Memory system, memory resource adjustment method and apparatus, and electronic device and medium

Through the CXL protocol and high-speed interconnection channel interface, multiple server memory resource sharing and dynamic allocation are realized, solving the problem of insufficient server memory resources, expanding the memory pool capacity, optimizing the DRAM to kernel ratio, and improving computing resource capabilities.

WO2024230193A1PCT designated stage expired Publication Date: 2024-11-14INSPUR SUZHOU INTELLIGENT TECH CO LTD

Patent Information

Application Number
PCT/CN2023/142079
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-06
Filing Date
2023-12-26
Publication Date
2024-11-14

AI Technical Summary

Technical Problem

Servers are facing insufficient memory resources, especially after cloud computing applications are widely used, the memory resource demand in data centers has increased, and it is difficult for existing technology to effectively expand and optimize memory resource configuration, resulting in insufficient memory pool capacity and excessive DRAM and CORE ratio. Low problem.

Method used

By adopting the CXL protocol and high-speed interconnect channel interface, memory resource sharing and dynamic allocation between multiple servers is established, high-speed interconnect signals are converted into PCIE physical links using connection cables, and chips that support the CXL protocol convert IPB resources into JEDEC standard DIMM memory resources, use external boards to expand DRAM resources, and optimize the motherboard space layout.

Benefits of technology

It realizes effective sharing and dynamic allocation of memory resources between multiple servers, expands the memory pool capacity, optimizes the DRAM to kernel ratio, improves the ability to process storage and computing resources, and reduces system costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023142079_14112024_PF_FP_ABST
    Figure CN2023142079_14112024_PF_FP_ABST
Patent Text Reader

Abstract

Provided in the embodiments of the present application are a memory system, a memory resource adjustment method and apparatus, and an electronic device and a non-volatile readable storage medium. The memory system comprises: a first motherboard; a processor and a high-speed interconnection path interface, which are integrated on the first motherboard; a first adapter board; a high-speed interconnection path interface, a chip and a memory module, which are integrated on the first adapter board; and at least one connecting cable, wherein the connecting cable is used for establishing a physical link between the high-speed interconnection path interface of the first motherboard and the high-speed interconnection path interface of the first adapter board, and is used for converting a high-speed path interconnection signal into an intra-processor dedicated bus signal which is transmitted over a physical link based on a peripheral component interconnect express, and the chip of the first adapter board is used for acquiring the converted intra-processor dedicated bus signal and converting the converted intra-processor dedicated bus signal into a double-rate memory signal. High-speed path interconnection signals are converted into IPB resources by means of connecting cables, thereby expanding DRAM resources.
Need to check novelty before this filing date? Find Prior Art

Description

Memory system, memory resource adjustment method, device, electronic device and medium

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to the Chinese patent application filed with the China Patent Office on May 6, 2023, with application number 202310499307.X, and application name “Memory system, memory resource adjustment method, device, electronic device and medium”, all contents of which are incorporated by reference into this application. Technical Field

[0003] The present application relates to the field of computer technology, and in particular to a memory system, a memory resource adjustment method, a memory resource adjustment device, an electronic device, and a computer non-volatile readable storage medium. Background Art

[0004] Memory, also known as internal storage, temporarily stores processor data and data exchanged with external storage devices like hard drives. It bridges the gap between the processor and the outside world and is an integral part of the entire computing unit. Memory acts as a buffer between the CPU (Central Processing Unit) and the hard drive, serving as temporary storage for data. Programs are dispatched to memory while they run, and data is automatically released from memory when the server or program is shut down.

[0005] With the development of cloud computing applications, informatization has gradually permeated all areas of society. People's daily work and life are increasingly conducted online, and the amount of network data is constantly increasing. Server memory is also facing numerous challenges, and insufficient memory resources are often encountered in server configurations.

[0006] Summary of the Invention

[0007] In view of the above problems, embodiments of the present application are proposed to provide a memory system, a memory resource adjustment method, a memory resource adjustment device, an electronic device and a computer non-volatile readable storage medium that overcome the above problems or at least partially solve the above problems.

[0008] The present application discloses a memory system, including:

[0009] a first motherboard, a processor integrated in the first motherboard, and a high-speed interconnect channel interface;

[0010] A first adapter board, integrating a high-speed interconnect channel interface, a chip, and a memory module on the first adapter board;

[0011] at least one connecting cable;

[0012] A connecting cable, configured to establish a physical link between the high-speed interconnect channel interface of the first mainboard and the high-speed interconnect channel interface of the first adapter board, thereby establishing a communication connection between the first mainboard and the first adapter board; wherein the connection between the high-speed interconnect channel interface of the first mainboard and the high-speed interconnect channel interface of the first adapter board is configured to transmit a high-speed channel interconnect signal;

[0013] The connecting cable is also used to convert high-speed channel interconnect signals into processor internal dedicated bus signals transmitted on a physical link based on a high-speed serial computer expansion bus standard;

[0014] The chip of the first adapter board is used to obtain the converted internal dedicated bus signal of the processor and convert the internal dedicated bus signal of the processor into a double-rate memory signal, so that the memory resources of the memory module of the first adapter board can be accessed and used by the processor of the first mainboard.

[0015] In some embodiments, further comprising:

[0016] a second motherboard, a processor and a high-speed interconnection channel interface integrated on the second motherboard;

[0017] The connecting cable is also used to establish a physical link between the high-speed interconnection channel interface of the first mainboard and the high-speed interconnection channel interface of the second mainboard, so as to establish a communication connection between the first mainboard and the second mainboard.

[0018] In some embodiments, the connection cable includes a first connection cable and a second connection cable, and further includes:

[0019] a first connecting cable, used for transmitting data between the processor of the first mainboard and the processor of the second mainboard;

[0020] The second connecting cable is used to transmit logic control signals between the first mainboard and the second mainboard.

[0021] In some embodiments, further comprising:

[0022] Miscellaneous signal transmission interface integrated into the first mainboard;

[0023] Miscellaneous signal transmission interface integrated in the first adapter board;

[0024] The connecting cable is also used to establish a physical link between the miscellaneous signal transmission interface of the first mainboard and the miscellaneous signal transmission interface of the first adapter board; wherein, the connection between the miscellaneous signal transmission interface of the first mainboard and the miscellaneous signal transmission interface of the first adapter board is used to transmit miscellaneous signals.

[0025] In some embodiments, the miscellaneous signals include a clock signal and further include:

[0026] The chip of the first adapter board is also used to obtain a clock signal.

[0027] In some embodiments, the miscellaneous signals include a chip enable signal and further include:

[0028] A field programmable logic gate array module integrated into the first mainboard;

[0029] The field programmable logic gate array module of the first mainboard is used to send a chip enable signal;

[0030] The chip of the first adapter board is also used to start working after obtaining the chip enable signal, and convert the processor's internal dedicated bus signal into a double-rate memory signal.

