processing unit 13, the GPU pipeline (the second part of the liquid cooling link 20), the second leakage detection line 60 and other components are arranged in the upper 1U space, which can solve the layout problem caused by the increase of the above liquid cooling components in the limited space in the liquid-cooled server.

[0054] As shown in Figures 2 through 8, the liquid inlet port 21 and the liquid outlet port 22 are fixed in the center of the liquid-cooled server's rear window. They are primarily used to open and close the cooling liquid flow path within the liquid-cooled server. Specifically, after the liquid-cooled server is installed in the rack of the cabinet, the male quick-connect connectors on the liquid inlet and outlet ports 21 and 22, extending outside the chassis, can be connected to the cabinet manifold via quick connectors to open the liquid circuit. Coolant enters the liquid-cooled server's liquid cooling system from the liquid inlet port 21, passes through the liquid cooling link 20, and then sequentially passes through two sets of liquid cooling plates 14 and four sets of liquid cooling components 15 before reaching the conversion device 30. The conversion device 30 is located on the right side of the liquid-cooled server's rear window, dividing the coolant flow into four equal routes to supply the graphics processing units 13. Four graphics processing units 13 are arranged in two rows on the left side of the rear window, with two units per row. Each graphics processing unit 13 has a male quick-connect connector on its supply and return ports, which connect to the female connector on the liquid cooling link 20 to achieve liquid continuity. The four coolant lines then flow through the GPUs 13 and then converge into the converter 30 via the liquid cooling link 20, enabling parallel connection of the four GPUs 13. Finally, the converter 30 flows through the liquid cooling link 20 to the liquid outlet port 22, completing the liquid loop within the liquid-cooled server. Meanwhile, the quick-connect connectors for the GPUs 13 are located inside the liquid-cooled server.

[0055] It should be noted that in the present application, the quick connector of the graphics processing unit 13 is inside the liquid-cooled server. Therefore, in order to solve the risk of accidental leakage of coolant due to faults or abnormal conditions during the operation and maintenance of the liquid-cooled server, as shown in Figures 1 to 4 and Figure 6, the liquid cooling system also includes a first leakage detection line 40. The first leakage detection line 40 is located in the first accommodating area and is laid along the extension path of the liquid cooling link 20 located in the first accommodating area to detect whether there is any coolant leakage in the liquid cooling link 20, the liquid cooling plate 14, and the liquid cooling component 15 located in the first accommodating area.

[0056] As shown in Figures 1 to 4, 8 and 9, the liquid cooling system also includes a liquid collection tray 50 and a second leakage detection line 60, wherein the liquid collection tray 50 is located between the first accommodating area and the second accommodating area, and at least the liquid collection tray 50 can receive the coolant leaked from the connection between the liquid cooling link 20 and the graphics processing unit 13 located in the second accommodating area; the second leakage detection line 60 is located in the liquid collection tray 50, and is used to detect whether there is coolant in the liquid collection tray 50.

[0057] Preferably, the bottom receiving surface of the liquid collection tray 50 is inclined, and the bottom receiving surface corresponding to the connection is the lowest surface, and at least part of the second liquid leakage detection line 60 is laid on the lowest surface.

[0058] It should be noted that in the present application, the first leakage detection line 40 and the second leakage detection line 60 are plugged in and connected through the detection line male head 61 and the detection line female head 41, wherein the end of the first leakage detection line 40 away from the second leakage detection line 60 is connected to the control module signal to transmit the detected leakage signal to the control module; or, the end of the second leakage detection line 60 away from the first leakage detection line 40 is connected to the control module signal to transmit the detected leakage signal to the control module.

[0059] Preferably, the first liquid leakage detection line 40 has a male end connector 42 for signal connection with the control module.

[0060] It should be noted that in the present application, the above-mentioned control module, namely the server management module, is designed with a first leakage detection line 40 and a second leakage detection line 60 for the liquid cooling heat dissipation system of the liquid-cooled server. Once a liquid leakage occurs, the first leakage detection line 40 and / or the second leakage detection line 60 will immediately detect the abnormality and transmit it to the server management module, and issue an instruction to shut down the server power, which greatly improves the safety of the liquid-cooled server.

