Server

Abstract

The invention provides a server, and relates to the field of heat exchange. The server comprises a box body; the mainboard is arranged in the box body; the plurality of GPUs are stacked on one side of the server and are electrically connected to the mainboard through the PCIE expansion card; the processors are electrically connected to the mainboard; a processor liquid cooling unit; and a GPU liquid cooling unit, wherein the processor liquid cooling unit and the GPU liquid cooling unit are arranged in an up-down layered manner. The server can have relatively high heat dissipation efficiency.

Description

Technical Field

[0001] The present specification relates to the field of heat exchange technology, and in particular to a server. Background Art

[0002] With the development of network technology, the computing demand of servers that carry applications and data storage has gradually increased, and the demand for heat dissipation has also gradually increased. Traditional air cooling can hardly meet the demand. Therefore, the application of liquid cooling has gradually increased.

[0003] In a server, the main heat generating devices are CPU (Central Processing Unit) and GPU (Graphics Processing Unit). With the improvement of computing power, the number of CPUs and GPUs in a server increases. The CPU is set on the motherboard, and multiple GPUs are first connected to the PCIE (Peripheral Component Interconnect Express) expansion card, and then connected to the motherboard through the PCIE expansion card. How to dissipate heat more effectively for more heat generating devices is a problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0004] In order to overcome the problems existing in the related art, this specification provides a server.

[0005] According to a first aspect of an embodiment of this specification, a server is provided, including:

[0006] Box;

[0007] A mainboard is arranged in the box;

[0008] A plurality of image processors (GPUs) are stacked on one side of the server and electrically connected to the mainboard via a high-speed peripheral component interconnect (PCIE) expansion card;

[0009] A plurality of processors, electrically connected to the mainboard;

[0010] Processor liquid cooling unit; and,

[0011] GPU liquid cooling unit;

[0012] The GPU liquid cooling unit includes a GPU cold plate, a first input pipeline connected to the GPU cold plate, a first cascade pipeline and a first output pipeline, the GPU cold plate wraps the GPU, the first input pipeline is branched to connect the GPU cold plates in the first group of GPUs, the first output pipeline is branched to connect the GPU cold plates in the second group of GPUs, the first cascade pipeline is connected to the GPU cold plates in the first group of GPUs and the GPU cold plates in the second group of GPUs corresponding to the positions of the GPUs in the first group of GPUs;

[0013] The processor liquid cooling unit includes a processor cold plate, a first pipe bracket, a second input pipe connected to the processor cold plate, a second cascade pipe and a second output pipe, the processor cold plate covers the processor, the second input pipe is connected to the processor cold plate of the first processor, the second output pipe is connected to the processor cold plate of the last processor, the second cascade pipe is sequentially connected between the processor cold plate of the first processor and the processor cold plate of the last processor, the first pipe bracket is arranged above the processor cold plate, the first pipe bracket arranged on at least one processor cold plate supports the first input pipe, and the first pipe bracket arranged on at least another processor cold plate supports the first output pipe.

[0014] Optionally, the first pipeline support includes:

[0015] A bottom plate connected to the processor cold plate;

[0016] a support arm extending from the bottom plate in a direction away from the processor, the support arm being formed with a receiving portion; and,

[0017] First strap;

[0018] The first input pipeline or the first output pipeline of the first pipeline support is fastened by the first binding strap.

[0019] Optionally, the processor cold plate comprises:

[0020] A plate body, wherein a cold liquid cavity is formed in the plate body;

[0021] A first liquid inlet connector and a first liquid outlet connector, connected to the cold liquid chamber;

[0022] Wherein, the plate body is fixed to the main board, the second input pipeline is connected to the first liquid inlet joint on the first processor cold plate, the second output pipeline is connected to the first liquid outlet joint on the last processor cold plate, and the second cascade pipeline connects the first processor cold plate to the last processor cold plate through the first liquid outlet joint of the first processor cold plate, the first liquid inlet joint and the first liquid outlet joint between the first processor cold plate and the last processor cold plate, and the first liquid inlet joint of the last processor cold plate.

[0023] Optionally, the processor cold plate further comprises:

[0024] A first liquid leakage locking frame is arranged on the outer peripheral side of the first liquid inlet connector and the first liquid outlet connector, and protrudes from the plate body to a side away from the processor;

[0025] The server further comprises:

[0026] The first liquid leakage detection device is electrically connected to the main board and is arranged at the edges of the first liquid inlet connector and the first liquid outlet connector in the first liquid leakage locking frame.

[0027] Optionally, the server further includes:

[0028] A second pipeline support connected to the bottom wall of the box;

[0029] Wherein, the clamping arm of the second pipeline support arranged away from the bottom wall clamps at least one of the second input pipeline, the second cascade pipeline and the second output pipeline.

