An acceleration module and open acceleration computing system
By employing a dual-layer structure and optimized device distribution in the acceleration module, the size limitations of traditional OAM have been addressed, resulting in more efficient computing performance and reliability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHIHAOTONG (TIANJIN) INFORMATION TECHNOLOGY CO LTD
- Filing Date
- 2024-11-20
- Publication Date
- 2026-05-29
AI Technical Summary
Traditional OAM is limited by size, which prevents it from integrating more high-performance devices in the field of high-performance computing, thus limiting its application.
Design an acceleration module with a two-layer board structure, including a base plate, a first board and a second board, to achieve electrical interconnection through a connection structure, and equipped with a heat dissipation device and a power management system, and optimize the device layout to improve integration capability.
Integrating more components into a limited size structure improves the performance and heat dissipation of the acceleration module, thereby enhancing the overall performance of the computing system.
Smart Images
Figure CN122111938A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computing device technology, specifically to an acceleration module and an open accelerated computing system. Background Technology
[0002] With the rapid development of artificial intelligence technology, the demand for high-performance computing is growing. Traditional OAM (Open Accelerator Module) is limited by size, which prevents the integration of more high-performance devices, thus limiting the application of OAM in the field of high-performance computing. Summary of the Invention
[0003] In view of this, this application provides an acceleration module and an open accelerated computing system, which can integrate more devices on the acceleration module and improve working performance.
[0004] In a first aspect, embodiments of the present invention provide an acceleration module, comprising: a substrate; a first plate having at least one first device disposed thereon; a first connecting structure disposed on the lower side of the first plate and connected to the upper side of the substrate and the lower side of the first plate respectively; a second plate located above the first plate and having a preset distance between the second plate and the first plate, wherein at least one second device is disposed on the second plate; and a second connecting structure disposed between the first plate and the second plate and connected to the upper side of the first plate and the lower side of the second plate respectively.
[0005] In one specific embodiment, a heat dissipation device is further included, which is disposed on the first plate and / or the second plate; wherein the heat dissipation device includes a plurality of heat sinks, and heat dissipation channels are provided between adjacent heat sinks; and / or the heat dissipation device includes a fan; and / or the heat dissipation device includes a liquid cooling system.
[0006] In one specific implementation, a power management system is further included, which is used to provide power to the first device and the second device, and the power management system includes multiple power modules.
[0007] In one specific implementation, the power of the first device is less than a first power threshold, and the power of the second device is greater than a second power threshold; or the sum of the power of each first device on the first board is less than the first power threshold, and the sum of the power of each second device on the second board is greater than the second power threshold.
[0008] In one specific implementation, the first device is any of the following: an input / output controller, a processor; and / or the second device is any of the following: a GPU, a power module, DDR memory.
[0009] In one specific implementation, the first board is provided with at least one first interface, and the first device is provided with a first plug portion adapted to and connected to the first interface; and / or the second board is provided with at least one second interface, and the second device is provided with a second plug portion adapted to and connected to the second interface.
[0010] In one specific implementation, the first board is communicatively connected to the second board.
[0011] Secondly, embodiments of the present invention also provide an open accelerated computing system, the open accelerated computing system comprising: a chassis; a tray disposed within the chassis; and an acceleration module, the acceleration module being any of the acceleration modules described in the embodiments of this application, wherein the substrate of the acceleration module is detachably connected to the tray.
[0012] The acceleration module and open accelerated computing system provided by embodiments of the present invention include: a substrate, a first plate, a first connection structure, a second plate, and a second connection structure; the first plate has at least one first device; the first connection structure is disposed on the lower side of the first plate and is connected to the upper side of the substrate and the lower side of the first plate; the second plate is located above the first plate, and there is a preset distance between the second plate and the first plate, wherein the second plate has at least one second device; the second connection structure is disposed between the first plate and the second plate and is connected to the upper side of the first plate and the lower side of the second plate. This allows for the integration of more devices within the limited size of the acceleration module, improving the integration capability of the acceleration module and thus effectively enhancing its performance. Attached Figure Description
[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0014] Figure 1 This is a schematic diagram of an acceleration module provided in an embodiment of this application.
[0015] Explanation of key figure labels: 100 - Acceleration module; 10 - Substrate; 20 - First board; 201 - First device; 30 - First connection structure; 40 - Second board; 401 - Second device; 50 - Second connection structure. Detailed Implementation
[0016] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
[0017] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0018] Firstly, such as Figure 1 As shown, an embodiment of the present invention provides an acceleration module 100, which may include: a substrate 10, a first plate 20, a first connecting structure 30, a second plate 40, and a second connecting structure 50.