[0031] In some embodiments, the miscellaneous signal includes a power-on completion signal and further includes:

[0032] The memory module of the first adapter board is used to send a power-on completion signal after power-on is completed, and transmit the power-on completion signal to the field programmable logic gate array module of the first main board; and is also used to perform in-situ detection after obtaining the double-rate memory signal to generate a corresponding detection signal.

[0033] In some embodiments, the miscellaneous signal includes a detection signal and further includes:

[0034] A baseboard management controller module integrated into the first mainboard;

[0035] The baseboard management controller module of the first mainboard is used to obtain a detection signal and determine a system operating state based on the detection signal, and initialize the chip of the first adapter board when it is determined that the system operating state is normal.

[0036] In some embodiments, the miscellaneous signals include a reset signal and further include:

[0037] The field programmable logic gate array module of the first main board is further used to send a reset signal when initializing the chip of the first adapter board;

[0038] The memory module of the first adapter board is further used to provide corresponding memory resources to the processor of the first mainboard for access and use after obtaining the reset signal.

[0039] In some embodiments, further comprising:

[0040] A second adapter board, integrating a high-speed interconnect channel interface, miscellaneous signal transmission interface, chip and memory module on the second adapter board;

[0041] The connecting cable is further used to establish a physical link between the high-speed interconnect channel interface of the second mainboard and the high-speed interconnect channel interface of the second adapter board, and to establish a physical link between the miscellaneous signal transmission interface of the second mainboard and the miscellaneous signal transmission interface of the second adapter board, so as to establish a communication connection between the second mainboard and the second adapter board; wherein the connection between the high-speed interconnect channel interface of the second mainboard and the high-speed interconnect channel interface of the second adapter board is used to transmit high-speed channel interconnection signals;

[0042] The connecting cable is also used to convert high-speed channel interconnect signals into processor internal dedicated bus signals transmitted on a physical link based on a high-speed serial computer expansion bus standard;

[0043] The chip of the second adapter board is used to obtain the converted internal dedicated bus signal of the processor and convert the internal dedicated bus signal of the processor into a double-rate memory signal so that the memory resources of the memory module of the second adapter board can be accessed and used by the processor of the second mainboard.

[0044] In some embodiments, the first mainboard is communicatively connected to a plurality of first adapter boards; and the second mainboard is communicatively connected to a plurality of second adapter boards.

[0045] In some embodiments, the memory module of the first riser board is the first memory module or the second memory module; the memory module of the second riser board is the first memory module or the second memory module.

[0046] In some embodiments, the first mainboard integrates multiple processors, and the processors are interconnected to transmit internal dedicated bus signals of the processors, and the first mainboard is communicatively connected to multiple first adapter boards, further comprising:

[0047] Each processor of the first mainboard is used to call the memory resources of the memory modules of each first adapter board.

[0048] The present application also discloses a memory resource adjustment method, which is applied to a memory system. The memory system includes a first mainboard, a first adapter board, a second mainboard, and a second adapter board. The method includes:

[0049] Determine the current load information of the memory system and determine the target memory resources to be provided based on the load information;

[0050] Based on the target memory resources, determine whether to control the first mainboard to be connected to the first adapter board, and / or determine whether to control the second mainboard to be connected to the second adapter board, and / or determine the number of first adapter boards connected to the first mainboard, and / or determine the number of second adapter boards connected to the second mainboard.

[0051] In some embodiments, determining whether to control the first mainboard to be connected to the first riser board, and / or determining whether to control the second mainboard to be connected to the second riser board, and / or determining the number of first riser boards connected to the first mainboard, and / or determining the number of second riser boards connected to the second mainboard based on the target memory resource includes:

[0052] If the target memory resources do not exceed the memory resources provided by the first mainboard and the second mainboard, the first mainboard is controlled to be disconnected from the first adapter board, and the second mainboard is controlled to be disconnected from the second adapter board, so that only the first mainboard and the second mainboard are responsible for providing memory resources.

[0053] In some embodiments, determining whether to control the first mainboard to be connected to the first riser board, and / or determining whether to control the second mainboard to be connected to the second riser board, and / or determining the number of first riser boards connected to the first mainboard, and / or determining the number of second riser boards connected to the second mainboard based on the target memory resource includes:

[0054] If the target memory resources exceed the memory resources provided by the first main board and the second main board, and do not exceed the preset first memory resource usage threshold, the first main board is controlled to be connected to the first adapter board, and the second main board is controlled to be disconnected from the second adapter board, so that the first main board, the second main board and at least one first adapter board are responsible for providing memory resources.

[0055] In some embodiments, determining whether to control the first mainboard to be connected to the first riser board, and / or determining whether to control the second mainboard to be connected to the second riser board, and / or determining the number of first riser boards connected to the first mainboard, and / or determining the number of second riser boards connected to the second mainboard based on the target memory resource includes:

[0056] If the target memory resource exceeds the preset first memory resource usage threshold and does not exceed the preset second memory resource usage threshold, the first main board is controlled to be connected to the first adapter board, and the second main board is controlled to be connected to the second adapter board, so that the first main board, the second main board, at least one first adapter board and at least one second adapter board are responsible for providing memory resources.

[0057] In some embodiments, the method further comprises:

[0058] If the target memory resource exceeds a preset second memory resource usage threshold, the memory module of the first riser board and / or the second riser board is configured as a second memory module.

[0059] The present application also discloses a memory resource adjustment device, which is applied to a memory system. The memory system includes a first mainboard, a first adapter board, a second mainboard, and a second adapter board. The device includes:

[0060] A first determining module is used to determine the current load information of the memory system and determine the target memory resources to be provided according to the load information;

[0061] The second determination module is used to determine whether to control the first mainboard to be connected to the first adapter board based on the target memory resources, and / or determine whether to control the second mainboard to be connected to the second adapter board, and / or determine the number of first adapter boards connected to the first mainboard, and / or determine the number of second adapter boards connected to the second mainboard.

[0062] In some embodiments, the second determining module includes:

[0063] The first control submodule is used to control the first mainboard to be disconnected from the first adapter board, and to control the second mainboard to be disconnected from the second adapter board if the target memory resources do not exceed the memory resources provided by the first mainboard and the second mainboard, so that only the first mainboard and the second mainboard are responsible for providing memory resources.

[0064] In some embodiments, the second determining module includes:

[0065] The second control submodule is used to control the first mainboard to be connected to the first adapter board, and to control the second mainboard to be disconnected from the second adapter board if the target memory resources exceed the memory resources provided by the first mainboard and the second mainboard, and do not exceed the preset first memory resource usage threshold, so that the first mainboard, the second mainboard and at least one first adapter board are responsible for providing memory resources.

[0066] In some embodiments, the second determining module includes:

[0067] The third control submodule is used to control the first mainboard to be connected to the first adapter board, and to control the second mainboard to be connected to the second adapter board if the target memory resource exceeds the preset first memory resource usage threshold and does not exceed the preset second memory resource usage threshold, so that the first mainboard, the second mainboard, at least one first adapter board and at least one second adapter board are responsible for providing memory resources.