[0061] Specifically, the first liquid leakage detection line 40 is laid along the liquid loop, passing through the connection between the liquid inlet interface 21 and the liquid cooling link 20, the welding point between the liquid cooling plate 14 and the pipe joint 300 (wherein one end of the pipe joint 300 is used for welding and connecting with the liquid cooling plate 14, and the other end of the pipe joint 300 is used for connecting with the liquid cooling link 20), the direct connection between the liquid cooling plate 14 and the liquid cooling link 20, the welding point between the liquid separation structure 151 and the sub-cold plate 153 on the liquid cooling component 15, the welding point between the liquid collection structure 152 and the sub-cold plate 153 on the liquid cooling component 15, the connection between the conversion device 30 and the liquid cooling link 20, and the bottom of the intermediate liquid cooling link 20, etc. There are places where there is a risk of leakage. The first liquid leakage detection line 40 is combined and fixed with the liquid cooling link 20 by installing a heat shrink tubing on the outer layer or wrapping it with acetic tape. The second liquid leakage detection line 60 is laid on the liquid collection tray 50 and secured with waterproof tape. The liquid collection tray 50 is located below the quick connector between the GPU pipeline and the graphics processing unit 13, and above the liquid cooling plate 14 and liquid cooling assembly 15. If the quick connector fails during the connection or disconnection between the GPU pipeline and the graphics processing unit 13, causing leakage, the leaked liquid will drip into the liquid collection tray 50 and flow to the bottom under the guidance of the inclined surface of the liquid collection tray 50, reaching the area where the second liquid leakage detection line 60 is laid, thereby triggering an alarm. The advantages of this solution are that, on the one hand, it eliminates the need for liquid leakage detection lines laid along the GPU pipeline, saving space and detection line costs within the liquid-cooled server; on the other hand, unlike the above-mentioned liquid cooling plate 14 and liquid cooling assembly 15, which are laid along the liquid loop, the quick connector of the graphics processing unit 13 is an active component that needs to be connected when the GPU is working and disconnected during installation and maintenance. Laying the liquid leakage detection line along the liquid loop can easily cause the liquid leakage detection line to be damaged by external force. In addition, the liquid leakage detection line is difficult to fix at the quick connector, and cannot cover the entire area of ​​the liquid droplet leakage, resulting in low detection accuracy. Therefore, the above problems can be well solved by setting a liquid collection tray to collect the leaked liquid and laying the liquid leakage detection line on the liquid collection tray. In addition, clips can also be attached to the liquid collection tray to fix the GPU pipeline.

[0062] As shown in Figures 10 and 11, the liquid cooling assembly 15 includes a liquid separation structure 151, a liquid collection structure 152 and multiple sub-cold plates 153, wherein the liquid separation structure 151 has a first flow chamber and a second liquid inlet 1511 connected to the first flow chamber, and multiple first liquid separation ports 1512; the liquid collection structure 152 has a second flow chamber and a second liquid discharge port 1521 connected to the second flow chamber, and multiple second liquid separation ports; multiple sub-cold plates 153 are arranged in parallel between the liquid separation structure 151 and the liquid collection structure 152, and the number of sub-cold plates 153, the number of first liquid separation ports 1512, and the number of second liquid separation ports are all equal, so that the two ends of each sub-cold plate 153 are respectively connected to a corresponding first liquid separation port 1512 and a second liquid separation port.

[0063] It should be noted that in the present application, the liquid-cooled server includes four groups of liquid cooling components 15, which are connected in series through a liquid cooling link 20 and are locked to the chassis 10 through a bracket structure 70. Positioning holes 100 and positioning guide pins 200 are respectively designed on the bracket structure 70 and the liquid cooling components 15 to provide positioning for the installation of the liquid cooling components 15, thereby ensuring the assembly accuracy between the liquid cooling components 15 and devices such as the main control board and the memory stick 12.

[0064] Specifically, after the coolant enters the liquid-cooled server from the liquid inlet interface 21, it first passes through two groups of liquid cooling plates 14 connected in series through the liquid cooling link 20, continuously removing the heat generated by the CPU chip when it is working, ensuring that the CPU chip temperature is always within the normal range; then, the coolant flows into the four groups of liquid cooling components 15 in sequence.

[0065] As shown in FIG10 and FIG11 , the middle portion of each sub-cold plate 153 is concave, and both ends are bent.

[0066] As shown in FIG10 and FIG11 , the liquid cooling assembly 15 further includes a heat conducting plate 154 , which is disposed at the middle recessed portion of the sub-cold plate 153 , and has the same thickness as that of the sub-cold plate 153 .