[0030] Optionally, the server further includes:

[0031] A third pipeline support straddling two side walls of the box;

[0032] Wherein, the connection part of the third pipeline support is connected to two side walls, and a support rod is formed between the connection parts, and the support rod supports the first input pipeline and the first output pipeline.

[0033] Optionally, the third pipeline support further includes:

[0034] Second strap;

[0035] The second binding strap secures the first cascade pipeline.

[0036] Optionally, the PCIE expansion card includes:

[0037] A base, fixed to the bottom wall of the server;

[0038] A frame connected to the base;

[0039] Wherein, a second liquid leakage locking frame is formed on the base, and the second liquid leakage locking frame is located below the second liquid inlet joint and the second liquid outlet joint arranged on the GPU cold plate.

[0040] Optionally, also include:

[0041] The second liquid leakage detection device is electrically connected to the main board and is arranged in the second liquid leakage locking frame.

[0042] Optionally, the GPU liquid cooling unit includes an input manifold and an output manifold;

[0043] The first input pipeline includes a first front input pipeline and a first rear input pipeline, and the first output pipeline includes a first front output pipeline and a first rear output pipeline;

[0044] Among them, the two ends of the first rear input pipeline are respectively connected to the input manifold and the GPU cold plate, the two ends of the first rear output pipeline are respectively connected to the output manifold and the GPU cold plate, one end of the first front input pipeline is connected to the input manifold, and the other end is connected to the input end of the external cold source, and one end of the first front output pipeline is connected to the output manifold, and the other end is connected to the output end of the external cold source.

[0045] Optionally, the server further includes:

[0046] The liquid collection tray is arranged below the input manifold and the output manifold.

[0047] Optionally, the server further includes:

[0048] The third liquid leakage detection device is arranged at a position in the liquid accumulation tray corresponding to the cold liquid joints of the input manifold and the output manifold, and is electrically connected to the main board.

[0049] Optionally, a fan port and a power port are arranged side by side on one side of the box;

[0050] The first input pipe and the first output pipe extend out of the server from a fan port.

[0051] Optionally, a gap is formed between the fan port and the power port;

[0052] The second input pipe and the second output pipe extend out of the server from the gap.

[0053] The technical solutions provided by the embodiments of this specification may have the following beneficial effects:

[0054] In an embodiment of the present specification, in a server having a GPU and a processor, a GPU liquid cooling unit and a processor liquid cooling unit are respectively provided, and the two are arranged in upper and lower layers in the server, and a first input pipe and a first output pipe of the GPU liquid cooling unit located on the upper layer are supported by a first pipe bracket provided on the processor liquid cooling unit, so that independent liquid cooling units can be deployed in the server for heat dissipation for different heat-generating devices, thereby improving the heat dissipation effect of the server.

[0055] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the specification and, together with the description, serve to explain the principles of the specification.

[0057] Figure 1 It is a structural diagram of a server involved in an implementation mode of the present application;

[0058] Figure 2 It is a structural schematic diagram of a first pipe support in a processor liquid cooling unit involved in an embodiment of the present application;

[0059] Figure 3 It is a structural schematic diagram of a processor cold plate in a processor liquid cooling unit involved in an embodiment of the present application;

[0060] Figure 4 It is a structural schematic diagram of a part of the structure of a processor liquid cooling unit involved in an embodiment of the present application;

[0061] Figure 5 It is a structural schematic diagram of a second pipe support in a server involved in an implementation manner of the present application;

[0062] Figure 6 is a structural schematic diagram of a third pipe support in a server involved in an implementation manner of the present application;

[0063] Figure 7 It is a structural diagram of a PCIE expansion card in a server involved in an implementation manner of the present application;

[0064] Figure 8 It is a structural schematic diagram of part of the structure of a GPU liquid cooling unit involved in an implementation manner of the present application. DETAILED DESCRIPTION

[0065] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this specification. Instead, they are merely examples of devices and methods consistent with some aspects of this specification as detailed in the appended claims.