[0068] In some embodiments, the apparatus further comprises:

[0069] The configuration module is configured to configure the memory module of the first adapter board and / or the second adapter board as the second memory module if the target memory resource exceeds a preset second memory resource usage threshold.

[0070] The embodiments of the present application include the following advantages:

[0071] In an embodiment of the present application, the memory system includes a first motherboard, a processor and a high-speed interconnect channel interface integrated on the first motherboard, and also includes a first adapter board, a high-speed interconnect channel interface, a chip and a memory module integrated on the first adapter board. A communication connection between the first motherboard and the first adapter board can be established by connecting a cable. The connecting cable can convert the high-speed channel interconnect signal into a processor internal dedicated bus signal transmitted based on a physical link of a high-speed serial computer expansion bus standard. Then, the chip of the first adapter board can convert the processor internal dedicated bus signal into a double-rate memory signal, so that the memory resources of the memory module of the first adapter board can be accessed and used by the processor of the first motherboard. By adopting the above design, the high-speed channel interconnect signal is converted into an IPB resource that can go through a PCIE physical link through a connecting cable, and the IPB resource is converted into a DIMM memory resource that complies with the JEDEC (Joint Electron Device Engineering Council, Joint Electronic Device Engineering Council) protocol standard through a chip that supports the CXL protocol. Using an external board to expand DRAM resources so that the DIMM resources are not on the same motherboard as the CPU can optimize the spatial layout resources of the motherboard. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] FIG1 is a system architecture diagram of a memory system;

[0073] FIG2 is a system architecture diagram of another memory system;

[0074] FIG3 is a system architecture diagram of another memory system;

[0075] FIG4 is a schematic diagram of a memory system memory module insertion method;

[0076] FIG5 is a structural block diagram of a memory system provided in an embodiment of the present application;

[0077] FIG6 is a system architecture diagram of a memory system according to an embodiment of the present application;

[0078] FIG7 is a structural block diagram of another memory system provided in an embodiment of the present application;

[0079] FIG8 is a system architecture diagram of another memory system according to an embodiment of the present application;

[0080] FIG9 is a system architecture diagram of another memory system according to an embodiment of the present application;

[0081] FIG10 is a schematic diagram of signal transmission of a memory system according to an embodiment of the present application;

[0082] FIG11 is a system architecture diagram of yet another memory system according to an embodiment of the present application;

[0083] FIG12 is a schematic diagram of an adapter plate designed as a 2DPC according to an embodiment of the present application;

[0084] FIG13 is a structural block diagram of another memory system provided in an embodiment of the present application;

[0085] 14 is a system architecture diagram of a memory system in which a first motherboard is integrated with four processors according to an embodiment of the present application;

[0086] 15 is a system architecture diagram of a memory system in which a first motherboard is integrated with eight processors according to an embodiment of the present application;

[0087] FIG16 is a flowchart of a method for adjusting memory resources according to an embodiment of the present application;

[0088] FIG17 is a structural block diagram of a memory resource adjustment device provided in an embodiment of the present application;

[0089] FIG18 is a structural block diagram of an electronic device provided in an embodiment of the present application;

[0090] FIG19 is a structural block diagram of a computer non-volatile readable storage medium provided in an embodiment of the present application. DETAILED DESCRIPTION

[0091] To make the above-mentioned objects, features, and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the embodiments described are only some of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field fall within the scope of protection of this application.

[0092] In related technologies, the link directly from the processor to the memory currently uses DDR5 (DDR5 SDRAM, a computer memory specification). When the memory pool is insufficient, DDR5 can be configured from 1DPC (DIMM Per Channel) to 2DPC or even 3DPC, or DDR5 memory modules with larger memory capacity can be configured. Alternatively, when memory cannot be further expanded, data can be preferentially exchanged within the server to free up more memory space. Therefore, many manufacturers have added CPU interconnect channels and channels that can directly connect to DDR5 in their next-generation CPUs.

[0093] As shown in Figure 1, this is a system architecture diagram of a memory system. Each CPU supports 8 DIMM (Dual-Inline-Memory-Modules) channels, and the two CPUs are interconnected by a 4x24 UPI (Ultra Path Interconnect). Based on the current maximum configuration of DDR5 (256GB), the maximum capacity of the memory pool of this two-way server (1DPC) is 4096GB (4TB). This memory system composed of two-way servers also includes an FPGA (Field Programmable Gate Array) module, a PCH (Platform Controller Hub, integrated south bridge) module, BOOT LOGIC (server startup logic, indicating the order in which logical control signals occur), and a BMC (Baseboard Manager Controller) module. The FPGA module is a digital integrated circuit that uses software to modify and configure the device's internal connection structure and logic units to complete the established design function. The basic principle of FPGAs is to integrate a large number of digital circuit gates, memory, and interconnect resources within the chip. Users can define the functions of these gates and the connections between modules through FPGA programming. The BMC (Baseboard Management Controller) module is a specialized service processor that uses sensors to monitor the status of a computer, network server, or other hardware-driven device.

[0094] Figure 2 shows the architecture of another memory system. Each CPU supports 12 DIMM channels, and the UPIs connecting the two CPUs are 6x24. Based on the current maximum DDR5 configuration (256GB), the maximum memory pool capacity of this two-socket server (1DPC) is 6144GB (6TB). This memory system, comprised of two-socket servers, also includes an FPGA module and a BMC module.

[0095] Figure 3 shows the architecture of another memory system. Each CPU can accommodate eight DIMMs, and the two CPUs are interconnected using three xGMIs (Global Memory Interconnects). Based on the current maximum DDR5 configuration (256GB), the maximum memory pool capacity of this two-socket server (1DPC) is 4096GB (4TB). This memory system, comprised of two socket servers, also includes a CPLD (Complex Programmable Logic Device) module and a BMC module.

[0096] Figure 4 shows a schematic diagram of memory module insertion in a memory system. DDR5 supports 2-3 DIMMs per channel, but this can introduce other disadvantages. DRAM (Dynamic Random Access Memory) typically uses a transistor and a capacitor to represent a bit. Both main types of random access memory (dynamic and static) lose their stored data when power is removed.

[0097] For the memory systems shown in Figures 1 and 2, when memory needs to be increased, memory channels need to be added. However, this eventually reaches the maximum value that can be increased and cannot be further optimized. In addition, the smaller the processor process, the more chips can be integrated per unit area, making it difficult to meet heat dissipation requirements.

[0098] The memory system in Figure 3 is difficult to deploy and does not easily meet the requirements of current server chassis. The design of circuit board layout and wiring is too difficult. 2DPC is feasible, but 3DPC is too unfeasible. When switching from 1DPC to 2DPC, the DDR (Double Data Rate SDRAM) operating rate will drop from 4800-5600MT / s in 1DPC to 4400MT / s, and the transmission delay will increase by tens of nanoseconds, which is not advisable for commercial users such as cloud users.