[0019] The acceleration module 100 in this embodiment can be OAM (Open Accelerator Module 100) conforming to the definition of OAI (Open Accelerator Infrastructure). The acceleration module 100 can be used to improve the performance and efficiency of processing, and achieve flexible compatibility and scalability between different systems, adapting to the ever-increasing high computing load and technological development, thus being applicable to a variety of fields, including AI inference, deep learning, machine learning, scientific simulation and data analysis application scenarios.
[0020] The substrate 10 can be a UBB (Universal Base Board) conforming to the OCP (Open Compute Project) design specifications. For example, the substrate 10 can be made of PCB (Printed Circuit Board). The dimensions and mechanical specifications of the substrate 10 conform to OCP standards to ensure compatibility with other OCP projects and facilitate manufacturing and installation. The power architecture of the substrate 10 supports multiple power options, including DC and AC, to meet the needs of different application scenarios. The substrate 10 can be configured with various data transmission interfaces, such as Ethernet, fiber optic communication, PCIe, etc., to meet the connection requirements of different devices. In addition, the substrate 10 can also support hardware management functions, including remote monitoring and fault diagnosis, to improve its operational reliability and maintainability.
[0021] The first board 20 and the second board 40 can also be made of PCB, with tiny apertures or line widths formed on the PCB to achieve dense wiring, thereby realizing electrical connection or signal transmission through the PCB. The second board 40 is located above the first board 20, and there is a preset distance between the second board 40 and the first board 20. The first board 20 has at least one first device 201, and the second board 40 has at least one second device 401. The first device 201 can be disposed on the upper side of the first board 20, and the second device 401 can be disposed on the upper side of the second board 40. The first device 201 and the second device 401 are hardware modules with corresponding functions, and are connected to the first board 20 and the second board 40 respectively in a detachable and replaceable manner. In addition, the first board 20 may also be provided with multiple first devices 201, which may be the same device or different devices; the second board 40 may also be provided with multiple second devices 401, which may be the same device or different devices.
[0022] The first connecting structure 30 is disposed on the lower side of the first plate 20, and the first connecting structure 30 is connected to the upper side of the substrate 10 and the lower side of the first plate 20 respectively; the second connecting structure 50 is disposed between the first plate 20 and the second plate 40, and the second connecting structure 50 is connected to the upper side of the first plate 20 and the lower side of the second plate 40 respectively. The first connecting structure 30 can be used to connect the first plate 20 and the substrate 10, realizing the electrical interconnection between the first plate 20 and the substrate 10. It can also act as a stiffener to provide better reinforcement support for the connection between the first plate 20 and the substrate 10, ensuring that the dimensional and positional tolerances of the first plate 20 and the substrate 10 after assembly meet the usage requirements and have good connection performance in various usage environments. Similarly, the second connecting structure 50 can be used to connect the second plate 40 and the first plate 20, realizing the electrical interconnection between the second plate 40 and the first plate 20 or between the second plate 40 and the substrate 10. It can also act as a stiffener to provide better reinforcement support for the connection between the second plate 40 and the first plate 20, ensuring that the dimensional and positional tolerances of the second plate 40 and the first plate 20 after assembly meet the usage requirements and have good connection performance in various usage environments, thereby ensuring the reliability of the operation of the acceleration module 100 structure.
[0023] The acceleration module 100 provided in the embodiments of the present invention includes: a substrate 10, a first plate 20, a first connecting structure 30, a second plate 40, and a second connecting structure 50; at least one first device 201 is provided on the first plate 20; the first connecting structure 30 is disposed on the lower side of the first plate 20 and is connected to the upper side of the substrate 10 and the lower side of the first plate 20 respectively; the second plate 40 is located above the first plate 20, and there is a preset distance between the second plate 40 and the first plate 20, wherein at least one second device 401 is provided on the second plate 40; the second connecting structure 50 is disposed between the first plate 20 and the second plate 40, and is connected to the upper side of the first plate 20 and the lower side of the second plate 40 respectively. This allows for the integration of more devices within the limited size of the acceleration module 100, improving the integration capability of the acceleration module 100 and thus effectively enhancing its performance.
[0024] Optionally, in one embodiment of the present invention, the acceleration module 100 further includes a heat dissipation device disposed on the first plate 20 and / or the second plate 40; wherein the heat dissipation device includes a plurality of heat sinks, and heat dissipation channels are formed between adjacent heat sinks; and / or the heat dissipation device includes a fan; and / or the heat dissipation device includes a liquid cooling system. For example, the heat dissipation device may be an extruded heatsink made of aluminum material, which may be disposed on the upper part of the second plate 40 to have a more unobstructed heat dissipation space. The heat dissipation device forms heat dissipation channels between adjacent heat sinks, so that the heat of each heat sink can be conducted outward in a timely manner, thereby improving the heat dissipation effect. Furthermore, depending on the operating conditions, the heat dissipation device may be further configured with a fan to improve the air cooling effect, or a liquid cooling system, etc., and the installation position of the heat dissipation device can be flexibly set according to the structural characteristics of the first plate 20 and the power of the configured first device 201, or the structural characteristics of the second plate 40 and the power of the configured second device 401. For example, a liquid cooling system can be arranged on the first plate 20 where the ventilation space is limited, or a fan can be arranged on the second plate 40 where ventilation is convenient, so as to improve the environmental adaptability and heat dissipation effect of the heat dissipation device.