[0066] The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this specification. The singular forms "a", "the" and "the" used in this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0067] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this specification, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0068] The present application provides a server 100, such as Figure 1 As shown, including:

[0069] The box body 1 comprises four walls, namely a bottom wall 10, two side walls 11 and a top wall (not shown);

[0070] A main board 2 is disposed in the box body 1 and is fixedly connected to the bottom wall 10 of the box body 1;

[0071] A plurality of GPUs (not shown) are stacked on one side of the server 100 and electrically connected to the mainboard 2 via a PCIE expansion card 3;

[0072] A plurality of processors (not shown), electrically connected to the mainboard 2;

[0073] Processor liquid cooling units 4, wherein the number of processor liquid cooling units 4 may be consistent with the number of processors disposed on the motherboard 2, or may be less than the number of processors deployed on the motherboard 2 (i.e., processor liquid cooling units 4 are disposed for some processors on the motherboard 2, and other processors are configured with other heat dissipation methods, such as air cooling), and the number of processors disposed on the motherboard 2 is at least two; and,

[0074] GPU liquid cooling unit 5, several GPUs are grouped in the server 100, at least two groups of GPUs are provided, each group of GPUs is connected to a PCIE expansion card 3, the PCIE expansion card 3 is plugged into the motherboard 2, the PCIE expansion card 3 can be wrapped by a frame 30 on the outside, and fixed to the bottom wall 10 through a base 31;

[0075] The GPU liquid cooling unit 5 includes a GPU cold plate 50, a first input pipeline 51 connected to the GPU cold plate 50, a first cascade pipeline 52 and a first output pipeline 53, the GPU cold plate 50 wraps the GPU, the first input pipeline 51 is branched to connect the GPU cold plate 50A in the first group of GPUs, the first output pipeline 53 is branched to connect the GPU cold plate 50B in the second group of GPUs, the first cascade pipeline 52 connects the GPU cold plate 50A in the first group of GPUs and the GPU cold plate 50B in the second group of GPUs corresponding to the position of the GPU in the first group of GPUs;

[0076] The processor liquid cooling unit 4 includes a processor cold plate 40, a second input pipeline 41 connected to the processor cold plate, a second cascade pipeline 42, a second output pipeline 43 and a first pipeline support 44, the processor cold plate 40 covers the processor, the second input pipeline 41 is connected to the processor cold plate 40A of the first processor, the second output pipeline 43 is connected to the processor cold plate 40B of the last processor, the second cascade pipeline 42 is sequentially connected between the processor cold plate 40A of the first processor and the processor cold plate 40B of the last processor, the first pipeline support 44 is arranged above the processor cold plate 40, the first pipeline support 44A arranged on at least one processor cold plate 40 (such as the processor cold plate 40A) supports the first input pipeline 51, and the first pipeline support 44B arranged on at least another processor cold plate 40 (such as the processor cold plate 40B) supports the first output pipeline 53.

[0077] The GPU cold plate 50 of the GPU liquid cooling unit 5 can be fixed to the frame 30 of the PCIE expansion card 3. The GPU cold plate 50 is provided with a second liquid inlet connector 50A and a second liquid outlet connector 50B. The second liquid inlet connector 50A is connected to the first input pipeline 51 of the GPU liquid cooling unit 5, and the second liquid outlet connector 50B is connected to the first output pipeline 53 of the GPU liquid cooling unit 5.

[0078] Since the GPU needs to be connected to the motherboard 2 through the PCIE expansion card 3, the GPU can be grouped according to the different PCIE expansion cards 3 connected to it. The GPU is divided into two groups, one group of GPUs is connected to the PCIE expansion card 3A, and the other group of GPUs is connected to the PCIE expansion card 3B. The figure is only shown as an example. If the size requirement can be met, more PCIE expansion cards 3 can be configured and more GPUs can be plugged in.

[0079] In the server 100, since the processor is installed on the motherboard 2 and the GPU is stacked above the motherboard 2 through the PCIE expansion card 3, the processor liquid cooling unit 4 is arranged as a whole above the GPU liquid cooling unit 5. The liquid cooling devices for different heat-generating components in the server 100 can be arranged in layers, so that the main heat-generating components in the server 100 can be effectively liquid-cooled and dissipated, thereby improving the heat dissipation efficiency of the server.

[0080] Various pipelines of the GPU liquid cooling unit 5 and the processor liquid cooling unit 4 need to be led out from the server 100. In order to avoid the occupation of the server 100 space by the pipelines, a fan port 12 and a power port 13 are optionally arranged side by side on one side of the box body 1. The fan port 12 is used to set a fan for air cooling, and the power port 13 is used to plug in a power supply.

[0081] The first input pipe 51 and the first output pipe 53 extend out of the server 100 from a fan port 12 .

[0082] Since the main heat-generating components in the server 100 dissipate heat through liquid cooling, the use of some fans can be reduced accordingly. Therefore, the first input pipeline 51 and the first output pipeline 53 can extend out by occupying one fan port 12.

[0083] In addition, if space permits, the second input pipeline 41 and the second output pipeline 43 may also extend through the same fan port 12, or occupy another fan port 12 to extend, and it can be set according to actual conditions without any limitation.

