Reinforced computing power server

By adopting a horizontally placed GPU computing cards and control circuit boards stacked vertically in the AI ​​computing server, combined with PCIe cables and metal arched brackets, the space and vibration resistance issues of the vehicle-mounted AI computing server are solved, achieving space saving and improved stability.

CN223566092UActive Publication Date: 2025-11-18SHENZHEN AOSHENGTONG TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202422878775.0
Authority / Receiving Office
CN · China
Patent Type
Utility models(China)
Current Assignee / Owner
Filing Date
2024-11-22
Publication Date
2025-11-18
Estimated Expiration
2034-11-22

AI Technical Summary

Technical Problem

Existing AI computing servers face challenges in balancing space constraints and GPU card installation stability requirements in automotive applications, particularly due to their high height and depth requirements and insufficient vibration resistance.

Method used

The design adopts a horizontally placed GPU computing card and control circuit board stacked vertically, combined with PCIe cable and metal arch bracket, to optimize space utilization and enhance vibration resistance.

Benefits of technology

It saves on the server's depth dimensions, improves the GPU's vibration resistance, adapts to vehicle space constraints, and ensures normal operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN223566092U_ABST
    Figure CN223566092U_ABST
Patent Text Reader

Abstract

The utility model discloses a reinforced computing power server, and relates to the technical field of servers. The server comprises a server body, a control circuit board and a GPU computing power card. Wherein a cavity is formed in the server body; the control circuit board is fixedly arranged in the cavity; the GPU computing power card is horizontally arranged; the GPU computing power card and the control circuit board are stacked up and down, and the GPU computing power card is electrically connected with the control circuit board. The reinforced computing power server further comprises a PCIE (Peripheral Component Interconnect Express) flat cable arranged in the GPU computing power card; the control circuit board is provided with a PCIE slot for the PCIE flat cable to be plugged in. And the PCIE flat cable is matched with the PCIE slot so as to realize data transmission between the GPU computing power card and the control circuit board. By adopting the technical scheme, the method has the advantages of saving the space of the whole computing power server and improving the anti-vibration characteristic of the GPU computing power card.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The utility model relates to a server technical field, concretely relates to a reinforced computing power server. BACKGROUND

[0002] The AI computing power server is a server specially used for supporting artificial intelligence computing tasks. The AI computing power server has excellent graphic processing capability and high-performance computing capability, and is widely applied in the medical health field, the security field and the financial field and the like. The AI computing power server is gradually popularized and applied in the vehicle-mounted field, and the AI computing power server usually has a large size of AI acceleration card, GPU card and the like, and the height of the server needs to be 4U (177mm) or above, and the depth size needs to be 600mm or above; when applied in the vehicle-mounted field, the space in the vehicle is limited, the capacity of the vehicle-mounted cabinet is limited, the height and the depth size of the server are required to be high, and the stability of the installation of the GPU card is also required to be high. SUMMARY

[0003] The utility model discloses a reinforced computing power server, and at least one of the above technical problems is solved, which has the advantages of saving the space of the overall computing power server and improving the vibration resistance of the GPU computing power card.

[0004] To achieve the above object, the utility model adopts the technical scheme of a reinforced computing power server, which comprises:

[0005] A server body is internally provided with a cavity.

[0006] A control circuit board is fixedly arranged in the cavity.

[0007] A GPU computing power card is horizontally placed, and the GPU computing power card and the control circuit board are arranged in a vertical stack, and the GPU computing power card and the control circuit board are electrically connected.

[0008] The utility model further sets up that the reinforced computing power server further includes: the PCIE flat cable placed in the GPU computing power card;The control circuit board is provided with the PCIE slot for the PCIE flat cable plug-in;The PCIE flat cable and the PCIE slot cooperate to realize the data transmission between the GPU computing power card and the control circuit board.

[0009] The utility model further sets up that the reinforced computing power server further includes: the support between the control circuit board and the GPU computing power card;The support is used for installing the GPU computing power card.

[0010] The utility model further sets up that the support is in the shape of an arch.

[0011] The utility model further sets up, the support includes: first bending section, second bending section and third bending section, first bending section and the one end of third bending section are fixedly arranged in the both sides of control circuit board, and all perpendicular extension towards GPU algorithm power card direction, the both ends of second bending section are connected first bending section and the one end of third bending section close to GPU algorithm power card, first bending section, second bending section and third bending section cooperate so that control circuit board radiates heat.

[0012] The utility model further sets up, the second bending section is provided with the openwork structure.

