Immersion liquid cooling server, matched chip heat sink and heat sink optimization design method

Abstract

The embodiment of the invention provides an immersion liquid cooling server, a matched chip heat sink and a heat sink optimization design method, and belongs to the technical field of data center temperature control. The matched chip heat sink comprises a heat sink upper cover plate and a heat sink core body, the heat sink upper cover plate is used for uniformly applying pressure to the upper surface of the heat sink core body through the spring bolts, so that the pressure is uniformly transmitted to the surface of the chip; the heat sink core body comprises a soft cover plate and a substrate fin type metal body; the soft cover plate and the substrate fin type metal body are locked and fixed through bolts; the heat sink core body is used for heat exchange through cooling liquid flowing and absorbing heat generated by the chip. The matched chip heat sink can effectively solve the problem that a traditional brazing type heat sink process is high in cost.

Description

Technical Field

[0001] The present application relates to the field of temperature control technology for data centers, and in particular to an immersion liquid cooling server, a matching chip heat sink, and a heat sink optimization design method. Background Art

[0002] Immersion liquid cooling offers excellent energy-saving, silent, and safe operation, and has significant potential for efficient temperature control in data centers. However, because servers are completely immersed in coolant, immersion cooling has inherent drawbacks. These include the small size of IT equipment's structural components and electronic devices, which results in high liquid consumption; and low flow rates, which are insufficient for high-power chips. These drawbacks significantly impact the cost-effectiveness and market penetration of immersion cooling, particularly for the new generation of high-power AI chips, for which traditional immersion cooling performance is no longer sufficient.

[0003] The heat sink is a key component of the temperature control system and the primary carrier for chip heat dissipation. Its performance plays a crucial role in determining the performance and cost of the chip temperature control system. Currently, most commonly used immersion-cooled heat sinks use the same design as air-cooled heat sinks. Due to the low liquid flow rate within immersion-cooled servers and the large cross-sectional area of ​​the servers, the liquid cannot centrally cool key heat sources such as chips, resulting in low heat exchange efficiency. Current immersion cooling is primarily targeted at low- to medium-power chips under 300 watts. For high-power chips over 300 watts, immersion cooling is rarely used due to the limited heat exchange performance of traditional immersion heat sinks. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to provide an immersion liquid cooling server, a matching chip heat sink, and a heat sink optimization design method.

[0005] The technical solution adopted by the present invention is:

[0006] In one aspect, an embodiment of the present invention provides a supporting chip heat sink, the supporting chip heat sink comprising a heat sink upper cover plate and a heat sink core;

[0007] The heat sink upper cover is used to uniformly apply pressure to the upper surface of the heat sink core through spring bolts, thereby uniformly transferring the pressure to the surface of the chip;

[0008] The heat sink core comprises a soft cover plate and a base plate fin type metal body; the soft cover plate and the base plate fin type metal body are locked by bolts;

[0009] The heat sink core is used to absorb the heat generated by the chip through the flow of cooling liquid for heat exchange.

[0010] Furthermore, the heat sink upper cover is made of a rigid material;

[0011] The heat sink upper cover plate has a first strip-shaped inlet;

[0012] The soft cover has a second strip-shaped entrance;

[0013] The first strip-shaped inlet and the second strip-shaped inlet are used to provide an inlet for the cooling liquid to enter the substrate fin-type metal body for heat exchange.

[0014] Furthermore, the soft cover is made of elastic material;

[0015] The soft cover plate generates elastic deformation under pressure to evenly distribute the pressure.

[0016] Furthermore, the substrate-fin type metal body comprises fins and a substrate;

[0017] The fins are used to increase the surface area;

[0018] The fin includes a cut-off structure;

[0019] The base plate includes a plurality of first grooves and a plurality of second grooves; the first grooves are parallel to the cutting structures of the fins; and the second grooves are parallel to the fins.

[0020] Furthermore, the heat sink upper cover plate is a semi-enclosed shell structure.

