A server hardware architecture and control method

By introducing adjustable installation substrate and intelligent resource management system into the server hardware architecture, the problems of low space utilization and unbalanced resource allocation within the server are solved, and compact and efficient resource utilization and low waste processing efficiency are achieved.

CN118732789BActive Publication Date: 2025-05-16WUHAN TONGWEI ELECTRONICS CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410913575.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-09
Publication Date
2025-05-16
Estimated Expiration
2044-07-09

AI Technical Summary

Technical Problem

In the existing server hardware architecture, the installation location of electrical components such as CPUs is fixed, resulting in low utilization of internal space of the server and cannot be compactly adjusted; at the same time, the existing technology cannot accurately allocate complex tasks when allocating resources, resulting in waste of resources and unbalanced utilization.

Method used

Design an adjustable server hardware architecture, and realize flexible installation and adjustment of electrical components through multiple installation substrates and limit adjustment components, forming high-heat and low-heat areas, and intelligent resource allocation through the main control system, task management system and resource management system.

Benefits of technology

It realizes compact partitioning and efficient utilization of the internal structure of the server, improves resource balance and utilization efficiency, and reduces resource waste and processing time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118732789B_ABST
    Figure CN118732789B_ABST
Patent Text Reader

Abstract

The present invention provides a server hardware architecture and a control method, comprising a server main body and a dust cover plate, wherein a plurality of mounting substrates are arranged inside the server main body, a circuit board main body and electrical components are arranged on the surface of the mounting substrate, guide sleeves are movably engaged at both ends of the back side of the mounting substrate, adjustment sleeves are arranged on both sides of the mounting substrate, a limit adjustment component is penetrated inside the adjustment sleeve, a guide plate and a telescopic sleeve are arranged between adjacent mounting substrates, a telescopic sleeve is slidably sleeved outside the guide plate, high-heat generating components inside the server such as CPU, GPU, memory bar, etc. are centrally installed, and low-heat generating components such as batteries, chips, etc. are centrally installed, so that a high-heat generating area and a low-heat generating area are formed inside the server main body, which is convenient for centralized cooling according to the partition environment, and the mounting substrate can autonomously adjust the spacing between adjacent electrical components or circuit board main bodies, thereby increasing the utilization rate, making the internal structure of the server main body more compact and highly adjustable.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of servers, and in particular to a server hardware architecture and a control method. Background Art

[0002] According to the Chinese publication number "CN217587961U", an artificial intelligence server hardware architecture based on a dual-channel domestic CPU includes a controller, a storage module, a heat dissipation module, a GPU module and a redundant power supply installed in a server chassis; the server chassis is divided into two layers, the lower layer of the server chassis is installed with a controller and a redundant power supply, and the upper layer of the server chassis is installed with a storage module, a heat dissipation module and a GPU module in sequence; the controller contains two domestic CPUs, the controller is connected to the storage module through a high-speed connector, the controller is connected and communicated with the GPU module through a high-speed connector, and the controller and the GPU module are both powered by a redundant power supply. The utility model has high integration and excellent computing performance.

[0003] According to a data center server control method based on artificial intelligence disclosed in China Publication No. "CN113806077A", the method includes the following steps: obtaining data center servers and average utilization, establishing a relationship between performance and average utilization, obtaining data center task sets and task deadlines, allocating resources based on performance and task sets, adjusting task sets at intervals of T, calculating the shortest running time of tasks, adjusting task running sequences, and calculating and outputting the time of task execution. The present invention implements a dynamic server scheduling algorithm to enable tasks to be executed efficiently, and by establishing a relationship between utilization and performance, tasks can be more efficiently allocated.

[0004] The above patent documents and prior art have the following technical problems when used:

[0005] Problem 1: Patent Document 1 has a fixed installation position for the CPU and other electrical components inside the server, which is not easy to adjust. In addition, the installation position of the electrical components inside the existing server body is fixed, and it is impossible to distinguish the heating zone according to the installation position. The installation position is fixed and cannot be adjusted compactly, resulting in low utilization of the internal space of the server body and poor adjustable performance.

[0006] Question 2: Patent Document 2 monitors the utilization rate of each component inside the server, establishes the relationship between performance and average utilization rate, and allocates resources according to performance and task sets. Although this method can increase utilization rate, it can only be used for simple tasks. For complex tasks, such as time-consuming and energy-intensive tasks, direct execution will cause resources to be occupied for a long time. When other tasks are issued, they cannot be executed. The resource allocation for complex tasks is insufficient. In addition, the above documents and existing technologies are prone to high-performance hardware being used for a long time, and low-performance hardware being used less frequently, resulting in uneven resource utilization within the server and obvious differentiation.

[0007] Question three: When allocating resources for server tasks, the above-mentioned patent documents and prior arts generally adopt the method of directly calling resources for centralized processing. They are unable to analyze the tasks and make accurate functional allocations, resulting in insufficient utilization of internal resources, large time loss, and low processing efficiency. Summary of the invention

[0008] Technical issues solved

[0009] In view of the deficiencies of the prior art, the present invention provides a server hardware architecture and a control method, which solves the following problems:

[0010] 1. The server's internal hardware installation is poorly adjustable and the server's internal space utilization is insufficient;

[0011] 2. When processing tasks within the server, there is a waste of resources and large differences between resources;

[0012] 3. The server entity is not sufficiently targeted when allocating task resources, resulting in long processing time and low processing efficiency.

[0013] Technical Solution

[0014] To achieve the above objectives, the present invention is implemented through the following technical solutions: a server hardware architecture, including a server body and a dust cover, the inner wall of the server body is provided with a T-shaped mounting groove, a plurality of mounting substrates are provided inside the server body, a circuit board body and electrical components are provided on the surface of the mounting substrate, both ends of the back side of the mounting substrate are movably engaged with guide sleeves, adjustment sleeves are provided on both sides of the mounting substrate, a limit adjustment component is penetrated through the inside of the adjustment sleeve, and one end of the limit adjustment component is movably connected to the inner wall of the guide sleeve, a guide plate and a telescopic sleeve are provided between adjacent mounting substrates, a telescopic sleeve is slidably sleeved on the outside of the guide plate, connecting support plates and T-shaped sliders are provided at both ends of the surface of the guide plate and the telescopic sleeve, the end of the connecting support plate is connected to the limit adjustment component, a T-shaped slide groove is provided on the surface of the guide sleeve, and the T-shaped slider is slidably located inside the T-shaped slide groove.