[0025] Optionally, in one embodiment of the present invention, the acceleration module 100 further includes a power management system, which provides power to the first device 201 and the second device 401, and the power management system includes multiple power modules. The power management system can employ various technologies such as intelligent adjustment, energy-saving management, and real-time monitoring. For example, the power modules can use ultra-small size packaging to effectively save circuit board area, improve conversion efficiency to reduce power consumption and save energy, and ensure a stable power supply.
[0026] Since the first board 20 and the second board 40 of the acceleration module 100 form a double-layer structure, in order to further utilize the heat dissipation potential of the double-layer structure, optionally, in one embodiment of the present invention, the power of the first device 201 is less than a first power threshold, and the power of the second device 401 is greater than a second power threshold; or the sum of the power of each first device 201 on the first board is less than the first power threshold, and the sum of the power of each second device 401 on the second board is greater than the second power threshold. Since the first board 20 is located between the substrate 10 and the second board 40, the heat dissipation space is limited, while the upper part of the second board 40 has a relatively unobstructed heat dissipation space. Therefore, the first power threshold and the second power threshold can be specifically set according to the device arrangement requirements of the acceleration module 100. For example, the second power threshold can be greater than or equal to the first power threshold. In this way, a first device 201 with lower power can be arranged on the lower first board 20, and a second device 401 with higher power can be arranged on the upper second board 40. The power of each second device 401 on the second board 40 is greater than or equal to that of each first device 201 on the first board 20. This method takes the device as the analysis object of heat dissipation, thereby realizing differentiated heat dissipation management for devices with different power and improving the overall heat dissipation effect of the acceleration module 100. In some embodiments, specific devices need to be arranged on the first board or the second board. In this case, the acceleration module 100 can also be configured such that the sum of the power of each first device 201 on the first board is less than a first power threshold, and the sum of the power of each second device 401 on the second board is greater than a second power threshold. This method uses the first board and the second board as heat dissipation analysis objects, and also realizes differentiated heat dissipation management for the first board and the second board with different heat dissipation capabilities.
[0027] Optionally, in one embodiment of the present invention, the first device 201 is any of the following: an input / output controller, a processor; and / or the second device 401 is any of the following: a GPU, a power module, DDR memory. The first device 201 can be a processor, an input / output controller, a network interface, a basic device, etc. The processor (Central Processing Unit, CPU) serves as the core for computation and control, used for information processing and instruction execution. The input / output controller is used to perform operations such as digital conversion, data checking, buffering, and control transmission. The GPU (Graphics Processing Unit) is used for image and graphics-related computations. DDR memory, also known as DDR SDRAM (Double Data Rate SDRAM), provides directly addressable storage space. The input / output controller and processor have relatively low power consumption and can therefore be located on the first board 20, while the GPU, power module, and DDR memory have relatively high power consumption and can therefore be located on the second board 40, to fully utilize the different heat dissipation capabilities of the first board 20 and the second board 40, thereby improving the thermal reliability of the acceleration module 100. The second device 401 can also be a GPGPU (General Purpose Computing on Graphics Processing Units). As a general-purpose graphics processor, the GPGPU can be used for non-graphics computing tasks, such as physics calculations and AI training, to assist the CPU in performing complex non-graphics related calculations.
[0028] Optionally, in one embodiment of the present invention, the first board 20 is provided with at least one first interface, and the first device 201 is provided with a first plug-in portion adapted to connect to the first interface; and / or the second board 40 is provided with at least one second interface, and the second device 401 is provided with a second plug-in portion adapted to connect to the second interface. The first interface and the first plug-in portion, as well as the second interface and the second plug-in portion, can be standardized interfaces conforming to specifications to improve the modularity of the acceleration module 100, allowing the first device 201 on the first board 20 and the second device 401 on the second board 40 to be replaced or upgraded according to usage. Furthermore, the first board 20 may be provided with one or more first interfaces, which may be the same interface or different interfaces; the second board 40 may be provided with one or more second interfaces, which may be the same interface or different interfaces.
[0029] Optionally, in one embodiment of the present invention, the first board 20 and the second board 40 are communicatively connected. For example, high-speed interconnect technology can be used to achieve high-speed data transmission between the first board 20 and the second board 40, and between the first device 201 and the second device 401.