[0021] On the other hand, an embodiment of the present invention provides a heat sink optimization design method, which is used to design the aforementioned supporting chip heat sink, including the following steps:

[0022] Obtaining design parameters of the heat sink structure; the design parameters include fin height, fin thickness and fin spacing;

[0023] According to the design parameters, obtaining a design parameter combination;

[0024] Set several flow values;

[0025] Acquiring temperature rise data and flow resistance data according to the design parameter combination and the flow value;

[0026] training a neural network for heat sink design based on the temperature rise data and the flow resistance data;

[0027] Heat sink design data is obtained according to the neural network of the heat sink design.

[0028] Furthermore, obtaining temperature rise data and flow resistance data according to the design parameter combination and the flow value includes the following steps:

[0029] For each of the design parameter combinations, simulation or testing is performed at several flow rate values ​​to obtain temperature rise data and flow resistance data for each of the design parameter combinations.

[0030] Furthermore, the training of a neural network for heat sink design based on the temperature rise data and the flow resistance data comprises the following steps:

[0031] Obtaining the temperature rise and flow resistance required for model training according to the temperature rise data and the flow resistance data;

[0032] Setting input parameters; the input parameters include the fin height, the fin thickness, the fin spacing, and the flow resistance;

[0033] Setting output parameters; the output parameters include the flow value and the temperature rise;

[0034] A neural network for heat sink design is obtained according to the input parameters and the output parameters.

[0035] On the other hand, an embodiment of the present invention provides an immersion liquid cooling server, the immersion liquid cooling server comprising a server cover plate, a liquid block, and the aforementioned supporting chip heat sink;

[0036] The server cover is used to cover the top of the server and protect the internal components of the server;

[0037] The liquid-occupying block is used to cover the entire mainboard area and maintain a preset distance from the mainboard to prevent interference with electronic devices;

[0038] The matching chip heat sink is used for chip heat dissipation.

[0039] Furthermore, the liquid-occupying block includes a groove structure and a slope surface structure; the distance between the bottom surface of the liquid-occupying block and the main board is several millimeters to more than ten millimeters;

[0040] The groove structure and the slope surface structure are connected;

[0041] The groove structure corresponds to the raised components on the motherboard; the raised components include chips and memory modules;

[0042] The groove structure and the slope surface structure are used to reduce the flow resistance of the coolant.

[0043] The embodiments of the present application include at least the following beneficial effects: The present application provides an immersion liquid cooling server, a matching chip heat sink, and a heat sink optimization design method. The matching chip heat sink of the present invention includes a heat sink upper cover plate and a heat sink core; the heat sink upper cover plate is used to uniformly apply pressure to the upper surface of the heat sink core through spring bolts, thereby uniformly transferring the pressure to the surface of the chip; the heat sink core includes a soft cover plate and a base plate fin-type metal body; the soft cover plate and the base plate fin-type metal body are locked by bolts; the heat sink core is used to absorb the heat generated by the chip through the flow of coolant for heat exchange. The matching chip heat sink of the present invention effectively avoids the high cost problem of traditional brazing heat sink processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a schematic diagram of a supporting chip heat sink provided by an embodiment of the present invention;

[0045] Figure 2 is an exploded schematic diagram of a heat sink core provided by an embodiment of the present invention;

[0046] Figure 3 is a schematic diagram of a heat sink upper cover plate provided by an embodiment of the present invention;

[0047] Figure 4 is a schematic diagram of a substrate fin-type metal body provided by an embodiment of the present invention;

[0048] Figure 5 is a schematic diagram of a heat sink optimization design method provided by an embodiment of the present invention;

[0049] Figure 6 This is a schematic diagram of the overall structure of the immersion liquid cooling server provided by an embodiment of the present invention;

[0050] Figure 7 This is a schematic diagram of the bottom of the liquid block provided by an embodiment of the present invention;

[0051] Figure 8 1 is a schematic cross-sectional view of the liquid-occupying block along the chip provided by an embodiment of the present invention;

[0052] Figure 9 This is a schematic cross-sectional view of the liquid occupying block along the memory bar provided by an embodiment of the present invention;

[0053] Figure 10 is a schematic diagram of a coolant flow path provided by an embodiment of the present invention;

[0054] Figure 11 This is a flow chart of a heat sink performance optimization method provided by an embodiment of the present invention with flow resistance as a limiting value and minimum temperature rise as a goal;

[0055] Figure 12 This is a flow chart of a heat sink performance optimization method provided by an embodiment of the present invention, with temperature rise as a limit value and minimum flow resistance as a goal.