[0067] As shown in FIG. 10 to FIG. 12 , the heat conducting plate 154 has a first avoidance notch 1541 . The first avoidance notch 1541 is used to avoid the memory chip 121 protruding from the memory module 12 .

[0068] Specifically, the sub-cold plate 153 is made by bending and flattening the ends of a copper tube, and the ends are welded to the liquid separation structure 151 and the liquid collection structure 152 to form a liquid passage, so that the four sub-cold plates 153 are connected in parallel. The thickness of the two heat-conducting plates 154 is consistent with the thickness of the flattened copper tube, and they are welded to the two ends of the sub-cold plate 153 respectively. The surface of the welding point with the sub-cold plate 153 needs to be flat and smooth, which is used to increase the heat conduction area with the memory particles, thereby improving the heat dissipation efficiency. The design of bending the ends of the sub-cold plate 153 avoids interference with the hippocampus on the memory slot and the memory chip 121 protruding from the center of the memory stick 12. At the same time, the disconnected area between the two heat-conducting plates 154 forms a first avoidance gap 1541, which can also effectively prevent the memory chip 121 protruding from the center of the memory from being squeezed and damaged during insertion and removal. The thermal pad 155 on each memory sub-cold plate 153 is integral, symmetrically overlapped on the left and right side surfaces and top surface of the memory sub-cold plate 153 and secured with adhesive. The thermal pad 155 has good compressibility in the thickness direction. On the one hand, it is used to fill the gap between the memory sub-cold plate 153 and the memory module 12 after installation, and to meet the dimensional tolerances of the processing and assembly process, ensuring good contact between the memory chips and the memory sub-cold plate 153, which is conducive to improving heat dissipation efficiency. On the other hand, the anti-scratch film and overlapping design on the outer layer of the thermal pad 155 help prevent deformation and damage to the thermal pad caused by the memory module disassembly and assembly process, thereby increasing the number of cycles. The thermal pad 155 has a second avoidance notch 1551 at a position opposite the first avoidance notch 1541.

[0069] Furthermore, unlike existing cold plate quick connectors located outside the server chassis 10, the quick connectors for the PCIe-based liquid-cooled GPU 13 are located inside the server chassis 10, creating the risk of liquid leaking onto the motherboard. Therefore, the design of a leak detection system is crucial. The additional liquid cooling components are difficult to implement within the limited space of a typical CPU server.

[0070] As shown in FIG. 4 to FIG. 6 , the liquid-cooled server further includes a support structure 70 . The support structure 70 is disposed in the accommodating cavity 11 . The liquid cooling assembly 15 is disposed on the support structure 70 so that the support structure 70 provides support for the liquid cooling assembly 15 .

[0071] (1) Compared with traditional general-purpose CPU liquid-cooled servers, this solution uses a liquid cooling link within a 2U general-purpose CPU liquid-cooled server to dissipate heat for high-power devices such as two CPUs, 32 memory modules, and up to four graphics processing units 13, ensuring that the temperature of each device is within the normal range. At the same time, the total flow of the liquid cooling system does not need to be increased, so the increase in the total pressure drop of the entire system caused by the above-mentioned liquid cooling components is very limited and is compatible with the existing refrigeration equipment in the data center. On the other hand, adding memory liquid cooling components and PCIe version graphics processing units 13 to the liquid cooling system further increases the proportion of liquid cooling in the total power consumption of the liquid-cooled server, which is conducive to further reducing the PUE of the data center.

[0072] (2) In order to prevent the chip protruding from the center of the memory from being squeezed and damaged during insertion and removal, the sub-cold plate 153 structure is designed to avoid the position. In this solution, the sub-cold plate 153 is made by flattening a copper tube. The difference is that the two ends of the sub-cold plate 153 are bent and welded with the water diversion pipe to form a liquid passage. In addition, the thickness of the two heat-conducting plates 154 is consistent with the thickness of the flattened copper tube. They are welded to the two ends of the sub-cold plate 153 respectively, and the surface of the welding part with the sub-cold plate 153 needs to be flat and smooth to increase the heat conduction area with the memory particles, thereby improving the heat dissipation efficiency. The design of bending the two ends of the sub-cold plate 153 avoids the interference problem between it and the hippocampus on the memory slot and the chip protruding from the center of the memory. At the same time, the disconnected area between the two heat-conducting plates 154 can also effectively avoid the chip protruding from the center of the memory, avoiding damage to the memory during operation.