[0099] With the development of cloud computing applications, informatization has gradually covered all areas of society. Data center servers often face insufficient memory resources.

[0100] DRAM represents a significant cost in every data center. However, servers are not homogeneous, and workloads are dynamic, constantly changing and evolving. The combination of compute resources, DRAM, NAND (NAND flash memory), and network types varies depending on the workload. Cloud providers offer a wide variety of configurations, attempting to optimize their hardware for different workloads. To address this, the Compute Express Link (CXL) protocol was proposed. CXL is an industry-supported cache-coherent interconnect protocol for processors, memory expansion, and accelerators. CXL technology maintains consistency between the CPU memory space and memory on attached devices, allowing resource sharing for higher performance, reducing software stack complexity, and lowering overall system costs. This enables users to focus simply on the target workload, rather than on redundant memory management hardware in the accelerator.

[0101] To address the above issues, this application proposes a memory system that overcomes or at least partially resolves them. Using CXL technology, multiple servers can share a portion of memory, dynamically allocating it to different servers. Rather than overprovisioning servers, it's better to configure them to a closer average DRAM-to-core ratio. This allows customers to address excess DRAM requirements through a memory pool, which communicates via the CXL protocol.

[0102] One of the core concepts of the embodiments of the present application is that the memory system includes a first motherboard, a processor and a high-speed interconnect channel interface integrated on the first motherboard, and also includes a first adapter board, a high-speed interconnect channel interface, a chip and a memory module integrated on the first adapter board. A communication connection between the first motherboard and the first adapter board can be established through a connecting cable. The connecting cable can convert the high-speed channel interconnect signal into a processor internal dedicated bus signal transmitted through a physical link based on the high-speed serial computer expansion bus standard. Then, the chip of the first adapter board can convert the processor internal dedicated bus signal into a double-rate memory signal, so that the memory resources of the memory module of the first adapter board can be accessed and used by the processor of the first motherboard. By adopting the above design, the high-speed channel interconnect signal is converted into an IPB resource that can go through a PCIE physical link through a connecting cable, and the IPB resource is converted into a DIMM memory resource that complies with the JEDEC protocol standard through a chip that supports the CXL protocol. Using an external board to expand DRAM resources so that the DIMM resources are not on the same motherboard as the CPU can optimize the spatial layout resources of the motherboard.

[0103] 5 , a block diagram of a memory system according to an embodiment of the present application is shown. The memory system 501 includes:

[0104] A first motherboard 5011, a processor integrated in the first motherboard 5011, and a high-speed interconnect channel interface;

[0105] A first adapter board 5012 , integrating a high-speed interconnect channel interface, a chip, and a memory module on the first adapter board 5012 ;

[0106] At least one connecting cable 5013;

[0107] A connecting cable 5013 is used to establish a physical link between the high-speed interconnect channel interface of the first mainboard 5011 and the high-speed interconnect channel interface of the first adapter board 5012, thereby establishing a communication connection between the first mainboard 5011 and the first adapter board 5012; wherein the connection between the high-speed interconnect channel interface of the first mainboard 5011 and the high-speed interconnect channel interface of the first adapter board 5012 is used to transmit high-speed channel interconnect signals;

[0108] The connection cable 5013 is also used to convert the high-speed channel interconnection signal into a processor internal dedicated bus signal transmitted on a physical link based on the high-speed serial computer expansion bus standard;

[0109] The chip of the first adapter board 5012 is used to obtain the converted processor internal dedicated bus signal and convert the processor internal dedicated bus signal into a double-rate memory signal so that the memory resources of the memory module of the first adapter board 5012 can be accessed and used by the processor of the first mainboard 5011.

[0110] Specifically, the high-speed interconnect channel interface can be a UPI interface or an IPB interface, the connecting cable can be a Cable cable, the physical link based on the high-speed serial computer expansion bus standard is specifically a PCIE physical link, the processor's internal dedicated bus signal is specifically an IPB signal, and the double-rate memory signal is specifically a DDR signal.

[0111] Multiple servers can share a portion of memory, and it can be dynamically allocated to different servers. Instead of over-provisioning servers, they can be configured closer to the average DRAM-to-core ratio, and customers' excess DRAM needs can be addressed through a memory pool. This memory pool will communicate via the CXL protocol.

[0112] IPB (Inter-processor Bus (Intel UPI, AMD xGMI, etc.)) is a protocol trace for interconnecting CPUs within a server. Taking Intel's UPI as an example, UPI is essentially a dedicated high-speed transmission protocol and an IPB resource. Compared to ordinary PCIE information transmission, it lacks management signals and clock signals. By adding a capacitor to the TX end of a cable, UPI can be converted into an IPB signal transmitted over a PCIE (peripheral component interconnect express) physical link. Then, according to the CXL protocol, the IPB signal traveling along the PCIE physical link can be converted into a standard DDR signal through a dedicated conversion chip. This allows UPI to be expanded to include more memory and pool the entire memory. The aforementioned high-speed channel interconnect signal can be a UPI signal.

[0113] In the memory system of the embodiment of the present application, the memory resources on the first adapter board can be provided to the first mainboard through signal conversion. As shown in Figure 6, a system architecture diagram of a memory system of the embodiment of the present application is shown. On the mainboard (MB), the UPI signal is transmitted to the UPI CON (high-speed interconnect channel interface), and other miscellaneous signals including (SMB, PRSNT, 100M_CLKDP\DN, etc.) are transmitted to the miscellaneous signal transmission interface (Slimline interface). All signals are transmitted to the switch board (adapter board) via a cable, and then transmitted to the switch IC (chip) via the internal wiring of the board. After conversion, the corresponding signal is transmitted to the corresponding DIMM SLOT (memory module).

[0114] In summary, in an embodiment of the present application, the memory system includes a first motherboard, a processor and a high-speed interconnect channel interface integrated on the first motherboard, and also includes a first adapter board, a high-speed interconnect channel interface, a chip and a memory module integrated on the first adapter board. A communication connection between the first motherboard and the first adapter board can be established through a connecting cable. The connecting cable can convert the high-speed channel interconnect signal into a processor internal dedicated bus signal transmitted based on a physical link of the high-speed serial computer expansion bus standard. Then, the chip of the first adapter board can convert the processor internal dedicated bus signal into a double-rate memory signal, so that the memory resources of the memory module of the first adapter board can be accessed and used by the processor of the first motherboard. By adopting the above design, the high-speed channel interconnect signal is converted into an IPB resource that can go through a PCIE physical link through a connecting cable, and the IPB resource is converted into a DIMM memory resource that complies with the JEDEC protocol standard through a chip that supports the CXL protocol. Using an external board to expand DRAM resources so that the DIMM resources are not on the same motherboard as the CPU can optimize the spatial layout resources of the motherboard.