[0056] Figure markings: 110-heat sink upper cover, 120-heat sink core, 200-chip, 121-soft cover, 1211-second strip inlet, 122-substrate fin-type metal body, 1221-fin, 123-bolt, 111-first strip inlet, 1222-substrate, 1221a-cutting structure, 1222a-first groove, 1222b-second groove, 310-server upper cover, 320-liquid block, 100-matching chip heat sink, 3211-groove above the chip heat sink, 3212-groove above the memory module, 322-slope surface structure, 323-edge inclined slope, 10A-path for the coolant to flow inside the shell, 10B-path for the coolant to be discharged from the heat sink upper cover. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are merely examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0058] It will be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, 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 words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0059] The terms "at least one", "plurality", "each", "any", etc. used in this application include "at least one", "two" or more, "plurality" or "each", "any" or "any one", "each" or "any one" as used herein.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0061] The new liquid-cooled server of the present invention can effectively save liquid and avoid a significant increase in flow resistance. The immersion-type liquid-cooled heat sink that comes with the new liquid-cooled server focuses on cooling high-power chips, and has a low process cost. The proposed liquid-cooled server and heat sink design method can eliminate the main defects of traditional immersion liquid cooling technology, and is of great value in applying immersion liquid cooling technology to a new generation of high-power GPUs and promoting energy conservation in AI computing centers. At the same time, a rapid optimization method based on measured data or simulation data is proposed for this type of heat sink, which is of great value in optimizing the flow resistance of the heat sink and improving the heat exchange performance of the heat sink.

[0062] The embodiments of the present invention are further described below with reference to the accompanying drawings.

[0063] On the one hand, the embodiment of the present invention provides a supporting chip heat sink 100, referring to Figure 1 and Figure 2 , including a heat sink upper cover plate 110 and a heat sink core 120;

[0064] The heat sink upper cover plate 110 is used to uniformly apply pressure to the upper surface of the heat sink core 120 through spring bolts, thereby uniformly transferring the pressure to the surface of the chip 200;

[0065] The heat sink core 120 includes a soft cover plate 121 and a base fin type metal body 122; the soft cover plate 121 and the base fin type metal body 122 are locked by bolts 123;

[0066] The heat sink core 120 is used to absorb the heat generated by the chip 200 through the flow of cooling liquid for heat exchange.

[0067] The heat sink upper cover plate 110 disclosed in the embodiment of the present invention is made of a rigid material; Figure 2 and Figure 3 ;

[0068] The heat sink upper cover plate 110 has a first strip-shaped inlet 111;

[0069] The soft cover 121 has a second strip-shaped inlet 1211;

[0070] The first strip-shaped inlet 111 and the second strip-shaped inlet 1211 are used to provide an inlet for the cooling liquid to enter the substrate fin-type metal body 122 for heat exchange.

[0071] The soft cover 121 disclosed in the embodiment of the present invention is made of elastic material;

[0072] The soft cover 121 generates elastic deformation under pressure to evenly distribute the pressure.

[0073] The substrate fin type metal body 122 disclosed in the embodiment of the present invention is referred to Figure 4 , the substrate fin type metal body 122 includes fins 1221 and a substrate 1222;

[0074] Fins 1221 are used to increase surface area;

[0075] The fin 1221 includes a cut-off structure 1221a;

[0076] The base plate 1222 includes a plurality of first grooves 1222 a and a plurality of second grooves 1222 b . The first grooves 1222 a are parallel to the cutting structures 1221 a of the fins 1221 . The second grooves 1222 b are parallel to the fins 1221 .