[0013] The utility model further sets up, the openwork structure is provided with a plurality, and parallelly arranged.

[0014] The utility model further sets up, the material of support is metal.

[0015] The utility model further sets up, the front side and rear side of server body are provided with the vent.

[0016] The utility model further sets up, the shell of server body is metal material.

[0017] After adopting above technical scheme, the utility model has the beneficial effects that: in the utility model, control circuit board is fixedly arranged in the cavity of server body, GPU algorithm power card is horizontally placed, GPU algorithm power card and control circuit board are arranged in the upper and lower stacking, reduce the requirement of the depth of this reinforced algorithm power server, be favorable to save the space of this reinforced algorithm power server whole, optimize the space setting.Compared with the placement mode of GPU algorithm power card in traditional algorithm power server, GPU algorithm power card is horizontally placed, improve its anti-vibration characteristics, avoid the influence of external vibration on its normal work. ACCURACY

[0018] In order to more clearly illustrate the technical scheme in the embodiment of the utility model or prior art, the following will briefly introduce the drawing needed to be used in the embodiment or prior art description, obviously, the drawing in the following description only some embodiments of the utility model, for ordinary skilled person in the art, under the premise of not paying the creative labor, still can obtain other drawings according to these drawings.

[0019] Figure 1 It is the structure schematic diagram of the utility model;

[0020] Figure 2 It is the structure schematic diagram of the utility model hiding one shell;

[0021] Figure 3is the structure explosion schematic view of the utility model after one of the shell is hidden;

[0022] Figure 4 is the structure schematic view of the support.

[0023] Mark explanation: 100, server body;110, cavity;120, vent;200, control circuit board;300, GPU algorithm card;400, PCIE slot;500, support;510, first bending section;520, second bending section;521, hollow structure;530, third bending section. DETAILED DESCRIPTION

[0024] The utility model will be further explained in detail below in combination with the drawings.

[0025] The specific embodiment is only the explanation of the utility model, and it is not the limitation of the utility model, and the person skilled in the art can make the modification without the creative contribution according to the need after reading the present specification, but as long as in the utility model's right claim range, it is protected by patent law.

[0026] The embodiment relates to a kind of reinforced algorithm servers, refer to Figures 1-2 , including: server body 100, control circuit board 200 and GPU algorithm card 300.Therein, server body 100 is equipped with cavity 110 in, to facilitate placing other functional components;Control circuit board 200 is fixedly arranged in cavity 110;GPU algorithm card 300 is horizontally placed;GPU algorithm card 300 and control circuit board 200 are electrically connected.GPU algorithm card 300 and control circuit board 200 are arranged in a stack, beneficial to save the overall space of the reinforced algorithm server, optimize space arrangement and reduce the depth size of the reinforced algorithm server.GPU algorithm card 300 is horizontally placed simultaneously, which is beneficial to enhance the anti-vibration characteristics of GPU algorithm card 300.When the reinforced algorithm server is applied in vehicle-mounted field, horizontally placed GPU algorithm card 300 is not only beneficial to ensure its normal operation;But also can be well adapted to the size requirement of algorithm server in vehicle-mounted field, improve the reliability of the reinforced algorithm server when being used in vehicle-mounted field.

[0027] In the embodiment, the reinforced algorithm server further includes: PCIE flat cable (not shown in the drawing).Wherein, PCIE flat cable is placed in GPU algorithm card 300;Refer to Figure 2The PCIE slot 400 is provided on the control circuit board 200 for the PCIE cable to be plugged in. The PCIE cable cooperates with the PCIE slot 400 to realize data transmission between the GPU computing card 300 and the control circuit board 200. By connecting the control circuit board 200 and the GPU computing card 300 through the PCIE cable, the height dimension of the server is reduced, and the overall space of the reinforced computing server is further saved. The PCIE cable is an ideal choice for data transmission between the control circuit board 200 and the GPU computing card 300 due to its high bandwidth and low delay. The provision of the PCIE slot 400 facilitates the plugging of the PCIE cable.