[0115] 7 , a block diagram of another memory system according to an embodiment of the present application is shown. The memory system 701 includes:

[0116] A first motherboard 7011, a processor integrated in the first motherboard 7011, and a high-speed interconnect channel interface;

[0117] A first adapter board 7012 , integrating a high-speed interconnect channel interface, a chip, and a memory module on the first adapter board 7012 ;

[0118] At least one connecting cable 7013;

[0119] A connecting cable 7013 is used to establish a physical link between the high-speed interconnect channel interface of the first mainboard 7011 and the high-speed interconnect channel interface of the first adapter board 7012, thereby establishing a communication connection between the first mainboard 7011 and the first adapter board 7012; wherein the connection between the high-speed interconnect channel interface of the first mainboard 7011 and the high-speed interconnect channel interface of the first adapter board 7012 is used to transmit high-speed channel interconnect signals;

[0120] The connection cable 7013 is also used to convert the high-speed channel interconnection signal into a processor internal dedicated bus signal transmitted on a physical link based on the high-speed serial computer expansion bus standard;

[0121] The chip of the first adapter board 7012 is used to obtain the converted processor internal dedicated bus signal and convert the processor internal dedicated bus signal into a double-rate memory signal so that the memory resources of the memory module of the first adapter board 7012 can be accessed and used by the processor of the first mainboard 7011.

[0122] The second mainboard 7014 includes a processor and a high-speed interconnection channel interface integrated therein;

[0123] The connecting cable 7013 is also used to establish a physical link between the high-speed interconnection channel interface of the first mainboard 7011 and the high-speed interconnection channel interface of the second mainboard 7014, so as to establish a communication connection between the first mainboard 7011 and the second mainboard 7014.

[0124] In the memory system of the embodiment of the present application, the two mainboards can belong to different servers, and the entire dual-way server can be converted into two single-way servers through cable interconnection, so that the important IPB resources in the system are no longer traditional dedicated resources, but can be connected to other places through cable for more efficient use.

[0125] In one embodiment of the present application, the connection cable includes a first connection cable and a second connection cable, and the memory system 701 further includes:

[0126] a first connecting cable, used for transmitting data between the processor of the first mainboard and the processor of the second mainboard;

[0127] The second connecting cable is used to transmit logic control signals between the first mainboard and the second mainboard.

[0128] Figure 8 shows the system architecture of another memory system according to an embodiment of the present application. Under normal circumstances, the system's memory resources are sufficient, and no memory optimization or expansion is required. The UPI CONs (high-speed interconnect channel interfaces) on the first motherboard (MB0) and the second motherboard (MB1) are interconnected via cables, forming a dual-channel system with two single-channel subsystems. Information between the CPUs is transmitted via Cable 4 (the first connecting cable), and logic control signals between MB0 and MB1 are transmitted via Cable 3 (the second connecting cable).

[0129] In one embodiment of the present application, the memory system 701 further includes:

[0130] Miscellaneous signal transmission interface integrated into the first mainboard;

[0131] Miscellaneous signal transmission interface integrated in the first adapter board;

[0132] The connecting cable is also used to establish a physical link between the miscellaneous signal transmission interface of the first mainboard and the miscellaneous signal transmission interface of the first adapter board; wherein, the connection between the miscellaneous signal transmission interface of the first mainboard and the miscellaneous signal transmission interface of the first adapter board is used to transmit miscellaneous signals.

[0133] Specifically, the miscellaneous signal transmission interface may be a Slimline interface.

[0134] In one embodiment of the present application, the miscellaneous signal includes a clock signal, and the memory system 701 further includes:

[0135] The chip of the first adapter board is also used to obtain a clock signal.

[0136] In one embodiment of the present application, the miscellaneous signal includes a chip enable signal, and the memory system 701 further includes:

[0137] A field programmable logic gate array module integrated into the first mainboard;

[0138] The field programmable logic gate array module of the first mainboard is used to send a chip enable signal;

[0139] The chip of the first adapter board is also used to start working after obtaining the chip enable signal, and convert the processor's internal dedicated bus signal into a double-rate memory signal.

[0140] In one embodiment of the present application, the miscellaneous signal includes a power-on completion signal, and the memory system 701 further includes:

[0141] The memory module of the first adapter board is used to send a power-on completion signal after power-on is completed, and transmit the power-on completion signal to the field programmable logic gate array module of the first main board; and is also used to perform in-situ detection after obtaining the double-rate memory signal to generate a corresponding detection signal.

[0142] In one embodiment of the present application, the miscellaneous signal includes a detection signal, and the memory system 701 further includes:

[0143] A baseboard management controller module integrated into the first mainboard;

[0144] The baseboard management controller module of the first mainboard is used to obtain a detection signal and determine a system operating state based on the detection signal, and initialize the chip of the first adapter board when it is determined that the system operating state is normal.

[0145] In one embodiment of the present application, the miscellaneous signal includes a reset signal, and the memory system 701 further includes:

[0146] The field programmable logic gate array module of the first main board is further used to send a reset signal when initializing the chip of the first adapter board;

[0147] The memory module of the first adapter board is further used to provide corresponding memory resources to the processor of the first mainboard for access and use after obtaining the reset signal.

[0148] As shown in Figure 9, this is a system architecture diagram of another memory system of an embodiment of the present application. When the memory resources are initially insufficient, the solution shown in Figure 9 can be adopted to optimize the system. Disconnect Cable5 (the difference between Cable4 and Cable5 is the length) that transmits the UPI signal, use Cable2 to connect to the SWITCH BOARD (adapter board), and connect the miscellaneous signals on the motherboard to the SWITCH BOARD through Cable1. The signal on the UPI CON is transmitted to the MXC chip (CXL DRAM memory controller, which belongs to the third device type defined by the CXL protocol) through the internal wiring of the board, and the miscellaneous signals on the Slimline interface are transmitted to the MXC chip and the corresponding memory module.

[0149] Among them, miscellaneous signals include:

[0150] SMB: Performs normal data transmission, allowing the device to be monitored by the BMC.

[0151] PRSNT0: Performs in-place detection to check whether the adapter board is in place. If it is not in place, the high and low levels of this signal will be sensed by the field programmable logic gate array module on the mainboard.

[0152] I3C_SPD: Implements serial presence monitoring of DIMM strips and performs communication.

[0153] 100M_CLKDP\DN: Provides clock signal to MXC chip.

[0154] PWR_BREAK: Connected to the MXC, it monitors whether the MXC voltage is abnormal. When the voltage is abnormal, it directly triggers the protection of the MXC and DDR DIMM.

[0155] RST: Reset MXC.

[0156] FPGA_RESET: RESET the DIMM.

[0157] MXC_EN: Enable signal for the MXC chip to work.

[0158] PWRGD: A signal that notifies the FPGA that the DDR DIMM has completed powering on.