[0077] The heat sink upper cover plate 110 disclosed in the embodiment of the present invention is a semi-enclosed shell structure.

[0078] On the other hand, the embodiment of the present invention further provides a heat sink optimization design method, which is used to design the aforementioned supporting chip heat sink 100, referring to Figure 5 , including the following steps:

[0079] S100, obtaining design parameters of a heat sink structure; the design parameters include fin height, fin thickness, and fin spacing;

[0080] S200, obtaining a design parameter combination according to the design parameters;

[0081] S300, setting several flow values;

[0082] S400, obtaining temperature rise data and flow resistance data according to the design parameter combination and flow value;

[0083] S500: Training a neural network for heat sink design based on temperature rise data and flow resistance data;

[0084] S600: Acquire heat sink design data according to the neural network of heat sink design.

[0085] S400 disclosed in the embodiment of the present invention obtains temperature rise data and flow resistance data according to the design parameter combination and the flow value, including the following steps:

[0086] S410 , for each design parameter combination, simulation or testing is performed at several flow rate values ​​to obtain temperature rise data and flow resistance data for each design parameter combination.

[0087] The S500 disclosed in the embodiment of the present invention trains a neural network for heat sink design based on temperature rise data and flow resistance data, including the following steps:

[0088] S510, obtaining the temperature rise and flow resistance required for model training based on the temperature rise data and the flow resistance data;

[0089] S520, setting input parameters; the input parameters include fin height, fin thickness, fin spacing, and flow resistance;

[0090] S530, setting output parameters; output parameters include flow value and temperature rise;

[0091] S540: Obtain a neural network for heat sink design based on the input parameters and the output parameters.

[0092] As an optional implementation, the present invention uses simulation or measured data to train a model, quickly evaluating a large number of design parameters and selecting the optimal heat sink design parameters. The heat sink performance optimization method, with flow resistance as the limiting value and minimum temperature rise as the goal, proceeds as follows:

[0093] Step 1: Generate training data using simulation or testing:

[0094] 1. Determine the design parameter range: Take N1 fin height values; take N2 fin thickness values; take N3 fin spacing values; combine these parameters to form N types of heat sink structure design parameters (N = N1 × N2 × N3).

[0095] 2. Determine the flow range: take M flow values.

[0096] 3. Obtain data through simulation or testing: For each heat sink structural design parameter (N types), simulate or test at M flow rates; record the temperature rise and flow resistance data for each combination; and ultimately generate M×N sets of data.

[0097] Step 1 of the embodiment of the present invention provides sufficient data samples for neural network training, so that the model can accurately predict the heat sink performance under different design parameters.

[0098] Step 2: Use the generated data to train the neural network:

[0099] 1. Determine input and output:

[0100] Input parameters: fin height, fin thickness, fin spacing, flow resistance.

[0101] Output parameters: flow rate, temperature rise.

[0102] 2. Use the M×N data generated in step 1 to train the neural network model.

[0103] 3. After training, the neural network can predict the corresponding flow rate and temperature rise based on the input design parameters (wing height, wing thickness, wing spacing, flow resistance).

[0104] Step 3: Input the design parameters to be evaluated into the neural network:

[0105] 1. Determine the range of design parameters to be evaluated: Take K1 fin height values; take K2 fin thickness values; take K3 fin spacing values; combine these parameters to form K types of heat sink structural design parameters to be evaluated (K = K1 × K2 × K3).

[0106] 2. Enter the target flow resistance value.

[0107] 3. Input each set of design parameters (wing height, wing thickness, wing spacing) and target flow resistance into the trained neural network.

[0108] 4. Neural network reasoning calculates the flow rate and temperature rise corresponding to each set of parameters, generating a total of K1×K2×K3 sets of data.

[0109] Step 4: Select the best design parameters:

[0110] 1. From the K1×K2×K3 data set generated in step 3, select the design that meets the target flow resistance requirements.