[0084] Optionally, a gap 15 is formed between the fan port 12 and the power port 13;

[0085] The second inlet line 41 and the second outlet line 43 extend out of the server 100 from the gap 15 .

[0086] A gap 15 may be reserved between the fan port 12 and the power port 13. The gap 15 may be set to a smaller size to accommodate a pipe. The second input pipe 41 and the second output pipe 43 may be arranged side by side in the longitudinal direction to lead the server 100 out of the gap 15.

[0087] Through the above arrangement, it is possible to avoid the outlet pipes from occupying too much air cooling space, thereby further improving the heat dissipation efficiency of the server.

[0088] If space permits, the first input pipeline 51 and the first output pipeline 53 may also be led out through the gap 15 , and may be arranged according to actual needs without limitation.

[0089] In addition, in some cases, the first input pipe 51 and the first output pipe 53 can be led out from a fan port 12, and the second input pipe 41 and the second output pipe 43 can be led out from the gap 15, so as to more reasonably utilize the outlet space of the cabinet 1 of the server 100, thereby further improving the heat dissipation effect of the server.

[0090] Furthermore, the leading out of the first input pipeline 51 and the first output pipeline 53 can also be achieved by separately setting up a pipeline outlet, which is located between two fan ports 12, or between the fan port 12 and the power port 13, etc., and can be set according to actual needs without limitation.

[0091] Since the pipeline has a certain span in the server 100, and since the cold liquid in the pipeline has a certain weight, in order to prevent the pipeline from sinking in the server 100 due to weight, it is optional, such as Figure 1 , 2 As shown, the first pipeline support 44 includes:

[0092] The bottom plate 440 is connected to the processor cold plate 50, which can be realized by clamping, screwing, riveting, etc.;

[0093] A support arm 441 extends from the bottom plate 440 in a direction away from the processor, and the support arm 441 is formed with a receiving portion 442. The support arm 441 may be formed in a nearly U-shape with an opening facing upward, and the receiving portion 442 is a space surrounded by the support arm 441; and

[0094] The first binding strap 443 may be integrally provided with the first pipe support 44, i.e., formed by a sheet metal structure protruding outward from the bottom plate 440, or may be separately provided with the first pipe support 44, i.e., the first binding strap 443 is provided separately, and a first through hole 444 is formed on the bottom plate 440, and the first binding strap 443 is provided through the first through hole 444;

[0095] The first input pipeline 51 or the first output pipeline 53 of the first pipeline support 44 is fastened by the first binding belt 443 .

[0096] After the first input pipeline 51 and the first output pipeline 53 are deployed, the support arm 441 supports the first input pipeline 51 or the first output pipeline 53. Specifically, according to different positions, such as Figure 1 As shown, if the first input pipeline 51 passes through the processor cold plate 40A of the processor liquid cooling unit 4, the first input pipeline 51 is fastened by the first strap 443 on the first pipeline bracket 44A; if the first output pipeline 53 passes through the processor cold plate 40B of the processor liquid cooling unit 4, the first output pipeline 53 is fastened by the first strap 443 on the first pipeline bracket 44B.

[0097] That is, at least two processors and corresponding processor liquid cooling units 4 may be provided in the server 100, the first pipe bracket 44A on one processor liquid cooling unit 4 fixes the first input pipe 51, and the first pipe bracket 44B on the other processor liquid cooling unit 4 fixes the first output pipe 53. If the server 100 includes more processors and corresponding processor cold plates, the processor cold plate may not be provided with brackets, and may be provided according to actual needs.

[0098] Specifically, if Figure 3 , 4 As shown, the processor cold plate 40 includes:

[0099] A plate body 400, wherein a cold liquid cavity (not shown) is formed in the plate body 400;

[0100] The first liquid inlet joint 401 and the first liquid outlet joint 402 are connected to the cold liquid cavity. The processor cold plate 40 is a sealed structure and is connected to the outside through the first liquid inlet joint 401 and the first liquid outlet joint 402. Cold liquid can flow into the cold liquid cavity through the first liquid inlet joint 401 and flow out of the cold liquid cavity from the first liquid outlet joint 402.

[0101] In which, the plate body 400 is fixed to the main board 2, the second input pipeline 41 is connected to the first liquid inlet connector 401A on the first processor cold plate 40A, the second output pipeline 43 is connected to the first liquid outlet connector 402A on the last processor cold plate 40B, and the second cascade pipeline 42 connects the first processor cold plate 40A to the last processor cold plate 40B through the first liquid outlet connector 401B of the first processor cold plate 40A, the first liquid inlet connector 401 and the first liquid outlet connector 402 between the first processor cold plate 40A and the last processor cold plate 40B, and the first liquid inlet connector 401B of the last processor cold plate 40B.