[0015] Preferably, the limit adjustment assembly includes an adjustment shaft and a limit tooth, the adjustment shaft passes through the interior of the adjustment sleeve, one end of the adjustment shaft is engaged with the top of the connecting support plate, the other end of the adjustment shaft is provided with a limit spring, the end of the limit spring is provided with a sliding bearing, and the sliding bearing is movably embedded in the guide sleeve, the surface of the adjustment shaft is provided with adjustment gears in a linear array, and one side of the inner wall of the guide sleeve is provided with limit teeth in a linear array, and the limit teeth are meshed with the surface of the adjustment gear.

[0016] Preferably, positioning screw holes are evenly distributed on the surface of the mounting substrate, the electrical components and the circuit board body are connected to the surface of the mounting substrate through the positioning screw holes, and the electrical components and the circuit board body are installed on the top and bottom surfaces of the mounting substrate.

[0017] Preferably, the telescopic sleeve surface is provided with positioning pin holes in a uniform array in vertical lines, the guide plate surface is provided with a spring positioning pin near the bottom end, and the spring positioning pin is engaged with the positioning pin hole, and the telescopic sleeve surface near the top and bottom surfaces of the server body is provided with a T-shaped mounting block, and the T-shaped mounting block is clearance-matched with the T-shaped mounting slot.

[0018] Preferably, the edge of the dust cover is connected to the front of the server body by screws, a heat dissipation hole is opened at the center of the back of the server body, heat dissipation arc grooves are distributed around the heat dissipation hole, the heat dissipation hole and the heat dissipation arc groove The inner groove profiles are both conical grooves, and a connection interface is provided on one side of the top surface of the server body.

[0019] Preferably, the spacing between adjacent adjusting gears is the same as the spacing between adjacent limiting teeth, and the length of the limiting teeth is the same as the thickness of the adjusting gears, the central axis of the adjusting shaft, the central axis of the adjusting sleeve and the central axis of the guide sleeve coincide with each other, the maximum sliding spacing between the T-shaped slider along the T-shaped slide groove is the same as the tooth length of the limiting teeth, the end of the adjusting shaft close to the connecting support plate is located outside the guide sleeve, and the connecting support plate and the T-shaped slider are both vertically mounted on the guide plate and the surface of the telescopic sleeve.

[0020] Preferably, the internal system structure of the server body includes a main control system, an early warning system, a task management system and a resource management system, wherein:

[0021] The main control system controls the electrical components and circuit board bodies inside the server body as a whole, issues instructions and monitors the start and completion of tasks, monitors the task management system and resource management system, and issues warnings through the warning system based on the monitoring results:

[0022] The early warning system receives the early warning signal from the main control system and connects with the early warning device to perform sound and light early warning;

[0023] The task management system monitors the tasks received by the server body and analyzes, decomposes, sets priorities and dynamically adjusts the order of the tasks through internal algorithm content;

[0024] The resource management system monitors the resource status inside the server body in real time, and intelligently controls and allocates the internal resources of the server body to the corresponding tasks based on the task priority and task content of the task management system in combination with the internal algorithm structure. It also monitors the resource structure in real time while the task is being executed, and intelligently adjusts the resource allocation based on the task execution status, time, and completion efficiency.

[0025] Preferably, the system controls the server body as follows:

[0026] Sp1: Resource monitoring, the resource management system monitors the resources inside the server in real time, including CPU, GPU, memory, and the monitoring types include temperature status, operation status, and occupancy rate;

[0027] Sp2: Task analysis. The task management system receives external task instruction content and analyzes the instruction content in combination with the internal algorithm. The analysis categories include task type, task requirements, task dependencies, task completion time, and the amount of resources that may be occupied when the task is enabled. The task is weighted according to the algorithm and a weighted threshold is set. If the actual weighted threshold is less than the set weighted threshold, it is a simple task that can be executed directly and enters Sp5. Otherwise, it is a complex task and enters Sp3 for task decomposition.

[0028] Sp3: Task decomposition, the task management system receives the complex task in Sp2 and the category of its analysis, and decomposes the task into multiple subtasks according to the completion process and steps of the task;

[0029] Sp4: Task priority setting and dynamic adjustment. Analyze each subtask in Sp3 according to the analysis category in Sp2, and add task urgency, task importance, task time limit, task impact and the degree of correlation with other tasks to the analysis category. Weight each item in the analysis category to obtain the overall weighted value, and sort the weighted values ​​of multiple subtasks. The higher the sorting, the higher the priority.

[0030] Sp5: Intelligent resource allocation. According to the resource monitoring content inside the server body, the CPU usage time data set in the resource management system is set to C, the GPU usage time data set is set to G, and the memory usage data set is set to T. Resource allocation is performed according to the priority of the subtask set in Sp4. When allocating resources, the CPU, GPU, and memory processing subtasks with small C values, G values, and T values ​​during the current resource monitoring are selected, and the corresponding processing priorities are reduced in the order of increasing C values, G values, and T values.

[0031] Sp6: Task execution and monitoring. The main control system starts the subtask after resource allocation and monitors the tasks during execution through the task management system and resource management system. According to the task completion status information and the real-time monitoring of current resources, the priority and resource allocation are dynamically adjusted until the task is completed.

[0032] Sp7: Task evaluation and optimization iteration. The main control system evaluates the entire process of task completion. The evaluation content includes all operating status and adjustment status of each system in steps Sp1-Sp6, forms a database, and continuously optimizes and iterates the algorithms within the task management system and resource management system.