[0111] 2. In the design that meets the flow resistance requirements, select the set of parameters with the smallest temperature rise. This set of parameters is the optimal heat sink design parameters.

[0112] On the other hand, the embodiment of the present invention further provides an immersion liquid cooling server 300, referring to Figure 6 , including a server upper cover plate 310, a liquid block 320, and a front supporting chip heat sink 100;

[0113] The server cover 310 is used to cover the top of the server and protect the internal components of the server;

[0114] The liquid block 320 is used to cover the entire mainboard area and maintain a preset distance from the mainboard to prevent interference with electronic devices;

[0115] The supporting chip heat sink 100 is used to dissipate heat from the chip 200.

[0116] The liquid block 320 disclosed in the embodiment of the present invention includes a groove structure 321 and a slope surface structure 322; the distance between the bottom surface of the liquid block 320 and the main board is several millimeters to more than ten millimeters;

[0117] The groove structure 321 and the slope surface structure 322 are connected;

[0118] The groove structure 321 corresponds to the raised components on the motherboard; the raised components include the chip 200 and the memory module;

[0119] The groove structure 321 and the slope surface structure 322 are used to reduce the flow resistance of the coolant.

[0120] As an optional implementation, refer to Figure 7 Bottom view of the liquid block, Figure 8Cross-sectional view of the liquid block along the chip position and Figure 9 The cross-sectional view of the liquid block along the memory bar position shows that the groove structure 321 includes a groove 3211 above the chip heat sink and a groove 3212 above the memory bar. The liquid block 320 has an edge slope 323 on the side where the coolant flows in.

[0121] The present invention provides a dedicated liquid-cooled server for use in immersion liquid-cooled data centers, a matching high-power chip heat sink, and an optimized design method thereof. The server is equipped with a specially designed liquid-occupying block within the server, resulting in low liquid consumption and low flow resistance.

[0122] The matching chip heat sink proposed in the embodiment of the present invention consists of two major parts: a heat sink upper cover plate 110 and a heat sink core 120. The heat sink upper cover plate 110 is made of a material with good rigidity, and the heat sink core 120 is assembled by a soft cover plate 121 and a substrate fin-type metal body 122. The bottom of the heat sink core 120 is attached to the chip 200 to absorb the heat of the chip 200. The substrate thickness of the heat sink core 120 is small, the thermal resistance is small, the contact pressure between the heat sink and the chip is uniform, the flow resistance is easy to control, and the potential for heat exchange performance optimization is large; the heat sink adopts an assembly process as a whole, which effectively avoids the high cost problem of traditional brazing heat sink process.

[0123] The present invention also proposes an optimization design method and process for this type of heat sink, which can train a neural network model based on limited measured data or simulation data, and utilize the high efficiency of neural network reasoning to evaluate a large number of design parameters and select the optimal heat sink design parameters. It has important application value for optimizing the performance of high-power chip heat sinks, reducing flow resistance, and addressing the high energy consumption problems of the new generation of computing power centers.

[0124] As an optional embodiment, the supporting chip heat sink 100 of the present invention includes the following key designs:

[0125] 1) The server is designed with a special integrated liquid block (see Figure 6 The server cover 310 and liquid block 320 are shown in the figure: Liquid block 320 covers the entire motherboard area, with its bottom surface a few to ten millimeters away from the motherboard. This distance prevents collision and interference between liquid block 320 and most electronic components. Because liquid block 320 occupies a large amount of space inside the server, it saves a large amount of coolant and reduces the overall weight of the liquid cooling system.

[0126] 2) The liquid block has grooves in the areas where the memory sticks, chips and other components have more protrusions (see Figure 7 The groove 3211 above the chip heat sink, the groove 3212 above the memory bar), the front and rear sides of the groove are sloped ( Figure 7Slope surface structure 322): The presence of the groove prevents interference between the liquid block and the protruding electronic components. The slope surface structures 322 in front and behind the groove help the coolant flow uniformly in the groove area and reduce flow resistance. It should be noted that the slope surface structure 322 is not limited to a flat surface and can be other types of curved surfaces.