[0102] by Figure 3 , 4Take the server 100 shown as an example, which includes two processors and two corresponding processor liquid cooling units 4. The processor cold plate 40A and the processor cold plate 40B in the two processor liquid cooling units 4 are respectively provided with a first liquid inlet joint 401A, a first liquid outlet joint 402A, a first liquid inlet joint 401B and a first liquid outlet joint 402B. Among them, the processor cold plate 40A is the first processor cold plate, and the processor cold plate 40B is the last processor cold plate. It should be noted that Figure 3 , 4 This is only an exemplary diagram. Multiple processor cold plates can be arranged between the processor cold plate 40A and the processor cold plate 40B. The specific number can be set according to actual needs. The first processor cold plate and the last processor cold plate are only arranged in sequence. It can be understood that the processor cold plate connected to the second input pipeline 41 is the first processor cold plate, and the processor cold plate connected to the second output pipeline 43 is the last processor cold plate. The processor cold plate between the first processor cold plate and the last processor cold plate can be called an intermediate processor cold plate. The first liquid inlet joint and the first liquid outlet joint of the intermediate processor cold plate are connected to the second cascade pipeline 42 to form a cold liquid passage from the first processor cold plate, the intermediate processor cold plate to the last processor cold plate.

[0103] When the cold liquid flows into the server 100 through the second input pipeline 41, the cold liquid starts from the first processor cold plate, flows through the middle processor cold plate to the last processor cold plate, and finally flows out of the server 100 through the second output pipeline 43. If the server 100 includes only two processor cold plates, it can be understood that there is no middle processor cold plate, and the first processor cold plate and the last processor cold plate are directly connected.

[0104] Through the above arrangement, a cascade is formed between the processor cold plates (ie, the processor liquid cooling units), so that the inflowing cold liquid can dissipate heat for multiple processors in turn.

[0105] Optional, such as Figure 3 As shown, the processor cold plate 40 also includes:

[0106] A first liquid leakage locking frame 403 is disposed on the outer peripheral side of the first liquid inlet connector 401 and the first liquid outlet connector 402 and protrudes from the plate body 400 to a side away from the processor;

[0107] The server 100 further includes:

[0108] The first liquid leakage detection device 60 is electrically connected to the main board 2 and is arranged at the edges of the first liquid inlet connector 401 and the first liquid outlet connector 402 in the first liquid leakage locking frame 403 .

[0109] In a liquid-cooled server, leakage generally occurs at the connection between pipes and connectors. For example, in the processor liquid cooling unit 4, leakage may occur between the first liquid inlet connector 401 and the second input pipe 41 and between the first liquid outlet connector 402 and the second output pipe 43. Therefore, it is necessary to prevent the leakage from dripping onto the mainboard 2 and to implement detection and alarm for the leakage.

[0110] In order to prevent the leakage from dripping, a first leakage locking frame 403 needs to be formed on the plate body 400. The first leakage locking frame 403 can be integrally formed with the plate body 400, or can be separately formed and then fixed to the plate body 400 by welding, sealing connection, etc. It can be set according to actual needs without limitation.

[0111] When liquid leakage occurs at the connection between the joint and the pipeline, since the first liquid leakage locking frame 403 protrudes upward from the plate body 400 , a small amount of leaked liquid is blocked by the first liquid leakage locking frame 403 and will not drip onto the main board 2 .

[0112] In order to detect liquid leakage, a first liquid leakage detection device 60 needs to be provided, and the first liquid leakage detection device 60 can be a liquid leakage detection rope, etc. In order to detect liquid leakage quickly, the first liquid leakage detection device 60 can be deployed at the edge of the first liquid inlet joint 401 and the first liquid outlet joint 402, close to the two. In this way, when cold liquid seeps out, the leakage can be quickly confirmed, and the location of the leakage can be reported to the staff to achieve a rapid alarm.

[0113] Optionally, the server 100, such as Figure 1 , 5 As shown, it also includes:

[0114] The second pipe support 70 is connected to the bottom wall 10 of the box body 1 and can be fixed by screw connection or clamping;

[0115] The clamping arm 70A of the second pipeline support 70 disposed away from the bottom wall 10 clamps at least one of the second input pipeline 41 , the second cascade pipeline 42 and the second output pipeline 43 .