[0159] As shown in Figure 10, it is a signal transmission diagram of an embodiment of the present application. When this system is working normally, CLK_100M (clock signal) is first sent to the MXC chip, and the IPB signal based on the PCIE physical link transmission converted from the UPI signal is sent to MXC. At this time, FPGA needs to send MXC_EN (chip enable signal) to start the MXC chip to start working. After waiting for the communication between SMB and the motherboard to be completed, MXC is initialized. At the same time, the DIMM SLOT (memory module) is powered on. After power-on is complete, it sends a PWRGD (power-on completion signal) to the FPGA. The IPB signal transmitted based on the PCIE physical link will be converted by the MXC chip, and the DDR signal that complies with the DDR5 standard protocol will be output from the dedicated interface on the MXC chip. The DIMM SLOT obtains these signals and performs in-place detection through I3C_SPD, notifying the BMC of the detection signal. After the BMC determines that the system is working normally based on the detection signal, it will initialize the resources converted by MXC. The FPGA sends an FPGA_RESET (reset signal) to the DIMM SLOT. At this time, the system officially starts working. The memory resources on the DIMM will be directly accessible to the CPU. These memory resources are the same as the memory resources on the DIMM directly connected to the CPU, which increases the DRAM capacity of the system, optimizes the core ratio of the system, and improves the ability to handle storage and computing resources.

[0160] In one embodiment of the present application, the memory system 701 further includes:

[0161] A second adapter board, integrating a high-speed interconnect channel interface, miscellaneous signal transmission interface, chip and memory module on the second adapter board;

[0162] The connecting cable is further used to establish a physical link between the high-speed interconnect channel interface of the second mainboard and the high-speed interconnect channel interface of the second adapter board, and to establish a physical link between the miscellaneous signal transmission interface of the second mainboard and the miscellaneous signal transmission interface of the second adapter board, so as to establish a communication connection between the second mainboard and the second adapter board; wherein the connection between the high-speed interconnect channel interface of the second mainboard and the high-speed interconnect channel interface of the second adapter board is used to transmit high-speed channel interconnection signals;

[0163] The connecting cable is also used to convert high-speed channel interconnect signals into processor internal dedicated bus signals transmitted on a physical link based on a high-speed serial computer expansion bus standard;

[0164] The chip of the second adapter board is used to obtain the converted internal dedicated bus signal of the processor and convert the internal dedicated bus signal of the processor into a double-rate memory signal so that the memory resources of the memory module of the second adapter board can be accessed and used by the processor of the second mainboard.

[0165] As shown in Figure 11, this is a system architecture diagram of another memory system according to an embodiment of the present application. When the system memory resources are further insufficient, the solution shown in Figure 11 can be adopted to optimize the system. The UPI of CPU0 on MB0 can be brought out for conversion, and the UPI of CPU1 on MB1 can also be brought out for conversion. Among them, the relevant logic control signals need to be transmitted to the FPGA of MB0 through the MISC CON (Miscellaneous connector) and Cable3. This increases the DRAM capacity of the system and improves the ability to process storage and computing resources.

[0166] Furthermore, according to the design principle of a dual-socket server system, only one UPI connection is required for the server system to operate normally. Therefore, all UPI CON1-5 can be disconnected and connected to the switch board for conversion. This increases the system's DRAM capacity and significantly improves its ability to handle storage and computing resources.

[0167] In one embodiment of the present application, the first main board is communicatively connected to the plurality of first adapter boards; and the second main board is communicatively connected to the plurality of second adapter boards.

[0168] In one embodiment of the present application, the memory module of the first adapter board is the first memory module or the second memory module; the memory module of the second adapter board is the first memory module or the second memory module.

[0169] Specifically, the first memory module may be a 1DPC memory module, and the second memory module may be a 2DPC memory module.

[0170] In one embodiment of the present application, if memory resources are still insufficient, the memory resources on the adapter board can be designed as 2DPC. Figure 12 shows a schematic diagram of an adapter board designed as 2DPC in accordance with an embodiment of the present application. Given the current design, the bandwidth of a CXL x8 channel is approximately the same as that of a DDR5 memory channel. This allows the adapter board to be optimized so that the two DIMM slots can be designed identically, using the same set of links.

[0171] When this system is working normally, CLK_100M is first sent to the MXC chip, and the IPB signal based on the PCIE physical link transmission converted from the UPI signal is sent to the MXC. At this time, the FPGA needs to send MXC_EN to start the MXC chip working. After waiting for the communication between the SMB and the motherboard to be completed, the MXC is initialized. At the same time, both DIMM slots are powered on simultaneously. After power-on, both send PWRGD signals to the FPGA. (If only one DIMM slot sends PWRGD, the FPGA will not be able to obtain the high level of PWRGD due to the wired-AND relationship.) The IPB signals transmitted based on the PCIE physical link will be converted by the MXC chip, and DDR signals compliant with the DDR5 standard protocol will be output from the dedicated interface on the MXC chip. The DIMM slots receive these signals and perform in-place detection through I3C_SPD, notifying the BMC of the detection signal. After the BMC determines that the system is operating normally based on the detection signal, it will initialize the resources converted by MXC. The FPGA sends FPGA_RESET to both DIMM slots, and the system officially begins operation. The memory resources on the two DIMMs will be directly accessible to the CPU. These memory resources are the same as the memory resources on the DIMMs directly connected to the CPU. This increases the system's DRAM capacity, optimizes the core ratio of the system, and improves the ability to handle storage and computing resources.

[0172] In summary, in the embodiment of the present application, the dual-core server is separated by Cable, so that the IPB resources are no longer dedicated. The dedicated IPB resources can be converted into IPB resources based on PCIE physical link transmission through another Cable, and combined with the CXL cache consistency protocol, this resource is converted into DIMM resources, thereby expanding the memory pool capacity of the entire system. In addition, not only the problem of low IPB utilization efficiency is solved, but also the memory resources are enriched by using external boards, solving the design difficulties of onboard 2DPC or 3DPC layout and routing. The present application can increase the DRAM capacity of the memory system, optimize the ratio with the core of the system, improve the ability to process storage and computing resources, increase the memory density on a single system, and reduce the cost of memory resources per unit system. The present application combines the advantages of 1DPC and 2DPC for design, using the adapter board design of 1DPC when a high transmission rate is required, and using the adapter board design of 2DPC when a larger DRAM memory space is required.

[0173] This application can effectively solve the problem of a server system with a too small memory pool and a low DRAM to CORE ratio; the problem of low utilization of IPB dedicated resources when the system interaction frequency is low; the difficulty in adding DIMM space on the server system motherboard; the problem of reduced DIMM transmission rate when the memory pool has a large capacity, and the problem of low memory capacity in a system with a high memory transmission rate.