[0127] 3) The edge of the liquid block is provided with an inclined slope on the side where the coolant flows in ( Figure 7 The edge slope 323 is used to guide the coolant flow, thereby preventing the increase in flow resistance caused by vertical walls in the coolant flow path. It should be noted that the edge slope 323 is not limited to a flat surface and can be other types of curved surfaces.

[0128] 4) The supporting chip heat sink 100 is made of a heat sink cover plate ( Figure 1 heat sink upper cover plate 110) and heat sink core ( Figure 1 The heat sink core 120 is composed of a heat sink upper cover 110, which applies pressure to the upper surface of the heat sink core 120: the heat sink upper cover 110 is made of a metal or non-metallic material with good rigidity. When the heat sink upper cover 110 is pressed against the circuit board using spring bolts passing through the fixing holes, the heat sink upper cover 110 evenly distributes the pressure applied by the spring bolts to the upper surface of the heat sink core 120, and then evenly applies pressure to the surface of the chip 200 through the heat sink core 120, so that the heat sink core 120 and the chip 200 fit well. This force-bearing structure has low rigidity requirements for the heat sink core 120, which can prevent the heat sink core 120 from being subjected to concentrated force at the corners and causing warping deformation, providing a larger structural parameter design space for optimizing the heat sink's heat dissipation performance;

[0129] 5) The heat sink core 120 is composed of a soft cover plate 121 and a base plate fin type metal body 122: The heat sink core 120 proposed in the embodiment of the present invention is composed of two parts: a soft cover plate 121 (see Figure 2 Soft cover 121), substrate fin type metal body 122 (see Figure 2 The soft cover 121 and the fin 1221 form a series of parallel flow channels for efficient flow and heat exchange of the coolant;

[0130] 6) The soft cover plate 121 is made of an elastic material: the soft cover plate 121 can produce elastic deformation under the pressure of the heat sink upper cover plate 110, which is used to compensate for the slight deformation of the heat sink upper cover plate 110 under the pressure of the spring bolts, and evenly distribute the pressure applied by the heat sink upper cover plate 110 to the top of the fins 1221, avoiding uneven force caused by contact between hard materials;

[0131] 7) The soft cover 121 and the base fin type metal body 122 are assembled in a simple and inexpensive manner: the soft cover 121 and the base fin type metal body 122 of the present invention are directly locked by bolts 123 ( Figure 2 The bolts 123 do not require special sealing, and slight leakage is allowed at the joint. Because the soft cover 121 itself has a certain degree of elasticity, there is no need to use sealing strips, adhesives or other auxiliary sealing materials between the soft cover 121 and the base fin-type metal body 122, and there is no need to use high-cost connection methods such as welding. The process is simple and the cost is low.

[0132] 8) The heat sink upper cover plate 110 and the soft cover plate 121 are provided with strip-shaped inlets (the strip-shaped inlet of the heat sink upper cover plate 110 is shown in FIG. Figure 3 The first strip entrance 111 and the strip entrance of the soft cover 121 are shown in FIG. Figure 2 The second strip-shaped inlet 1211 is perpendicular to the fins 1221. The coolant enters the strip-shaped liquid distribution hole at high speed under the pressure of the pipeline, and is then distributed to the gaps between the fins 1221. It flows out from the gaps between the fins 1221 and takes away the heat energy absorbed by the bottom of the core.