[0116] Since the laying distance of the pipeline in the server 100 may be long and it is easy to be deformed, the second pipeline support 70 is arranged at an appropriate position, such as at Figure 1 , 5The second pipe support 70 is arranged at a position close to the fan port 12, and supports the second input pipe 41 and the second cascade pipe 42 through the second pipe support 70. It should be noted that since the processor liquid cooling unit 4 and the GPU liquid cooling unit 5 are arranged in layers in the server 100, and the second input pipe 41, the second cascade pipe 42 and the second output pipe 43 of the processor liquid cooling unit 4 are located in the lower layer, closer to the bottom wall 10, therefore, the second pipe support 70 arranged on the bottom wall 10 preferably supports the second input pipe 41, the second cascade pipe 42 and the second output pipe 43. In addition to its supporting effect, it can also prevent the second pipe support 70 from occupying more longitudinal space of the server 100, thereby improving the space utilization rate in the server.

[0117] Optionally, the server 100, such as Figure 1 , 6 As shown, it also includes:

[0118] A third pipeline support 71 is connected across the two side walls 11 of the box body 1;

[0119] The connection portion 71A of the third pipeline support 71 is connected to the two side walls 11 , and a support rod 71B is formed between the connection portions 71A. The support rod 71B supports the first input pipeline 51 and the first output pipeline 53 .

[0120] Since the first group of GPUs and the second group of GPUs are connected through the first cascade pipe 52, which has a certain span, and as the number of GPUs in a group increases, the first cascade pipe 52 may sink or even become loose from the joint due to the pressure between them. In order to avoid the above situation, a third pipe bracket 71 may be arranged at the upper position of the server 100, and the third pipe bracket 71 may be fixed by the two side walls 11 of the box body 1, such as by clamping or screwing.

[0121] After the first cascade pipeline 52 is deployed, it can be supported by the third pipeline support 71 to prevent the first cascade pipeline 52 from sinking.

[0122] Furthermore, in order to enable the third pipeline support 71 to more reliably support the first cascade pipeline 52, the third pipeline support 71 further includes:

[0123] Second strap 71C;

[0124] The second binding band 71C binds the first cascade pipeline 52 .

[0125] Similar to the first strap, the second strap 71C can also be integrally formed with the third pipe bracket 71 or separately formed. When separately formed, a second through hole 71D can be formed on the support rod 71B of the third pipe bracket 71. The second strap 71C passes through the second through hole 71D to connect with the support rod 71B, and locks the first cascade pipeline 52 on the support rod 71B, so that the third pipe bracket 71 can more reliably support the first cascade pipeline 52.

[0126] Optionally, the PCIE expansion card 3, such as Figure 1 , 7 As shown, including:

[0127] A base 31 fixed to the bottom wall 10 of the server 100;

[0128] The frame 30 is connected to the base 31;

[0129] A second liquid leakage locking frame 32 is formed on the base 31 , and the second liquid leakage locking frame 32 is located below a second liquid inlet joint 500 and a second liquid outlet joint 501 provided on the GPU cold plate 50 .

[0130] On the side of the PCIE expansion card 3 where the GPU is set, a joint is also set on the GPU cold plate 50, namely, a second liquid inlet joint 500 and a second liquid outlet joint 501. In order to prevent the second liquid inlet joint 500 and the second liquid outlet joint 501 from being loosely connected to the pipeline, causing the cold liquid to drip onto the bottom wall 10 and flow to the mainboard 2 and other electrical components, a second liquid leakage locking frame 32 is set on the base 31 below the second liquid inlet joint 500 and the second liquid outlet joint 501. The second liquid leakage locking frame 32 is protruded from the base 31 to the side away from the bottom wall 10. When part of the cold liquid drips, the cold liquid can be locked in the second liquid leakage locking frame 32 to prevent the dripping cold liquid from flowing to the mainboard 2, thereby avoiding damage to the server caused by leakage to a certain extent, and improving the reliability of the server with liquid cooling.

[0131] Furthermore, in order to reduce the possibility of excessive leakage overflowing the second leakage locking frame 32 and promptly remind the staff to conduct abnormality inspection, the server 100 also includes:

[0132] The second liquid leakage detection device 61 is electrically connected to the main board 2 and is disposed in the second liquid leakage locking frame 32 .

[0133] Similar to the first liquid leakage detection device 60, the second liquid leakage detection device 61 may also be a liquid leakage detection rope, etc. By providing the second liquid leakage detection device 61, the server 100 can prompt the server 100 to issue an alarm when a fault is detected in the second liquid inlet connector 500 and the second liquid outlet connector 501, so as to remind the staff to conduct abnormality inspection and reduce the possibility of damage to the server.