[0033] Preferably, the resource management system makes the CPU usage time data set C = {C1, C2, ...C n}, where C1, C2, ...C n The corresponding C value gradually increases, making the GPU usage time data set G = {G1, G2, ...G m} Where G1, G2, …G m The corresponding G value gradually increases, making the memory usage data set T = {T1, T2, ...T p}, where T1, T2, …T p The corresponding T value gradually increases.

[0034] Preferably, when allocating resources in the resource management system, the allocation is performed according to the resource categories mainly consumed by the completion of the subtasks. For example, when a large amount of memory is required, the memory is matched first for resource application.

[0035] Beneficial Effects

[0036] The present invention provides a server hardware architecture and a control method, which has the following beneficial effects:

[0037] 1. The present invention realizes compact partitioning and adjustable height of the internal structure of the server body. Multiple mounting substrates are arranged inside the server body for mounting the circuit board body and electrical components. Multiple CPU and GPU components can be installed, and limit adjustment components are arranged at the ends of adjacent mounting substrates for angle adjustment. The height spacing is adjusted by the cooperation of the guide plate and the telescopic sleeve. Through the function of the mounting substrate, high-heat generating components inside the server such as the CPU, GPU, and memory stick can be centrally installed, and low-heat generating components such as batteries and chips can be centrally installed, so that high-heat generating areas and low-heat generating areas are formed inside the server body. The areas are connected by line arrangement, which is convenient for centralized cooling according to the partition environment. At the same time, the spacing between adjacent electrical components or circuit board bodies can be independently adjusted through the foldable mounting substrate, thereby increasing the utilization rate and making the internal structure of the server body more compact and highly adjustable.

[0038] 2. The present invention realizes homogenization of functions and resources inside the server body, averages the loss of the server body, reduces the loss difference, adopts the internal installation baseboard to connect multiple CPUs, GPUs, memories and other electrical components for use, and internally regulates the task instructions of the server body through the main control system, early warning system, task management system and resource management system, and allocates the internal resources of the server body through task analysis and classification. Through weighted analysis and dynamic resource regulation, the task management system can make the resource utilization of the entire server more balanced, high-priority tasks can obtain the required resources first, and low-priority or simple tasks will not occupy too many resources, thereby reducing the waste of server resources and differentiated losses, realizing the homogenization of functions and resources of sub-servers inside the server, averaging the loss of the entire server, and improving the overall utilization efficiency of resources.

[0039] 3. The present invention realizes the allocation by function within the server body, improves efficiency, and reduces time loss. When managing tasks within the server body, the task management system cooperates with the resource management system to analyze the task type, requirements, dependencies, completion time, and resource requirements when enabled, and accurately allocates corresponding resources according to the functional requirements of the task such as CPU, GPU, memory, etc., to ensure that each task obtains the most suitable execution environment. This allocation by function not only improves the efficiency of task execution, but also makes the best use of the server's hardware facilities, avoids resource waste and unnecessary waiting time, and can effectively prioritize and evenly allocate the functions and resources of the sub-servers within the server. The task management system can significantly reduce the time loss during task execution through intelligent resource allocation and priority setting. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a main structure diagram of the server of the present invention;

[0041] Figure 2 This is a front structural diagram of the server body of the present invention;

[0042] Figure 3 The overall structure diagram of the mounting substrate of the present invention;

[0043] Figure 4 It is a partial structural diagram of the mounting substrate of the present invention;

[0044] Figure 5 This is a connection structure diagram of the limit adjustment component of the present invention;

[0045] Figure 6 It is a connection structure diagram of the mounting substrate of the present invention;

[0046] Figure 7This is a front internal structure diagram of the mounting substrate of the present invention;

[0047] Figure 8 This is a top surface internal structure diagram of the mounting substrate of the present invention;

[0048] Fig. 9 This is a diagram showing the internal structure of the server body of the present invention;

[0049] Fig.10 This is a back structural diagram of the server body of the present invention;

[0050] Fig.11 A diagram showing the internal system connections of the server body of the present invention;

[0051] Fig.12 This is a diagram of the internal control method of the server body of the present invention.

[0052] Among them: 1. Server body; 2. Dust cover; 3. Connection interface; 4. Heat dissipation arc groove; 5. Heat dissipation hole; 6. T-shaped mounting groove; 7. Mounting base plate; 8. Positioning screw hole; 9. Adjustment sleeve; 10. Guide sleeve; 11. T-shaped slide groove; 12. Limit adjustment assembly; 1201. Limit tooth; 1202. Adjustment shaft; 1203. Adjustment gear; 1204. Sliding bearing; 1205. Limit spring; 13. Guide plate; 14. Telescopic sleeve; 15. Spring positioning pin; 16. Positioning pin hole; 17. T-shaped slider; 18. Connecting support plate; 19. Circuit board body; 20. Electrical component; 21. T-shaped mounting block. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Specific embodiment one:

[0055] like Figure 1-10As shown, a server hardware architecture includes a server body 1 and a dust cover 2. A plurality of mounting substrates 7 are arranged inside the server body 1. A circuit board body 19 and electrical components 20 are arranged on the surface of the mounting substrate 7. Positioning screw holes 8 are evenly distributed on the surface of the mounting substrate 7. The electrical components 20 and the circuit board body 19 are connected to the surface of the mounting substrate 7 through the positioning screw holes 8. The top and bottom surfaces of the mounting substrate 7 are both installed with electrical components 20 and the circuit board body 19. The entire server body 1 is mainly installed with a plurality of circuit board bodies 19, electrical components 20, etc. through the mounting substrate 7. Compared with the single mounting plate structure inside the traditional server body 1, the present application sets a plurality of mounting substrates 7 to install a plurality of circuit board bodies 19 and electrical components 20, so as to realize capacity expansion inside the server body 1. For example, a plurality of CPU components are installed on the surface of the same circuit board body 19 for use. The electrical components 20 are actually used. The conventional component structure used for the operation of the server body 1 is used, such as cooling fans, graphics cards, memory sticks, CPUs, GPUs, etc. The same type of electrical components 20 can be arranged in multiple groups during actual installation. Since there are multiple mounting substrates 7, during actual installation, they can be partitioned and installed according to the actual heating conditions of the electrical components 20. High-heating components inside the server, such as CPUs, GPUs, memory sticks, etc., can be centrally installed, and low-heating components, such as batteries and chips, can be centrally installed. For example, the high-heating zone can be installed on the surface of the mounting substrate 7 close to the top surface of the server body 1. Due to the air flow, the temperature flows upward, so that the heat of the high-heating zone on the top can be quickly dissipated to reduce the influence of thermal radiation. The coconut shell sets a separate heat dissipation structure for the high-heating zone for centralized heat dissipation, so that high-heating zones and low-heating zones are formed inside the server body 1. The areas are connected by line layout, which is convenient for centralized cooling according to the partition environment.