[0073] (3) The conversion device 30 is designed in the node and is divided into two layers, upper and lower. Each layer has a flow channel, namely the first flow channel 31 and the second flow channel 32 mentioned above, and each layer has 5 interfaces, of which 1 is a main interface and 4 are branch interfaces. As shown in Figures 2, 3, 6 and 7, the first flow channel 31 has 1 main input interface 311 and 4 branch output interfaces 312, and the second flow channel 32 has 1 main output interface 321 and 4 branch input interfaces 322, which can realize the equalization of the coolant flow into 4 channels to supply the graphics processing unit 13. In addition, since the branches are in a parallel structure, the design of the conversion device 30 can support the flexible configuration of 2 to 4 graphics processing units 13, that is, when the number of graphics processing units 13 changes, the conversion device 30 can also evenly divide the main flow according to the actual number of graphics processing units 13 configured and supply it to each graphics processing unit 13, while not affecting other components in the liquid cooling system. The GPU pipeline features a female quick-connect connector that connects to the male quick-connect connector on the graphics processing unit 13, thereby ensuring coolant flow and removing heat from the chips on the graphics processing unit 13 (i.e., the GPU card) without leaking. This also ensures a liquid seal when the quick-connect connector on the GPU pipeline is disconnected from the quick-connect connector on the graphics processing unit 13 during installation, maintenance, or when a branch circuit is not equipped with a graphics processing unit 13.

[0074] (4) Compared with the traditional general-purpose CPU liquid-cooled server, liquid cooling components such as memory liquid cooling components, PCIe version graphics processing unit 13, conversion device 30, GPU pipeline, etc. are added. Therefore, the designed leakage detection device covers a wider area and has a different form. The difference is that the leakage detection line is divided into two parts, which are connected and transmit detection signals through the terminal block. Among them, the first leakage detection line 40 is laid along the liquid loop, passing through the connection between the liquid inlet interface 21 and the liquid outlet interface 22 and the liquid cooling link 20, the welding point between the liquid cooling plate 14 and the pipe joint 300 (wherein one end of the pipe joint 300 is used for welding connection with the liquid cooling plate 14, and the other end of the pipe joint 300 is used for connection with the liquid cooling link 20), the connection between the liquid cooling plate 14 and the liquid cooling link 20, the welding points between the liquid separation structure 151 and the liquid collection structure 152 and the sub-cold plate 153 respectively, the connection between the conversion device 30 and the liquid cooling link 20, and the bottom of the intermediate liquid cooling link 20 and other parts with leakage risks. A second leak detection line 60 is laid on a liquid collection tray and secured with waterproof tape. The liquid collection tray is located below the quick connector between the GPU pipeline and the graphics processing unit 13, and above the liquid cooling plate 14 and liquid cooling assembly 15. If the quick connector fails during the connection or disconnection between the GPU pipeline and the graphics processing unit 13, causing leakage, the leaked liquid will drip into the liquid collection tray and, guided by the inclined surface of the liquid collection tray, flow to the bottom and reach the area where the leak detection line is laid, thereby triggering an alarm. The advantages of this solution are that, on the one hand, it eliminates the need for leak detection lines laid along the GPU pipeline, saving space and detection line costs within the liquid-cooled server; on the other hand, unlike the above-mentioned liquid cooling plate 14 and liquid cooling assembly 15, which are laid along the liquid loop, the quick connector of the graphics processing unit 13 is an active component that needs to be connected when the GPU is working and disconnected during installation and maintenance. Laying the leak detection line along the liquid loop can easily cause the leak detection line to be damaged by external forces. In addition, the leak detection line is difficult to fix at the quick connector, and cannot cover the entire area of ​​the liquid droplet leakage, resulting in low detection accuracy. Therefore, the above problems can be well solved by setting a liquid collection tray to collect the leaked liquid and laying the leak detection line on the liquid collection tray.

[0075] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0076] Unless otherwise specifically stated, the relative arrangement of the parts and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present application. At the same time, it should be understood that, for ease of description, the sizes of the various parts shown in the drawings are not drawn according to actual proportional relationships. The techniques, methods and equipment known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, the techniques, methods and equipment should be considered as part of the authorization specification. In all examples shown and discussed here, any specific values ​​should be interpreted as being merely exemplary and not as limitations. Therefore, other examples of the exemplary embodiments may have different values. It should be noted that similar numbers and letters represent similar items in the following figures, and therefore, once an item is defined in one figure, it does not need to be further discussed in subsequent figures.