[0134] Optionally, the GPU liquid cooling unit 5 comprises an input manifold 54 and an output manifold 55, and the manifold is used to realize the branching transportation of the cooling liquid;

[0135] The first input pipeline 51 includes a first front input pipeline 51A and a first rear input pipeline 51B, and the first output pipeline 53 includes a first front output pipeline 53A and a first rear output pipeline 53B, wherein the first front input pipeline 51A and the first rear input pipeline 51B are distinguished by the input manifold 54, that is, the first input pipeline 51 before the external connection to the input manifold 54 should be understood as the first front input pipeline 51A, and the first output pipeline 53A from the input manifold 54 to the GPU cold plate 5 0 should be understood as the first rear input pipeline 51B, and correspondingly, the first front output pipeline 53A and the first rear output pipeline 53B are also distinguished by the output manifold 55, that is, the first output pipeline 53 from the output manifold 55 to the outside of the server 100 should be understood as the first front output pipeline 53A, and the first output pipeline 53 from the output manifold 55 to the GPU cold plate 50 should be understood as the first rear output pipeline 53B;

[0136] Among them, the two ends of the first rear input pipeline 51B are respectively connected to the input manifold 54 and the GPU cold plate 50, the two ends of the first rear output pipeline 53B are respectively connected to the output manifold 55 and the GPU cold plate 50, one end of the first front input pipeline 51A is connected to the input manifold 54, and the other end is connected to the input end of the external cold source, and one end of the first front output pipeline 53A is connected to the output manifold 55, and the other end is connected to the output end of the external cold source.

[0137] The external cooling source may be provided by a CDU (Cooling Distribution Units) or the like.

[0138] Cold liquid can be provided to multiple GPU cold plates 50 through the input manifold 54 and recycled through the output manifold 55, so that multiple GPUs in a group can receive cold liquid with lower temperature to dissipate heat, thereby improving the heat dissipation effect of the server 100.

[0139] Optionally, the server 100, such as Figure 1 , 8 As shown, it also includes:

[0140] The liquid collection tray 8 is arranged below the input water collector 54 and the output water collector 55 .

[0141] Since the input water collecting and distributing device 54 and the output water collecting and distributing device 55 also need to be connected to the first input pipeline 51 and the first output pipeline 53 through joints, leakage may also occur. Therefore, a liquid collection tray 8 is set under the input water collecting and distributing device 54 and the output water collecting and distributing device 55 to collect the leaked liquid to avoid damage caused by the leaked liquid dripping onto the main board 2.

[0142] Furthermore, the server 100 further includes:

[0143] The third liquid leakage detection device 62 is arranged at a position in the liquid collection tray 8 corresponding to the cold liquid joints of the input manifold 54 and the output manifold 55 , and is electrically connected to the main board 2 .

[0144] Similar to the first leakage detection device 60 and the second leakage detection device 61 , a third leakage detection device 62 may be provided in the liquid collection tray 8 to detect leakage, and realize detection and alarm by connecting to the mainboard 2 .

[0145] The technical solutions provided by the embodiments of this specification may have the following beneficial effects:

[0146] In an embodiment of the present specification, in a server having a GPU and a processor, a GPU liquid cooling unit and a processor liquid cooling unit are respectively provided, and the two are arranged in upper and lower layers in the server, and a first input pipe and a first output pipe of the GPU liquid cooling unit located on the upper layer are supported by a first pipe bracket provided on the processor liquid cooling unit, so that independent liquid cooling units can be deployed in the server for heat dissipation for different heat-generating devices, thereby improving the heat dissipation effect of the server.

[0147] Those skilled in the art will readily appreciate other embodiments of the specification after considering the specification and practicing the invention claimed herein. The specification is intended to cover any variations, uses or adaptations of the specification that follow the general principles of the specification and include common knowledge or customary techniques in the art that are not claimed in the specification. The specification and examples are to be considered exemplary only, and the true scope and spirit of the specification are indicated by the following claims.

[0148] It should be understood that the present description is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present description is limited only by the appended claims.

[0149] The above description is only a preferred embodiment of this specification and is not intended to limit this specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of this specification should be included in the scope of protection of this specification.