[0056] When adjusting the angle between adjacent mounting substrates 7, both ends of the back side of the mounting substrate 7 are movably engaged with guide sleeves 10, and adjusting sleeves 9 are provided on both sides of the mounting substrate 7. A limit adjustment component 12 is penetrated inside the adjusting sleeve 9, and one end of the limit adjustment component 12 is movably connected to the inner wall of the guide sleeve 10. When adjusting the angle of the mounting substrate 7, the positioning is mainly carried out by the limit engagement between the adjusting sleeve 9 and the surface of the limit adjustment component 12. The specific structure is that the limit adjustment component 12 includes an adjusting shaft 1202 and a limit tooth 1201, the adjusting shaft 1202 penetrates along the inside of the adjusting sleeve 9, one end of the adjusting shaft 1202 is engaged with the top end of the connecting support plate 18, and the other end of the adjusting shaft 1202 is provided with a limit spring 1205. The limit spring 120 A sliding bearing 1204 is provided at the end of the adjusting shaft 1202, and the sliding bearing 1204 is movably embedded in the guide sleeve 10. The movable bearing at the end of the adjusting shaft 1202 is supported by the guide sleeve 10 to ensure that when one end of the adjusting shaft 1202 is squeezed, the end connected to the guide sleeve 10 will not be separated. The surface of the adjusting shaft 1202 is provided with an adjusting gear 1203 in a linear array, and one side of the inner wall of the guide sleeve 10 is provided with a limiting tooth 1201 in a linear array. The limiting tooth 1201 is meshed with the surface of the adjusting gear 1203. The spacing between adjacent adjusting gears 1203 is the same as the spacing between adjacent limiting teeth 1201, and the length of the limiting tooth 1201 is the same as the thickness of the adjusting gear 1203, so that the adjusting gear 1203 and the limiting tooth 1201 are connected. When in the engaged state, each position is kept securely engaged to increase stability. The maximum sliding distance between the T-shaped slider 17 and the T-shaped slot 11 is the same as the tooth length of the limit tooth 1201. The adjusting shaft 1202 is meshed with the end of the limit tooth 1201 through the adjusting gear 1203 on the surface to limit the rotation angle of the adjusting sleeve 9. When the adjusting gear 1203 is not engaged with the limit tooth 1201, the mounting base 7 can be adjusted by rotating the adjusting sleeve 9 at the end along the outer surface of the adjusting circle to adjust the angle. Under normal conditions, the adjusting gear 1203 and the end of the limit tooth 1201 remain in an engaged state. When adjustment is required, since the end of the adjusting shaft 1202 close to the connecting support plate 18 is located outside the guide sleeve 10, the connecting support plate 18 and the T-shaped slider 17 are connected. They are vertically mounted on the surfaces of the guide plate 13 and the telescopic sleeve 14, and are pressed inward along one end of the adjusting shaft 1202 and the connecting support plate 18 to move the adjusting shaft 1202, driving the adjusting gear 1203 on the surface to move and stagger the meshing position with the limiting tooth 1201. At this time, the limiting spring 1205 is squeezed and contracted, and the staggered limiting tooth 1201 is no longer restricted. The mounting base 7 can be rotated along the outer surface of the adjusting shaft 1202 by the adjusting sleeve 9. When rotating the angle, the end of the guide sleeve 10 is movably engaged with the surface of the mounting base 7, and the central axis of the adjusting shaft 1202, the central axis of the adjusting sleeve 9 and the central axis of the guide sleeve 10 coincide with each other. Therefore, when the mounting base 7 rotates the angle, the guide sleeve 10 keeps its position.When the mounting base plate 7 rotates, it rotates along the connection position with the guide sleeve 10. When adjusting the position, the squeezing force on the adjusting shaft 1202 is continuously maintained. After adjusting the angle, the squeezing force on the adjusting shaft 1202 is withdrawn, so that the limit spring 1205 rebounds, driving the adjustment gear 1203 to rebound to the initial position, and maintaining the engagement with the surface of the limit tooth 1201 after rotation, completing the positioning of the mounting base plate 7 after the angle adjustment. By adjusting the angle of the surface of the mounting base plate 7, the angle folding degree between adjacent mounting base plates 7 can be adjusted according to actual use needs, and the space utilization rate inside the server body 1 is controlled and adjusted to achieve compact adjustment of the space.