[0174] 13 , there is shown a block diagram of a memory system according to another embodiment of the present application. The memory system 1301 includes:

[0175] A first motherboard 13011, a processor integrated in the first motherboard 13011, and a high-speed interconnect channel interface;

[0176] A first adapter board 13012 , integrating a high-speed interconnect channel interface, a chip, and a memory module on the first adapter board 13012 ;

[0177] At least one connecting cable 13013;

[0178] A connecting cable 13013 is used to establish a physical link between the high-speed interconnect channel interface of the first mainboard 13011 and the high-speed interconnect channel interface of the first adapter board 13012, thereby establishing a communication connection between the first mainboard 13011 and the first adapter board 13012; wherein the connection between the high-speed interconnect channel interface of the first mainboard 13011 and the high-speed interconnect channel interface of the first adapter board 13012 is used to transmit high-speed channel interconnect signals;

[0179] The connection cable 13013 is also used to convert the high-speed channel interconnection signal into a processor internal dedicated bus signal transmitted on a physical link based on the high-speed serial computer expansion bus standard;

[0180] The chip of the first adapter board 13012 is used to obtain the converted internal processor dedicated bus signal and convert the internal processor dedicated bus signal into a double data rate memory signal, so that the memory resources of the memory module of the first adapter board 13012 can be accessed and used by the processor of the first main board 13011;

[0181] The first mainboard 13011 integrates multiple processors 130111, and each processor 130111 is interconnected to transmit internal dedicated bus signals of the processor. The first mainboard 13011 is in communication with multiple first adapter boards 13012, and further includes:

[0182] Each processor 130111 of the first mainboard 13011 is used to call the memory resources of the memory modules of each first adapter board 13012.

[0183] Specifically, the first motherboard can be integrated with four or eight processors. Referring to FIG14 , which is a system architecture diagram of a memory system in which the first motherboard is integrated with four processors according to an embodiment of the present application, any two processors are interconnected to transmit internal processor-specific bus signals. The memory system includes four first adapter boards. When the cables are correctly connected, the memory resources of any memory module on any first adapter board can be accessed by any processor on the first motherboard. For example, CPU0 can access the memory resources of any first adapter board.

[0184] 15 , which is a system architecture diagram of a memory system in which eight processors are integrated into the first mainboard of an embodiment of the present application, the processors on the first mainboard are interconnected to transmit internal dedicated bus signals of the processors, and the memory resources of the first adapter board can be called arbitrarily.

[0185] In the embodiments of the present application, when the server is a multi-processor server, the technical solutions of the embodiments of the present application will be implemented according to different server hardware architectures, which are mainly determined by the IPB interconnection topology of multiple CPUs. The CPU has multiple groups of IPB resources, for example, one CPU has four groups of IPB buses, and another CPU has six groups of IPB buses. Different hardware connection schemes can be designed based on different topologies to expand the memory resources of the memory system.

[0186] When the server is a multi-way server, the technical solutions of the above embodiments can be combined to maximize the capacity of the memory pool.

[0187] In summary, in an embodiment of the present application, the memory system includes a first motherboard, a processor and a high-speed interconnect channel interface integrated on the first motherboard, and also includes a first adapter board, a high-speed interconnect channel interface, a chip and a memory module integrated on the first adapter board. A communication connection between the first motherboard and the first adapter board can be established through a connecting cable. The connecting cable can convert the high-speed channel interconnect signal into a processor internal dedicated bus signal transmitted based on a physical link of the high-speed serial computer expansion bus standard. Then, the chip of the first adapter board can convert the processor internal dedicated bus signal into a double-rate memory signal, so that the memory resources of the memory module of the first adapter board can be accessed and used by the processor of the first motherboard. By adopting the above design, the high-speed channel interconnect signal is converted into an IPB resource that can go through a PCIE physical link through a connecting cable, and the IPB resource is converted into a DIMM memory resource that complies with the JEDEC protocol standard through a chip that supports the CXL protocol. Using an external board to expand DRAM resources so that the DIMM resources are not on the same motherboard as the CPU can optimize the spatial layout resources of the motherboard.

[0188] 16 , a flowchart of a memory resource adjustment method according to an embodiment of the present application is shown. The method is applied to a memory system including the first mainboard, the first adapter board, the second mainboard, and the second adapter board as described above. Specifically, the method may include the following steps:

[0189] Step 1601: determine the current load information of the memory system, and determine the target memory resources to be provided based on the load information.

[0190] Step 1602, based on the target memory resources, determines whether to control the first main board to be connected to the first adapter board, and / or, determines whether to control the second main board to be connected to the second adapter board, and / or, determines the number of first adapter boards connected to the first main board, and / or, determines the number of second adapter boards connected to the second main board.

[0191] In an embodiment of the present application, the current load information of the memory system can be determined, and the target memory resources to be provided by the memory system can be determined based on the load information. The load information can refer to the size of the memory space occupied by the programs running in the system; the target memory resources are used to maintain the processing, storage, and computing resource capabilities of the memory system at a high level.

[0192] In one embodiment of the present application, determining whether to control the connection between the first mainboard and the first adapter board based on the target memory resource, and / or determining whether to control the connection between the second mainboard and the second adapter board, and / or determining the number of first adapter boards connected to the first mainboard, and / or determining the number of second adapter boards connected to the second mainboard, may specifically include the following sub-steps:

[0193] If the target memory resources do not exceed the memory resources provided by the first mainboard and the second mainboard, the first mainboard is controlled to be disconnected from the first adapter board, and the second mainboard is controlled to be disconnected from the second adapter board, so that only the first mainboard and the second mainboard are responsible for providing memory resources.

[0194] If the target memory resources exceed the memory resources provided by the first main board and the second main board, and do not exceed the preset first memory resource usage threshold, the first main board is controlled to be connected to the first adapter board, and the second main board is controlled to be disconnected from the second adapter board, so that the first main board, the second main board and at least one first adapter board are responsible for providing memory resources.

[0195] If the target memory resource exceeds the preset first memory resource usage threshold and does not exceed the preset second memory resource usage threshold, the first main board is controlled to be connected to the first adapter board, and the second main board is controlled to be connected to the second adapter board, so that the first main board, the second main board, at least one first adapter board and at least one second adapter board are responsible for providing memory resources.

[0196] In one embodiment of the present application, the following steps may be further included:

[0197] If the target memory resource exceeds a preset second memory resource usage threshold, the memory module of the first riser board and / or the second riser board is configured as a second memory module.

[0198] In an embodiment of the present application, as more and more target memory resources need to be provided, steps can be gradually performed to establish a connection between the first mainboard and the first adapter board, establish a connection between the second mainboard and the second adapter board, increase the number of first adapter boards connected to the first mainboard, increase the number of second adapter boards connected to the second mainboard, and configure the memory modules of the first adapter board and / or the second adapter board as 2DPC memory modules.

[0199] In summary, in the embodiment of the present application, the dual-core server is separated by Cable, so that the IPB resources are no longer dedicated. The dedicated IPB resources can be converted into IPB resources that go through the PCIE physical link through another Cable. Combined with the CXL cache consistency protocol, this resource is converted into DIMM resources, thereby expanding the memory pool capacity of the entire system. In addition, not only does it solve the problem of low IPB utilization efficiency, but it also uses external boards to enrich memory resources and solve the design difficulties of onboard 2DPC or 3DPC layout and routing. The present application can increase the DRAM capacity of the memory system, optimize the ratio with the system core, improve the ability to process storage and computing resources, increase the memory density on a single system, and reduce the cost of memory resources per unit system. The present application combines the advantages of 1DPC and 2DPC for design, using the 1DPC adapter board design when a high transmission rate is required, and using the 2DPC adapter board design when a larger DRAM memory space is required.