Claims

1. A server, characterized in that: include: Box; A mainboard is arranged in the box; A plurality of image processors (GPUs) are stacked on one side of the server and electrically connected to the mainboard via a high-speed peripheral component interconnect (PCIE) expansion card; A plurality of processors, electrically connected to the mainboard; Processor liquid cooling unit; and, GPU liquid cooling unit; The GPU liquid cooling unit includes a GPU cold plate, a first input pipeline connected to the GPU cold plate, a first cascade pipeline and a first output pipeline, the GPU cold plate wraps the GPU, the first input pipeline is branched to connect the GPU cold plates in the first group of GPUs, the first output pipeline is branched to connect the GPU cold plates in the second group of GPUs, the first cascade pipeline is connected to the GPU cold plates in the first group of GPUs and the GPU cold plates in the second group of GPUs corresponding to the positions of the GPUs in the first group of GPUs; The processor liquid cooling unit includes a processor cold plate, a first pipe bracket, a second input pipe connected to the processor cold plate, a second cascade pipe and a second output pipe, the processor cold plate covers the processor, the second input pipe is connected to the processor cold plate of the first processor, the second output pipe is connected to the processor cold plate of the last processor, the second cascade pipe is sequentially connected between the processor cold plate of the first processor and the processor cold plate of the last processor, the first pipe bracket is arranged above the processor cold plate, the first pipe bracket arranged on at least one processor cold plate supports the first input pipe, and the first pipe bracket arranged on at least another processor cold plate supports the first output pipe.

2. The server according to claim 1, characterized in that: The first pipeline support comprises: A bottom plate connected to the processor cold plate; a support arm extending from the bottom plate in a direction away from the processor, the support arm being formed with a receiving portion; and, First strap; The first input pipeline or the first output pipeline of the first pipeline support is fastened by the first binding strap.

3. The server according to claim 2, characterized in that: The processor cold plate comprises: A plate body, wherein a cold liquid cavity is formed in the plate body; A first liquid inlet connector and a first liquid outlet connector, connected to the cold liquid chamber; Wherein, the plate body is fixed to the main board, the second input pipeline is connected to the first liquid inlet joint on the first processor cold plate, the second output pipeline is connected to the first liquid outlet joint on the last processor cold plate, and the second cascade pipeline connects the first processor cold plate to the last processor cold plate through the first liquid outlet joint of the first processor cold plate, the first liquid inlet joint and the first liquid outlet joint between the first processor cold plate and the last processor cold plate, and the first liquid inlet joint of the last processor cold plate.

4. The server according to claim 3, characterized in that: The processor cold plate also includes: A first liquid leakage locking frame is arranged on the outer peripheral side of the first liquid inlet connector and the first liquid outlet connector, and protrudes from the plate body to a side away from the processor; The server further comprises: The first liquid leakage detection device is electrically connected to the main board and is arranged at the edges of the first liquid inlet connector and the first liquid outlet connector in the first liquid leakage locking frame.

5. The server according to claim 1, characterized in that: Also includes: A second pipeline support connected to the bottom wall of the box; Wherein, the clamping arm of the second pipeline support arranged away from the bottom wall clamps at least one of the second input pipeline, the second cascade pipeline and the second output pipeline.

6. The server according to claim 1, characterized in that: Also includes: A third pipeline support straddling two side walls of the box; Wherein, the connection part of the third pipeline support is connected to two side walls, and a support rod is formed between the connection parts, and the support rod supports the first input pipeline and the first output pipeline.

7. The server according to claim 6, characterized in that: The third pipeline support further includes: Second strap; The second binding strap secures the first cascade pipeline.

8. The server according to claim 1, characterized in that: The PCIE expansion card includes: A base, fixed to the bottom wall of the server; A frame connected to the base; Wherein, a second liquid leakage locking frame is formed on the base, and the second liquid leakage locking frame is located below a second liquid inlet joint and a second liquid outlet joint arranged on the GPU cold plate.

9. The server according to claim 8, characterized in that: Also includes: The second liquid leakage detection device is electrically connected to the main board and is arranged in the second liquid leakage locking frame.

10. The server according to claim 1, characterized in that: The GPU liquid cooling unit comprises an input water collector and an output water collector; The first input pipeline includes a first front input pipeline and a first rear input pipeline, and the first output pipeline includes a first front output pipeline and a first rear output pipeline; Among them, the two ends of the first rear input pipeline are respectively connected to the input manifold and the GPU cold plate, the two ends of the first rear output pipeline are respectively connected to the output manifold and the GPU cold plate, one end of the first front input pipeline is connected to the input manifold, and the other end is connected to the input end of the external cold source, and one end of the first front output pipeline is connected to the output manifold, and the other end is connected to the output end of the external cold source.

11. The server according to claim 10, characterized in that: Also includes: The liquid collection tray is arranged below the input manifold and the output manifold.

12. The server according to claim 11, characterized in that Also includes: The third liquid leakage detection device is arranged at a position in the liquid accumulation tray corresponding to the cold liquid joints of the input manifold and the output manifold, and is electrically connected to the main board.

13. The server according to claim 1, characterized in that: A fan port and a power port are arranged side by side on one side of the box; The first input pipe and the first output pipe extend out of the server from a fan port.

14. The server according to claim 13, characterized in that: A gap is formed between the fan port and the power port; The second input pipe and the second output pipe extend out of the server from the gap.

Images