[0133] 9) The heat sink cover plate 110 is a semi-enclosed shell structure: the coolant discharged from the fin gap impacts the side wall of the heat sink cover plate 110 and then returns inside the semi-enclosed shell, forming a turbulent flow with high heat exchange capacity. The turbulent flow passes around the chip (after the coolant is discharged from the heat sink core 120, the flow path inside the shell is shown). Figure 10 10A), three-dimensional heat dissipation is performed around the chip 200; the coolant is guided through the semi-enclosed shell structure of the heat sink upper cover 110 and then discharged from the open side of the shell of the heat sink upper cover 110 (the coolant discharge path from the heat sink upper cover 110 is shown in FIG. Figure 10 10B);

[0134] 10) The fins 1221 of the substrate fin type metal body 122 are cut (see Figure 4 The cutting structure 1221a of the fin 1221 is cut into grooves at the bottom of the substrate 1222 (see FIG. Figure 4 The first groove 1222a) is engraved with a plurality of grooves in a direction parallel to the fin 1221 (see Figure 4 Second grooves 1222b): This design divides the base plate 1222 and fins 1221 into several smaller sections. Due to the thin connections between these sections and the excellent ductility of copper, the uniform pressure applied from the top further improves the microscopic fit between each section and the surface of the chip 200, further reducing the contact thermal resistance between the chip 200 and the accompanying chip heat sink 100. When liquid metal thermal interface material is used between the heat sink core 120 and the chip 200, the grooves at the bottom of the base plate 1222 can absorb some of the melted liquid metal, helping to prevent it from overflowing the contact surface.

[0135] As an optional embodiment, the present invention provides a performance optimization method for a matching chip heat sink, which can obtain the best heat dissipation performance and energy-saving characteristics under the set target technical parameters:

[0136] Flow resistance has an important impact on the energy-saving characteristics of liquid cooling equipment. It should be considered at the beginning of heat sink design. The heat sink performance optimization method with flow resistance as the limit value and minimum temperature rise as the goal is as follows: Figure 11 .

[0137] refer to Figure 12 , heat sink performance optimization method with temperature rise as the limit value and minimum flow resistance as the goal:

[0138] Step 1: Generate training data using simulation or testing:

[0139] 1. Determine the design parameter range: Take N1 fin height values; take N2 fin thickness values; take N3 fin spacing values; combine these parameters to form N types of heat sink structure design parameters (N = N1 × N2 × N3).

[0140] 2. Determine the flow range: take M flow values.

[0141] 3. Obtain data through simulation or testing: For each heat sink structural design parameter (N types), simulate or test at M flow rates; record the temperature rise and flow resistance data for each combination; and ultimately generate M×N sets of data.

[0142] Step 2: Use the generated data to train the neural network:

[0143] 1. Determine input and output:

[0144] Input parameters: fin height, fin thickness, fin spacing, temperature rise.

[0145] Output parameters: flow rate, flow resistance.

[0146] 2. Use the M×N data generated in step 1 to train the neural network model.

[0147] 3. After training, the neural network can predict the corresponding flow rate and flow resistance based on the input design parameters (wing height, wing thickness, wing spacing, flow resistance).

[0148] Step 3: Input the design parameters to be evaluated into the neural network:

[0149] 1. Determine the range of design parameters to be evaluated: Take K1 fin height values; take K2 fin thickness values; take K3 fin spacing values; combine these parameters to form K types of heat sink structural design parameters to be evaluated (K = K1 × K2 × K3).

[0150] 2. Enter the target temperature rise value.

[0151] 3. Input each set of design parameters (wing height, wing thickness, wing spacing) and target temperature rise into the trained neural network.

[0152] 4. Neural network reasoning calculates the flow rate and flow resistance corresponding to each set of parameters, generating a total of K1×K2×K3 sets of data.

[0153] Step 4: Select the best design parameters:

[0154] 1. From the K1×K2×K3 data set generated in step 3, select the design that meets the target temperature rise requirements.

[0155] 2. In the design that meets the temperature rise requirements, select the set of parameters with the smallest flow resistance. This set of parameters is the optimal heat sink design parameters.

[0156] On the other hand, an embodiment of the present invention further provides a heat sink optimization design device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the heat sink optimization design method described above is implemented.

[0157] The processor and the memory can be connected via a bus or other means. The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0158] On the other hand, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above heat sink optimization design method.