[0057] When the height between adjacent mounting base plates 7 is adjusted, since a guide plate 13 and a telescopic sleeve 14 are provided between adjacent mounting base plates 7, the telescopic sleeve 14 is slidably sleeved on the outside of the guide plate 13, and connecting support plates 18 and T-shaped sliders 17 are provided at both ends of the surfaces of the guide plate 13 and the telescopic sleeve 14, the end of the connecting support plate 18 is connected to the limit adjustment component 12, and a T-shaped slide groove 11 is provided on the surface of the guide shaft sleeve 10, and the T-shaped slider 17 is slidably located inside the T-shaped slide groove 11. The sliding adjustment between the guide plate 13 and the telescopic sleeve 14 can realize the vertical height adjustment between adjacent mounting base plates 7, and is used to connect the limit adjustment components 12 between adjacent mountings to maintain the continuous structure of multiple mounting base plates 7. Since the guide plate 13 and The surface of the telescopic sleeve 14 is provided with a connecting support plate 18 and a Y-shaped slider, which have the same structure when connected to the limit adjustment component 12, so the adjustment principle is the same. Taking the guide plate 13 as an example, the top of the guide plate 13 is connected to the end of the adjustment shaft 1202 through the connecting support plate 18, and the top of the T-shaped slider 17 is located inside the T-shaped slide groove 11 on the surface of the guide shaft sleeve 10. The limit adjustment component 12 and the guide plate 13 are connected at one end through the connecting support plate 18, and the other end of the guide plate 13 is hoisted by the engagement of the T-shaped slider 17 and the T-shaped slide groove 11 to maintain the stability of the connection at the end of the guide plate 13. When the conventional adjusting gear 1203 and the limit gear 1201 are kept in meshing state, the T-shaped slider 17 is located in the T-shaped slide groove. The groove 11 is close to one end of the surface of the mounting substrate 7, and there is a gap between the side of the connecting support plate 18 and the mounting substrate 7. When the adjusting shaft 1202 is pressed, the guide plate 13 and the connecting support plate 18 move following the adjusting shaft 1202. While moving, the T-shaped slider 17 slides along the inside of the T-shaped slide groove 11, so that when the surface of the connecting support plate 18 abuts against the surface of the mounting substrate 7, the limit tooth 1201 and the adjusting gear 1203 just disengage. The T-shaped slider 17 is just located inside the end of the T-shaped slide groove 11 away from the mounting substrate 7 at this time, and the gap between adjacent electrical components 20 or circuit board bodies 19 can be adjusted autonomously, thereby increasing the utilization rate, making the internal structure of the server body 1 more compact, and having a high degree of adjustability. When positioning and adjusting between the guide plate 13 and the telescopic sleeve 14, since the telescopic sleeve 14 surface is provided with a vertical straight line uniform array of positioning pin holes 16, the guide plate 13 surface is provided with a spring positioning pin 15 near the bottom end, and the spring positioning pin 15 is engaged with the positioning pin hole 16, so that the guide plate 13 can be adjusted in height by sliding along the inside of the telescopic sleeve 14. After the height is adjusted, the spring positioning pin 15 is engaged with the positioning pin hole 16 on the surface of the telescopic sleeve 14 to achieve the adjusted positioning. The surface of the positioning plate and the position of the spring positioning pin 15 are provided with grooves for the rebound of the spring positioning pin 15. The spring positioning pin 15 is a structure in which the spring is connected to the bottom end of the pin shaft, which is used to adjust the engaging position with the positioning pin hole 16 to achieve the height adjustment of the guide plate 13 and the telescopic sleeve 14.

[0058] When the entire mounting substrate 7 is connected and installed with the server body 1, since the inner wall of the server body 1 is provided with a T-shaped mounting groove 6, the surface of the telescopic sleeve 14 close to the top and bottom surfaces of the server body 1 is provided with a T-shaped mounting block 21, and the T-shaped mounting block 21 is clearance-matched with the T-shaped mounting groove 6, the mounting substrate 7 is positioned at both ends of the entire mounting substrate 7 through the engaging position of the T-shaped mounting block 21 and the T-shaped mounting groove 6, and the mounting position is adjusted according to actual use requirements. The inner wall of the server body 1 is provided with T-shaped mounting grooves 6 in a linear array on all sides of the surface, which is convenient for adaptive installation in multiple positions. Specific embodiment 2:

[0060] like Figure 1-12 As shown, according to the content of the specific embodiment 1, the following contents are further disclosed:

[0061] When the entire server body 1 is dissipating heat, the edge of the dust cover 2 is connected to the front of the server body 1 by screws, a heat dissipation hole 5 is opened at the center of the back of the server body 1, and heat dissipation arc grooves 4 are distributed around the heat dissipation hole 5. A connection interface 3 is provided on one side of the top surface of the server body 1. The entire server body 1 is connected to the external device through the connection interface 3, and convection between the internal air and the external space is carried out through the heat dissipation hole 5 and the heat dissipation arc groove 4 to dissipate air. In actual settings, the change in aperture will affect the flow rate and pressure distribution of the fluid. In the area where the aperture is reduced, the fluid velocity will increase. According to the Bernoulli principle, when the flow rate increases, the static pressure decreases, which helps to take heat away from the heat source because more fluid molecules contact the surface of the heat source and take away heat. The internal groove profiles of the heat dissipation hole 5 and the heat dissipation arc groove 4 are all conical grooves, and the aperture gradually changes from large to small, which can form an effect similar to a Venturi tube. Although this is usually used for acceleration and mixing of gas or liquid, in heat dissipation design, its ideas can be used to optimize the fluid flow path, reduce eddy currents and reflux, and improve heat dissipation efficiency. Specific embodiment three:

[0063] like Figure 1-12 As shown, according to the content of the specific embodiment 1, the following contents are further disclosed:

[0064] When the entire device is actually used, the connection structure between the connecting support plate 18 and the adjusting shaft 1202 can be installed by threading or bolting, so that the connecting support plate 18 can be detached from the end of the adjusting shaft 1202, and the guide sleeve 10 is penetrated at one end of the length of the T-shaped slide groove 11, and a limiting screw or bolt structure is set at the penetration position to prevent the T-shaped slider 17 from falling off during routine adjustment. At the same time, the T-shaped slider 17 can be slid and disassembled by removing the bolts, so as to install and disassemble the guide plate 13 and the telescopic sleeve 14 at the end of the limit adjustment component 12, thereby realizing the assembly and splicing of multiple mounting substrates 7. According to the actual installation needs inside the server body 1, the number of assembled mounting substrates 7 and the corresponding structural contents of the guide plate 13 and the telescopic sleeve 14 are selected. Specific embodiment four:

[0066] like Figure 1-12 As shown, according to the content of the specific embodiment 1, the following contents are further disclosed:

[0067] The internal system structure of the server body 1 includes a main control system, an early warning system, a task management system and a resource management system, among which:

[0068] The main control system controls the electrical components 20 and the circuit board 19 inside the server body 1 as a whole, issues instructions and monitors the start and completion of tasks, monitors the task management system and resource management system, and issues warnings through the warning system based on the monitoring results:

[0069] Early warning system, which receives early warning signals from the main control system and connects with early warning equipment for sound and light warning;

[0070] The task management system monitors the tasks received by the server body 1 and analyzes, decomposes, sets priorities and dynamically adjusts the order of the tasks through internal algorithm content;

[0071] The resource management system monitors the resource status inside the server body 1 in real time, and intelligently controls and allocates the internal resources of the server body 1 to correspond to tasks based on the task priority and task content of the task management system and the internal algorithm structure. It also monitors the resource structure in real time while the task is being executed, and intelligently adjusts resource allocation based on the task execution status, time, and completion efficiency.