[0028] In the embodiment, with reference to Figures 3-4 The reinforced computing server further includes a support 500. The support 500 is arranged between the control circuit board 200 and the GPU computing card 300, and is used for mounting the GPU computing card 300. While maintaining the up-down stacking design of the GPU computing card 300 and the control circuit board 200, the support 500 plays a role in reinforcing the GPU computing card 300 to increase its anti-vibration performance, and is also conducive to heat dissipation between the control circuit board 200 and the GPU computing card 300. Further, the support 500 is in an arc shape. The arc-shaped support 500 can effectively disperse the weight of the GPU computing card 300 and the vibration generated from the outside. Moreover, the arc-shaped support 500 can form a natural heat dissipation channel, which is conducive to the circulation of air in the cavity 110 of the reinforced computing server, accelerates the dissipation and discharge of heat. Specifically, the support 500 includes a first bent section 510, a second bent section 520, and a third bent section 530. One end of the first bent section 510 and one end of the third bent section 530 are fixedly arranged on the two sides of the control circuit board 200, respectively, and both extend vertically towards the GPU computing card 300. The second bent section 520 is connected to one end of the first bent section 510 and one end of the third bent section 530 close to the GPU computing card 300. The first bent section 510, the second bent section 520, and the third bent section 530 cooperate to dissipate heat from the control circuit board 200. The cooperation of the three bent sections improves the space utilization in the reinforced computing server. In other embodiments, the support 500 can also have other shapes.

[0029] In the embodiment, with reference to Figures 3-4 The second bent section 520 is provided with a hollow structure 521. The hollow structure 521 facilitates the plugging of the PCIE cable in the GPU computing card 300 and the PCIE slot 400 on the control circuit board 200. On the other hand, the hollow structure 521 further enhances the heat dissipation effect. Specifically, the hollow structure 521 is provided with a plurality of parallel structures, which facilitates the one-to-one plugging of the PCIE cable in the GPU computing card 300 and the PCIE slot 400 on the control circuit board 200. In some embodiments, the hollow structure 521 can also be provided with one.

[0030] In the embodiment, the material of the support 500 is metal, which is beneficial for heat dissipation. Specifically, the support 500 is made of aluminum. In some embodiments, the support 500 can also be made of aluminum alloy.

[0031] In the embodiment, the front side and the rear side of the server body 100 are both provided with the air vents 120, which are beneficial for air circulation to take away the heat in the cavity 110 of the reinforced computing power server, accelerate the heat dissipation, and further enhance the heat dissipation effect.

[0032] In the embodiment, the shell of the server body 100 is made of metal. Specifically, the shell of the server body 100 is made of aluminum. In some embodiments, the shell of the server body 100 can also be made of other alloy materials.

[0033] The above is only used to illustrate the technical scheme of the utility model and is not limited, and other modifications or equivalent replacements of the technical scheme of the utility model made by the ordinary skilled in the art should be covered in the scope of the claims of the utility model.

Claims

1. A hardened compute server, comprising: The server body (100) is internally provided with a cavity (110); A control circuit board (200) is fixedly arranged in the cavity (110); and A GPU computing card (300) is horizontally placed; the GPU computing card (300) and the control circuit board (200) are arranged in a vertical stack, and the GPU computing card (300) and the control circuit board (200) are electrically connected; The reinforced computing power server further comprises a support (500) arranged between the control circuit board (200) and the GPU computing card (300); the support (500) is used for mounting the GPU computing card (300); The support (500) comprises a first bending section (510), a second bending section (520) and a third bending section (530); one end of the first bending section (510) and one end of the third bending section (530) are fixedly arranged on both sides of the control circuit board (200) and vertically extend towards the GPU computing card (300); the second bending section (520) is connected to one end of the first bending section (510) and one end of the third bending section (530) close to the GPU computing card (300); the first bending section (510), the second bending section (520) and the third bending section (530) cooperate to dissipate heat of the control circuit board (200). The reinforced computing power server further comprises a PCIE cable arranged in the GPU computing card (300); the control circuit board (200) is provided with a PCIE slot (400) for plugging the PCIE cable; the PCIE cable and the PCIE slot (400) cooperate to realize data transmission between the GPU computing card (300) and the control circuit board (200).

2. The hardened compute server of claim 1, wherein, The support (500) is in an arch shape.

3. The hardened compute server of claim 1, wherein, The second bending section (520) is provided with a hollow structure (521).

4. The hardened compute server of claim 1, wherein, The hollow structure (521) is provided with a plurality of parallel structures.

5. The hardened compute server of claim 4, wherein, The material of the support (500) is metal.

6. The augmented compute server of any of claims 3-5, wherein, The server body (100) is provided with a ventilation opening (120) on the front side and the rear side.

7. The augmented compute server of any of claims 3-5, wherein, The shell of the server body (100) is made of metal.

8. The augmented compute server of any of claims 3-5, wherein, ​