[0077] For ease of description, spatially relative terms such as "above", "above", "on the upper surface of", "above", etc. may be used herein to describe the spatial positional relationship of a device or feature to other devices or features as shown in the figures. It should be understood that spatially relative terms are intended to include different orientations of the device in use or operation in addition to the orientation described in the figures. For example, if the device in the drawings is inverted, the device described as "above other devices or structures" or "above other devices or structures" will be positioned as "below other devices or structures" or "below other devices or structures". Thus, the exemplary term "above" can include both "above" and "below". The device can also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatially relative descriptions used here are interpreted accordingly.

[0078] It should be noted that the terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0079] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A liquid-cooled server, characterized in that: include: A chassis (10), the chassis (10) having a receiving cavity (11), a main control board, a central processing unit (CPU), a memory bar (12), and a graphics processing unit (13) being arranged in the receiving cavity (11), and the CPU, the memory bar (12), and the graphics processing unit (13) are all electrically connected to the main control board; A liquid cooling and heat dissipation system, at least part of which is located in the accommodating cavity (11), and the liquid cooling and heat dissipation system comprises: A liquid cooling link (20), wherein the cooling liquid in the liquid cooling link (20) flows through the liquid cooling plate (14) at the central processing unit, the liquid cooling assembly (15) at the memory bar (12), and the graphics processing unit (13), so as to remove the heat generated by the central processing unit through the liquid cooling plate (14), remove the heat generated by the memory bar (12) through the liquid cooling assembly (15), and remove the heat generated by the graphics processing unit (13) through the cooling liquid flowing through the graphics processing unit (13).

2. The liquid-cooled server according to claim 1, wherein: There are multiple liquid cooling plates (14), and the multiple liquid cooling plates (14) are arranged in series on the liquid cooling link (20) in sequence, so that the multiple liquid cooling plates (14) can perform heat exchange with one or more central processing units; and / or, There are multiple liquid cooling components (15), and the multiple liquid cooling components (15) are arranged in series on the liquid cooling link (20) in an end-to-end manner, so that the multiple liquid cooling components (15) can perform heat exchange with one or more memory bars (12); and / or, There are a plurality of graphics processing units (13), and the plurality of graphics processing units (13) are arranged in parallel on the liquid cooling link (20).

3. The liquid-cooled server according to claim 1, wherein: The liquid cooling system further includes: a conversion device (30), the conversion device (30) being arranged on the liquid cooling link (20) and located downstream of the liquid cooling component (15), the conversion device (30) having a first flow channel (31) and a second flow channel (32); The first flow channel (31) has a main input interface (311) and a plurality of branch output interfaces (312), the main input interface (311) is connected to the outlet of the liquid cooling component (15), so that the cooling liquid in the liquid cooling component (15) is collected in the first flow channel (31), and the plurality of branch output interfaces (312) are respectively connected to the first liquid inlets corresponding to the plurality of graphics processing units (13); The second flow channel (32) has a main output interface (321) and a plurality of branch input interfaces (322), the main output interface (321) is connected to an external coolant recovery device, and the plurality of branch input interfaces (322) are respectively connected to the first liquid discharge ports corresponding to the plurality of graphics processing units (13).

4. The liquid-cooled server according to claim 3, wherein: The first flow channel (31) and the second flow channel (32) are arranged vertically and spaced apart in the height direction of the conversion device (30).

5. The liquid-cooled server according to claim 3, wherein: The conversion device (30) further comprises a cover plate (33), wherein there are two cover plates (33), and the two cover plates (33) are respectively arranged to cover the first flow channel (31) and the second flow channel (32).

6. The liquid-cooled server according to claim 1, wherein: The accommodating cavity (11) has a first accommodating area and a second accommodating area, the second accommodating area is located above the first accommodating area, the central processing unit, the memory bar (12) and part of the liquid cooling link (20) are located in the first accommodating area, and the graphics processing unit (13) and at least part of the remaining liquid cooling link (20) are located in the second accommodating area.

7. The liquid-cooled server according to claim 6, wherein: The liquid cooling system further includes: A first liquid leakage detection line (40) is located in the first accommodating area and is laid along an extension path of the liquid cooling link (20) located in the first accommodating area, so as to be configured to detect whether there is a cooling liquid leak in the liquid cooling link (20) located in the first accommodating area.