[0030] Secondly, embodiments of the present invention also provide an open accelerated computing system that can integrate more devices on the acceleration module 100 to improve performance.
[0031] The acceleration module 100 and open accelerated computing system provided in the embodiments of the present invention include a chassis, a tray and the acceleration module 100; the tray is disposed in the chassis; the acceleration module 100 is any of the acceleration modules 100 described in the embodiments of this application, wherein the base plate 10 of the acceleration module 100 is detachably connected to the tray.
[0032] The open accelerated computing system provided in the embodiments of the present invention includes: a chassis, a tray, and an acceleration module 100. The acceleration module 100 is configured with a substrate 10, a first plate 20, a first connecting structure 30, a second plate 40, and a second connecting structure 50. At least one first device 201 is provided on the first plate 20. The first connecting structure 30 is disposed on the lower side of the first plate 20 and is connected to the upper side of the substrate 10 and the lower side of the first plate 20, respectively. The second plate 40 is located above the first plate 20, and a predetermined distance exists between the second plate 40 and the first plate 20. At least one second device 401 is provided on the second plate 40. The second connecting structure 50 is disposed between the first plate 20 and the second plate 40, and is connected to the upper side of the first plate 20 and the lower side of the second plate 40, respectively. This system enables the integration of more devices within the limited size of the acceleration module 100, improving its integration capability and thus effectively enhancing its performance.
[0033] The aforementioned open accelerated computing systems exist in various forms, including but not limited to: (1) Mobile communication devices: These devices are characterized by their mobile communication capabilities and are primarily designed to provide voice and data communication. These terminals include smartphones (such as iPhones), multimedia phones, feature phones, and low-end phones.
[0034] (2) Ultra-mobile personal computer devices: These devices fall under the category of personal computers, possessing computing and processing capabilities, and generally also have mobile internet access features. These terminals include PDAs, MIDs, and UMPCs, such as the iPad.
[0035] (3) Portable entertainment devices: These devices can display and play multimedia content. This category includes audio and video players (such as iPods), handheld game consoles, e-book readers, as well as smart toys and portable car navigation devices.
[0036] (4) Server: A device that provides computing services. The components of a server include a processor, hard disk, memory, system bus, etc. Servers are similar to general computer architectures, but because they need to provide highly reliable services, they have higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.
[0037] (5) Other electronic devices with data interaction functions.
[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0039] The various embodiments in this specification are described in a related manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
[0040] In particular, the device embodiment is basically similar to the method embodiment, so the description is relatively simple. For relevant details, please refer to the description of the method embodiment.
[0041] For ease of description, the above apparatus is described by dividing it into various functional units / modules. Of course, in implementing this invention, the functions of each unit / module can be implemented in one or more software and / or hardware.
[0042] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0043] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. An acceleration module, characterized in that, include: substrate; A first plate, wherein at least one first device is provided on the first plate; A first connecting structure is disposed on the lower side of the first plate, and the first connecting structure is connected to the upper side of the substrate and the lower side of the first plate, respectively. The second plate is located above the first plate, and there is a preset distance between the second plate and the first plate. The second plate is provided with at least one second device. The second connecting structure is disposed between the first plate and the second plate, and the second connecting structure is connected to the upper side of the first plate and the lower side of the second plate, respectively.
2. The acceleration module as described in claim 1, characterized in that, It also includes a heat dissipation device, which is disposed on the first plate and / or the second plate; The heat dissipation device includes multiple heat sinks, and there are heat dissipation channels between adjacent heat sinks; and / or the heat dissipation device includes a fan; and / or the heat dissipation device includes a liquid cooling system.
3. The acceleration module as described in claim 1, characterized in that, It also includes a power management system for providing power to the first device and the second device, wherein the power management system includes multiple power modules.
4. The acceleration module as described in claim 1, characterized in that, The power of the first device is less than a first power threshold, and the power of the second device is greater than a second power threshold; or The sum of the power of each first device on the first board is less than a first power threshold, and the sum of the power of each second device on the second board is greater than a second power threshold.
5. The acceleration module as described in claim 1, characterized in that, The first device is any of the following: an input / output controller, a processor; and / or the second device is any of the following: a GPU, a power module, DDR memory.
6. The acceleration module as described in claim 1, characterized in that, The first board has at least one first interface, and the first device has a first plug-in portion adapted to and connected to the first interface; and / or the second board has at least one second interface, and the second device has a second plug-in portion adapted to and connected to the second interface.
7. The acceleration module as described in claim 1, characterized in that, The first board is communicatively connected to the second board.
8. An open accelerated computing system, characterized in that, include: Chassis; A tray, which is disposed within the chassis; An acceleration module, wherein the acceleration module is the acceleration module according to any one of claims 1-7, wherein the base plate of the acceleration module is detachably connected to the tray.