[0072] The system regulates server entity 1 as follows:

[0073] Sp1: Resource monitoring. The resource management system monitors the resources inside the server body 1 in real time, including the CPU, GPU, and memory. The monitoring types include temperature status, operating status, and occupancy rate.

[0074] Sp2: Task analysis. The task management system receives external task instructions and analyzes the instruction content in combination with internal algorithms. The analysis categories include task type, task requirements, task dependencies, task completion time, and the amount of resources that may be occupied when the task is enabled. The task is weighted according to the algorithm and a weighted threshold is set. If the actual weighted threshold is less than the set weighted threshold, it is a simple task that can be executed directly and enters Sp5. Otherwise, it is a complex task and enters Sp3 for task decomposition.

[0075] Sp3: Task decomposition. The task management system receives the complex tasks and their analysis categories in Sp2, and decomposes the tasks into multiple subtasks according to the completion process and steps of the tasks;

[0076] Sp4: Task priority setting and dynamic adjustment. Analyze each subtask in Sp3 according to the analysis category in Sp2, and add task urgency, task importance, task time limit, task impact and the degree of correlation with other tasks to the analysis category. Weight each item in the analysis category to obtain the overall weighted value, and sort the weighted values ​​of multiple subtasks. The higher the sorting, the higher the priority.

[0077] Sp5: Intelligent resource allocation. According to the resource monitoring content inside the server main body 1, the CPU usage time data set in the resource management system is set to C, the GPU usage time data set is set to G, and the memory usage data set is set to T. The resource management system makes the CPU usage time data set C = {C1, C2, ...C n}, where C1, C2, ...C n The corresponding C value gradually increases, making the GPU usage time data set G = {G1, G2, ...G m} Where G1, G2, …G m The corresponding G value gradually increases, making the memory usage data set T = {T1, T2, ...T p}, where T1, T2, …T p The corresponding T value gradually increases, and resources are allocated according to the priority of the subtask set in Sp4. When allocating resources, select the subtasks with high CPU, GPU and memory processing priorities with small C value, G value and T value during current resource monitoring. In the order of increasing C value, G value and T value, the corresponding processing priority decreases in turn. When allocating resources in the resource management system, allocation is made according to the resource category mainly consumed by the completion of the subtask. If a large memory is required, the memory is matched first for resource application;

[0078] Sp6: Task Execution and Monitoring. The master control system executes the start instruction for the subtasks after resource allocation, and monitors the tasks during the execution process through the task management system and the resource management system. According to the status information of task completion and in coordination with the real-time monitoring of current resources, the priority and resource allocation are dynamically adjusted until the task is completed;

[0079] Sp7: Task Evaluation and Optimization Iteration. The master control system evaluates the entire process of task completion. The evaluation content includes all the running states and adjustment states of each system in steps Sp1 - Sp6, forms a database, and continuously optimizes and iterates the algorithms inside the task management system and the resource management system. Specific Embodiment Five:

[0081] As Figure 1-12 shown, based on the content in Specific Embodiment Four, the following content is further disclosed:

[0082] The task analysis algorithm adopted in the task management system is as follows:

[0083] The task analysis algorithm is mainly responsible for parsing the received task instructions and evaluating their complexity and resource requirements. Here, a weight - based scoring system can be used to determine the complexity of the task. Set the set of evaluation factors for task A as F = {f1, f2,... f n}, where f i represents the i - th evaluation factor, and each factor has a corresponding weight w, and there is:

[0084]

[0085] The task complexity score S can be calculated by weighted sum:

[0086]

[0087] where eval(f i , A) is the specific evaluation value of task T on evaluation factor f i , which may be quantitative, such as resource demand, or qualitative and needs to be converted into a quantitative score;

[0088] When classifying tasks, if S < Threshold (the preset threshold), the task is a simple task; otherwise, it is a complex task.

[0089] For the subtasks after the decomposition of complex tasks, priorities need to be set and dynamically adjusted. This usually involves multi - factor weighted sorting. Set the set of evaluation factors for subtask t j as P = {P1, P2,... P m}, where P kRepresents the kth evaluation factor, such as urgency, importance, time limit, etc. Each factor has a corresponding weight v k , and there are:

[0090]

[0091] Subtask priority score P j It can be calculated by weighted sum:

[0092]

[0093] Then, according to P j The value of is used to sort all subtasks. The higher the sorting, the higher the priority. Specific embodiment six:

[0095] like Figure 1-12 As shown, according to the contents of specific embodiments four and five, the following contents are further disclosed:

[0096] The contents of the intelligent resource allocation algorithm in the resource management system are as follows:

[0097] Resource monitoring data: First, the resource management system tries to collect and update the usage data of CPR, GPU, and memory. This data can be represented as three sets, as shown below:

[0098] CPR usage duration dataset C = {C1, C2, ...C n}, where C i Indicates the current usage time or load of the i-th CPU or CPU core;

[0099] GPU usage time dataset G = {G1, G2, ...G m}, where G j Indicates the current usage time or load of the jth GPU;

[0100] Memory usage dataset T = {T1, T2, ...T p}, set T k Indicates the usage of a certain area or the whole memory.