8. The liquid-cooled server according to claim 7, wherein: The liquid cooling system further includes: a liquid collection tray (50), the liquid collection tray (50) being located between the first accommodating area and the second accommodating area, and at least the liquid collection tray (50) being capable of receiving cooling liquid leaked from a connection between the liquid cooling link (20) and the graphics processing unit (13) located in the second accommodating area; A second liquid leakage detection line (60) is located in the liquid collection tray (50) and is configured to detect whether coolant is contained in the liquid collection tray (50).

9. The liquid-cooled server according to claim 8, wherein: The bottom receiving surface of the liquid collection tray (50) is inclined, and the bottom receiving surface corresponding to the connection is the lowest surface, and at least part of the second liquid leakage detection line (60) is laid on the lowest surface.

10. The liquid-cooled server according to claim 8, wherein: The first liquid leakage detection line (40) and the second liquid leakage detection line (60) are connected by plugging a detection line male connector (61) and a detection line female connector (41), wherein: One end of the first liquid leakage detection line (40) away from the second liquid leakage detection line (60) is connected to a control module signal to transmit the detected liquid leakage signal to the control module; or, One end of the second liquid leakage detection line (60) away from the first liquid leakage detection line (40) is connected to a control module signal to transmit the detected liquid leakage signal to the control module.

11. The liquid-cooled server according to claim 10, wherein: The first liquid leakage detection line (40) has a male end connector (42), and the male end connector (42) is connected to the control module signal.

12. The liquid-cooled server according to claim 8, wherein: The liquid accumulation tray (50) is provided with a buckle, and the buckle is configured to fix at least a portion of the liquid cooling link (20).

13. The liquid-cooled server according to any one of claims 1 to 12, characterized in that: The liquid cooling link (20) comprises a liquid inlet interface (21) and a liquid outlet interface (22); the cooling liquid flows into the liquid cooling link (20) through the liquid inlet interface (21) and flows into an external cooling liquid recovery device through the liquid outlet interface (22).

14. The liquid-cooled server according to any one of claims 1 to 12, characterized in that: The liquid cooling assembly (15) comprises: A liquid separation structure (151), the liquid separation structure (151) comprising a first flow cavity, a second liquid inlet (1511) communicating with the first flow cavity, and a plurality of first liquid separation ports (1512); A liquid collecting structure (152), the liquid collecting structure (152) comprising a second flow cavity, a second liquid discharge port (1521) communicating with the second flow cavity, and a plurality of second liquid separation ports; A plurality of sub-cold plate members (153) are arranged in parallel between the liquid separation structure (151) and the liquid collection structure (152), and the number of the sub-cold plate members (153), the number of the first liquid separation ports (1512), and the number of the second liquid separation ports are equal, so that two ends of each sub-cold plate member (153) are respectively connected to a corresponding first liquid separation port (1512) and a corresponding second liquid separation port.

15. The liquid-cooled server according to claim 14, wherein: The middle portion of each sub-cold plate (153) is concave, and both ends are bent.

16. The liquid-cooled server according to claim 15, wherein: The liquid cooling assembly (15) further comprises: A heat conducting plate (154) is provided at a middle recessed portion of the sub-cold plate (153), and the thickness of the heat conducting plate (154) is equal to the thickness of the sub-cold plate (153).

17. The liquid-cooled server according to claim 16, wherein: The heat conducting plate (154) has a first avoidance notch (1541), and the first avoidance notch (1541) is configured to avoid a chip (121) protruding from the memory bar (12).

18. The liquid-cooled server according to claim 17, wherein: A heat conduction pad (155) is provided on the sub-cold plate (153), and a second avoidance gap (1551) is provided at a position of the heat conduction pad (155) opposite to the first avoidance gap (1541).

19. The liquid-cooled server according to any one of claims 1 to 12, characterized in that: The liquid-cooled server further includes a support structure (70), wherein the support structure (70) is arranged in the accommodating cavity (11), and the liquid-cooling component (15) is arranged on the support structure (70), so that the support structure (70) provides support for the liquid-cooling component (15).

20. The liquid-cooled server according to claim 19, wherein: Positioning holes (100) and positioning guide pins (200) are respectively provided on the support structure (70) and the liquid cooling assembly (15) to provide positioning for the installation of the liquid cooling assembly (15).

Images