[0159] Those skilled in the art will appreciate that all or some of the steps and systems in the method disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, and the computer-readable medium can include computer storage media (or non-transitory media) and communication media (or temporary media). As known to those skilled in the art, the term computer storage media is included in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data) and is volatile and non-volatile, removable, and non-removable. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technology, CD-ROM, digital versatile disks (DVD), or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage, or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, as is well known to those skilled in the art, communication media typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0160] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A matching chip heat sink, characterized in that: The supporting chip heat sink includes a heat sink upper cover plate and a heat sink core; The heat sink upper cover is used to uniformly apply pressure to the upper surface of the heat sink core through spring bolts, thereby uniformly transferring the pressure to the surface of the chip; The heat sink core comprises a soft cover plate and a base plate fin type metal body; the soft cover plate and the base plate fin type metal body are locked by bolts; The heat sink core is used to absorb the heat generated by the chip through the flow of cooling liquid for heat exchange.

2. The supporting chip heat sink according to claim 1, characterized in that: The heat sink upper cover is made of a rigid material; The heat sink upper cover plate has a first strip-shaped inlet; The soft cover has a second strip-shaped entrance; The first strip-shaped inlet and the second strip-shaped inlet are used to provide an inlet for the cooling liquid to enter the substrate fin-type metal body for heat exchange.

3. The supporting chip heat sink according to claim 1, characterized in that: The soft cover is made of elastic material; The soft cover plate generates elastic deformation under pressure to evenly distribute the pressure.

4. The supporting chip heat sink according to claim 1, characterized in that: The substrate-fin type metal body comprises fins and a substrate; The fins are used to increase the surface area; The fin includes a cut-off structure; The base plate includes a plurality of first grooves and a plurality of second grooves; the first grooves are parallel to the cutting structures of the fins; and the second grooves are parallel to the fins.

5. The supporting chip heat sink according to claim 1, characterized in that: The heat sink upper cover plate is a semi-enclosed shell structure.

6. A heat sink optimization design method, characterized in that: The heat sink optimization design method is used to design the matching chip heat sink according to any one of claims 1 to 5, comprising the following steps: Obtaining design parameters of the heat sink structure; the design parameters include fin height, fin thickness and fin spacing; According to the design parameters, obtaining a design parameter combination; Set several flow values; Acquiring temperature rise data and flow resistance data according to the design parameter combination and the flow value; training a neural network for heat sink design based on the temperature rise data and the flow resistance data; Heat sink design data is obtained according to the neural network of the heat sink design.

7. The heat sink optimization design method according to claim 6, characterized in that: The step of obtaining temperature rise data and flow resistance data according to the design parameter combination and the flow value comprises the following steps: For each of the design parameter combinations, simulation or testing is performed at several flow rate values ​​to obtain temperature rise data and flow resistance data for each of the design parameter combinations.

8. The heat sink optimization design method according to claim 6, characterized in that: The step of training a neural network for heat sink design based on the temperature rise data and the flow resistance data comprises the following steps: Obtaining the temperature rise and flow resistance required for model training according to the temperature rise data and the flow resistance data; Setting input parameters; the input parameters include the fin height, the fin thickness, the fin spacing, and the flow resistance; Setting output parameters; the output parameters include the flow value and the temperature rise; A neural network for heat sink design is obtained according to the input parameters and the output parameters.

9. An immersion liquid cooling server, characterized in that: The immersion liquid cooling server comprises a server cover plate, a liquid block, and a matching chip heat sink according to any one of claims 1 to 5; The server cover is used to cover the top of the server and protect the internal components of the server; The liquid-occupying block is used to cover the entire mainboard area and maintain a preset distance from the mainboard to prevent interference with electronic devices; The matching chip heat sink is used for chip heat dissipation.

10. The immersion liquid cooling server according to claim 9, characterized in that: The liquid-occupying block includes a groove structure and a slope surface structure; the distance between the bottom surface of the liquid-occupying block and the main board is several millimeters to more than ten millimeters; The groove structure and the slope surface structure are connected; The groove structure corresponds to the raised components on the motherboard; the raised components include chips and memory modules; The groove structure and the slope surface structure are used to reduce the flow resistance of the coolant.

Images