[0101] Task resource requirements, each task t j There is a set of resource requirements for drinking in and Represents the requirements for CPU, GPU and memory respectively;

[0102] Intelligent resource allocation algorithm allocate(t j , the core logic of A) is as follows:

[0103] Input: Task t to be assigned j and the currently available resource set A, including the status information of CPU, GPU and memory resources;

[0104] Evaluation: According to the task j resource requirements and current resource monitoring data to assess which resources are best suited to be allocated to task t j This usually involves comparing the resource requirements of the task with the current status of the resources, evaluating the availability and remaining capacity of various resources based on the information in the resource status database, identifying which resources are idle or lightly loaded, and which resources are busy or close to full load, such as idleness, performance, etc.;

[0105] Selection: Select from the available resource set A to satisfy task t j The resource demand and the resources with lighter current load are allocated, which may involve multi-factor decision-making, such as giving priority to resources with better performance and lower current load.

[0106] Allocate: Assign the selected resources to the task j , and updates resource monitoring data and task status.

[0107] Output: Returns the value assigned to task t j Resource collection

[0108] After resource allocation, the resource usage and task execution are continuously monitored. The main control system will update the information in the resource status database in real time and evaluate the system performance after resource allocation, including task execution efficiency, resource utilization, system response time, etc. This can be achieved by comparing the performance indicators before and after allocation. If it is found that resource allocation is unreasonable or system performance does not meet expectations, the resource allocation strategy will be adjusted in time. The adjustment can be based on real-time data, historical data or prediction models. The adjusted allocation strategy will be reapplied to new tasks and resource requirements.

[0109] After the task is completed, feedback on the resource allocation effect is collected from users, system administrators or other relevant parties. The feedback may include the improvement of task execution efficiency, the improvement of resource utilization or the shortening of system response time. Based on the feedback and monitoring data, the resource allocation algorithm is continuously optimized. The optimization may include improving the allocation strategy, optimizing the prediction model, adjusting the algorithm parameters, etc. The goal of optimization is to improve the efficiency and accuracy of resource allocation to better meet the task requirements and improve the system performance. The logical process of the intelligent resource allocation algorithm is an iterative and optimization process, which needs to be continuously adjusted and optimized according to the actual situation and needs. Through this process, it can ensure that resources are used efficiently and reasonably, thereby improving the overall performance and efficiency of the system. Specific embodiment seven:

[0111] like Figure 1-12 As shown, according to the contents of specific embodiments four, five and six, the following contents are further disclosed:

[0112] When conducting task evaluation, collect various indicator data during the task execution process, including task execution time, resource utilization, system response time, error rate, etc., perform performance analysis on the collected data, evaluate the efficiency and effectiveness of task execution, and identify bottlenecks in the task execution process through data analysis, such as insufficient resources, inefficient algorithms, network delays, etc.

[0113] Use log collection and analysis tools such as ELK Stack and Splunk to collect and analyze task execution data, use data visualization tools such as Grafana and Tableau to display analysis results for easy understanding and decision-making, write performance evaluation reports, and summarize problems and optimization suggestions during task execution.

[0114] During the optimization iteration, according to the performance evaluation results, the algorithms of the task management system and resource management system are optimized, such as improving the resource allocation algorithm, optimizing the task scheduling strategy, etc. The system is regularly upgraded, and new technologies and tools are introduced to improve system performance and stability. A feedback loop mechanism is established, and the optimized system is redeployed to the production environment. Data continues to be collected and analyzed to verify the optimization effect and perform further iterative optimization. Version control systems such as Git are used to manage system code and configuration files. Automated test scripts are written to ensure the stability and performance of the system after each optimization. Continuous integration / continuous deployment CI / CD tools such as Jenkins and GitLab CI / CD are used to automate the testing and deployment process. Through the implementation of the above solutions and technologies, dynamic adjustment of priority and resource allocation during task execution, as well as comprehensive evaluation and optimization iteration of task execution effects can be achieved, thereby improving the overall performance and efficiency of the system.

[0115] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprising a reference structure" do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0116] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A server hardware architecture, comprising a server body (1) and a dust cover (2), characterized in that: The inner wall of the server body (1) is provided with a T-shaped installation groove (6), and a plurality of installation substrates (7) are provided inside the server body (1). The surface of the installation substrate (7) is provided with a circuit board body (19) and an electrical component (20). Both ends of the back of the installation substrate (7) are movably engaged with guide sleeves (10), and both sides of the installation substrate (7) are provided with adjustment sleeves (9). A limit adjustment component (12) is provided inside the adjustment sleeve (9), and one end of the limit adjustment component (12) is connected to the inner side of the guide sleeve (10). The wall is movably connected, a guide plate (13) and a telescopic sleeve (14) are provided between adjacent mounting base plates (7), the telescopic sleeve (14) is slidably sleeved on the outside of the guide plate (13), connecting support plates (18) and T-shaped sliders (17) are provided at both ends of the surfaces of the guide plate (13) and the telescopic sleeve (14), the end of the connecting support plate (18) is connected to the limit adjustment component (12), a T-shaped slide groove (11) is provided on the surface of the guide shaft sleeve (10), and the T-shaped slider (17) is slidably located inside the T-shaped slide groove (11); The surface of the telescopic sleeve (14) is provided with positioning pin holes (16) in a uniform array in a vertical straight line, the surface of the guide plate (13) is provided with a spring positioning pin (15) near the bottom end, and the spring positioning pin (15) is engaged with the positioning pin hole (16), and the surface of the telescopic sleeve (14) near the top and bottom surfaces of the server body (1) is provided with a T-shaped mounting block (21), and the T-shaped mounting block (21) is clearance-matched with the T-shaped mounting slot (6); The internal system structure of the server body (1) includes a main control system, an early warning system, a task management system and a resource management system, wherein: The main control system performs overall control over the various electrical components (20) and the circuit board body (19) inside the server body (1), issues instructions and monitors the start and completion of tasks, monitors the task management system and the resource management system, and issues warnings through the warning system based on the monitoring results: The early warning system receives the early warning signal from the main control system and connects with the early warning device to perform sound and light early warning; The task management system monitors the tasks received by the server body (1), and analyzes, decomposes, sets priorities and dynamically adjusts the order of the tasks through internal algorithm content; The resource management system monitors the resource status inside the server body (1) in real time, and intelligently controls and allocates the internal resources of the server body (1) to the corresponding tasks based on the task priority and task content of the task management system in combination with the internal algorithm structure, and monitors the resource structure in real time while the task is being executed, and intelligently adjusts the resource allocation based on the task execution status, time, and completion efficiency.

2. The server hardware architecture according to claim 1, characterized in that: The limit adjustment component (12) comprises an adjustment shaft (1202) and a limit tooth (1201); the adjustment shaft (1202) passes through the inside of the adjustment sleeve (9); one end of the adjustment shaft (1202) is engaged with the top end of the connecting support plate (18); the other end of the adjustment shaft (1202) is provided with a limit spring (1205); the end of the limit spring (1205) is provided with a sliding bearing (1204), and the sliding bearing (1204) is movably embedded in the inside of the guide sleeve (10); the surface of the adjustment shaft (1202) is provided with an adjustment gear (1203) in a linear array; one side of the inner wall of the guide sleeve (10) is provided with a limit tooth (1201) in a linear array; the limit tooth (1201) is meshed with the surface of the adjustment gear (1203).

3. The server hardware architecture according to claim 1, characterized in that: The surface of the mounting substrate (7) is evenly distributed with positioning screw holes (8); the electrical components (20) and the circuit board body (19) are connected to the surface of the mounting substrate (7) via the positioning screw holes (8); and the electrical components (20) and the circuit board body (19) are installed on the top and bottom surfaces of the mounting substrate (7).

4. The server hardware architecture according to claim 1, characterized in that: The edge of the dust cover plate (2) is connected to the front of the server body (1) by means of screws; a heat dissipation hole (5) is provided at the center of the back of the server body (1); heat dissipation arc grooves (4) are provided at the circumferential positions of the heat dissipation hole (5); the inner groove profiles of the heat dissipation hole (5) and the heat dissipation arc groove (4) are both conical grooves; and a connection interface (3) is provided on one side of the top surface of the server body (1).

5. The server hardware architecture according to claim 2, characterized in that: The spacing between adjacent adjusting gears (1203) is the same as the spacing between adjacent limiting teeth (1201), and the length of the limiting teeth (1201) is the same as the thickness of the adjusting gear (1203). The central axis of the adjusting shaft (1202), the central axis of the adjusting sleeve (9) and the central axis of the guide sleeve (10) coincide with each other. The maximum sliding spacing of the T-shaped slider (17) along the T-shaped slide groove (11) is the same as the tooth length of the limiting teeth (1201). One end of the adjusting shaft (1202) close to the connecting support plate (18) is located outside the guide sleeve (10). The connecting support plate (18) and the T-shaped slider (17) are both vertically mounted on the surfaces of the guide plate (13) and the telescopic sleeve (14).

6. The method for controlling the server hardware architecture according to any one of claims 1 to 5, characterized in that: The steps include: Sp1: Resource monitoring, the resource management system monitors the resources inside the server body (1) in real time, including the CPU, GPU, and memory, and the monitoring types include their temperature status, operating status, and occupancy rate; Sp2: Task analysis. The task management system receives external task instruction content and analyzes the instruction content in combination with the internal algorithm. The analysis categories include task type, task requirements, task dependencies, task completion time, and the amount of resources that may be occupied when the task is enabled. The task is weighted according to the algorithm and a weighted threshold is set. If the actual weighted threshold is less than the set weighted threshold, it is a simple task that can be executed directly and enters Sp5. Otherwise, it is a complex task and enters Sp3 for task decomposition. Sp3: Task decomposition, the task management system receives the complex task in Sp2 and the category of its analysis, and decomposes the task into multiple subtasks according to the completion process and steps of the task; Sp4: Task priority setting and dynamic adjustment. Analyze each subtask in Sp3 according to the analysis category in Sp2, and add task urgency, task importance, task time limit, task impact and the degree of correlation with other tasks to the analysis category. Weight each item in the analysis category to obtain the overall weighted value, and sort the weighted values ​​of multiple subtasks. The higher the sorting, the higher the priority. Sp5: Intelligent resource allocation. According to the resource monitoring content inside the server body (1), the CPU usage time data set is set as C, the GPU usage time data set is set as G, and the memory usage data set is set as T in the resource management system. Resource allocation is performed according to the priority of the subtasks set in Sp4. When allocating resources, the subtasks with high CPU, GPU and memory processing priorities with small C values, G values ​​and T values ​​during the current resource monitoring are selected. In the order of increasing C values, G values ​​and T values, the corresponding processing priorities are reduced in sequence. Sp6: Task execution and monitoring. The main control system starts the subtask after resource allocation and monitors the tasks during execution through the task management system and resource management system. According to the task completion status information and the real-time monitoring of current resources, the priority and resource allocation are dynamically adjusted until the task is completed. Sp7: Task evaluation and optimization iteration. The main control system evaluates the entire process of task completion. The evaluation content includes all operating status and adjustment status of each system in steps Sp1-Sp6, forms a database, and continuously optimizes and iterates the algorithms within the task management system and resource management system.

7. The control method according to claim 6, characterized in that: The resource management system makes the CPU usage time data set C = {C1, C2, ...C n }, where C1, C2, …C n The corresponding C value gradually increases, making the GPU usage time data set G = {G1, G2, ...G m }, G1, G2, …G m The corresponding G value gradually increases, and the memory usage data set T = {T1, T2, ...T p }, T1, T2, ...T p The corresponding T value gradually increases.

8. The control method according to claim 6, characterized in that: When allocating resources in the resource management system, the resources are allocated according to the resource categories mainly consumed in completing the subtasks.

Citation Information

Patent Citations

  • Data center server regulation and control method and device based on artificial intelligence

    CN113806077A

  • Artificial intelligence server hardware architecture based on two-way domestic CPU

    CN217587961U

  • Dynamic resource scheduling method and system under high-concurrency task complex scene

    CN115495231A

  • Replaceable modular baffle with different functions for director all-in-one machine case

    CN214896484U