[0200] It should be noted that for the method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present application are not limited by the order of the actions described, because according to some embodiments of the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all embodiments of the present application, and the actions involved are not necessarily required by the embodiments of the present application.

[0201] 17 shows a block diagram of a memory resource adjustment device according to an embodiment of the present application, which is applied to a memory system. The memory system includes the first main board, the first adapter board, the second main board, and the second adapter board as described above, and may specifically include the following modules:

[0202] A first determining module 1701 is configured to determine current load information of the memory system and determine target memory resources to be provided based on the load information;

[0203] The second determination module 1702 is used to determine whether to control the first mainboard to be connected to the first adapter board based on the target memory resources, and / or determine whether to control the second mainboard to be connected to the second adapter board, and / or determine the number of first adapter boards connected to the first mainboard, and / or determine the number of second adapter boards connected to the second mainboard.

[0204] In some embodiments of the present application, the second determining module includes:

[0205] The first control submodule is used to control the first mainboard to be disconnected from the first adapter board, and to control the second mainboard to be disconnected from the second adapter board if the target memory resources do not exceed the memory resources provided by the first mainboard and the second mainboard, so that only the first mainboard and the second mainboard are responsible for providing memory resources.

[0206] In some embodiments of the present application, the second determining module includes:

[0207] The second control submodule is used to control the first mainboard to be connected to the first adapter board, and to control the second mainboard to be disconnected from the second adapter board if the target memory resources exceed the memory resources provided by the first mainboard and the second mainboard, and do not exceed the preset first memory resource usage threshold, so that the first mainboard, the second mainboard and at least one first adapter board are responsible for providing memory resources.

[0208] In some embodiments of the present application, the second determining module includes:

[0209] The third control submodule is used to control the first mainboard to be connected to the first adapter board, and to control the second mainboard to be connected to the second adapter board if the target memory resource exceeds the preset first memory resource usage threshold and does not exceed the preset second memory resource usage threshold, so that the first mainboard, the second mainboard, at least one first adapter board and at least one second adapter board are responsible for providing memory resources.

[0210] In some embodiments of the present application, the apparatus further comprises:

[0211] The configuration module is configured to configure the memory module of the first adapter board and / or the second adapter board as the second memory module if the target memory resource exceeds a preset second memory resource usage threshold.

[0212] In summary, in the embodiment of the present application, the dual-core server is separated by Cable, so that the IPB resources are no longer dedicated. The dedicated IPB resources can be converted into IPB resources that go through the PCIE physical link through another Cable. Combined with the CXL cache consistency protocol, this resource is converted into DIMM resources, thereby expanding the memory pool capacity of the entire system. In addition, not only does it solve the problem of low IPB utilization efficiency, but it also uses external boards to enrich memory resources and solve the design difficulties of onboard 2DPC or 3DPC layout and routing. The present application can increase the DRAM capacity of the memory system, optimize the ratio with the system core, improve the ability to process storage and computing resources, increase the memory density on a single system, and reduce the cost of memory resources per unit system. The present application combines the advantages of 1DPC and 2DPC for design, using the 1DPC adapter board design when a high transmission rate is required, and using the 2DPC adapter board design when a larger DRAM memory space is required.

[0213] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0214] An embodiment of the present application also provides an electronic device, as shown in Figure 18, including: a processor 1801, a memory 1802, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the various processes of the above-mentioned memory resource adjustment method embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0215] An embodiment of the present application also provides a computer non-volatile readable storage medium. As shown in FIG19 , a computer program 1901 is stored on the computer non-volatile readable storage medium. When the computer program is executed by the processor, the various processes of the above-mentioned memory resource adjustment method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0216] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0217] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, devices, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0218] The present application embodiment is described with reference to the flow chart and / or block diagram of the method according to the embodiment of the present application, terminal equipment (system), and computer program product.It should be understood that each process and / or box in the flow chart and / or block diagram and the combination of the process and / or box in the flow chart and / or block diagram can be realized by computer program instructions.These computer program instructions can be provided to the processor of general-purpose computer, special-purpose computer, embedded processing machine or other programmable data processing terminal equipment to produce a machine, so that the instruction executed by the processor of computer or other programmable data processing terminal equipment produces the device for realizing the function specified in one process or multiple processes and / or one box or multiple boxes of flow chart.

[0219] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0220] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce computer-implemented processing, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0221] Although some embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the embodiments of the present application and all changes and modifications that fall within the scope of the embodiments of the present application.

[0222] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0223] The above is a detailed introduction to a memory system, a memory resource adjustment method, a memory resource adjustment device, an electronic device and a computer non-volatile readable storage medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, based on the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A memory system, It is characterized in that include: A first motherboard, a processor integrated in the first motherboard, and a high-speed interconnect channel interface; A first adapter board, a high-speed interconnect channel interface, a chip and a memory module integrated in the first adapter board; at least one connecting cable; The connection cable is used to establish a physical link between the high-speed interconnection channel interface of the first mainboard and the high-speed interconnection channel interface of the first adapter board, so as to establish a communication connection between the first mainboard and the first adapter board; wherein the connection between the high-speed interconnection channel interface of the first mainboard and the high-speed interconnection channel interface of the first adapter board is used to transmit a high-speed channel interconnection signal; The connecting cable is also used to convert the high-speed channel interconnection signal into a processor internal dedicated bus signal for physical link transmission based on the high-speed serial computer expansion bus standard; The chip of the first adapter board is used to obtain the converted internal dedicated bus signal of the processor and convert the internal dedicated bus signal of the processor into a double-rate memory signal, so that the memory resources of the memory module of the first adapter board can be accessed and used by the processor of the first mainboard, thereby realizing memory pooling.

2. The memory system according to claim 1, It is characterized in that Also includes: A second mainboard, a processor and a high-speed interconnect channel interface integrated in the second mainboard; The connection cable is further used to establish a physical link between the high-speed interconnection channel interface of the first mainboard and the high-speed interconnection channel interface of the second mainboard, so as to establish a communication connection between the first mainboard and the second mainboard; A second adapter plate having the same structure as the first adapter plate; The second adapter board is used to connect to the second main board; the memory resource adjustment device in the memory system is used to determine the current load information of the memory system and determine the target memory resources to be provided according to the load information; Based on the target memory resources, determine whether to control the first mainboard to be connected to the first adapter board, and / or determine whether to control the second mainboard to be connected to the second adapter board, and / or determine the number of the first adapter boards connected to the first mainboard, and / or determine the number of the second adapter boards connected to the second mainboard.

3. The memory system according to claim 2, It is characterized in that The connecting cable comprises a first connecting cable and a second connecting cable, and further comprises: The first connection cable is used for data transmission between the processor of the first mainboard and the processor of the second mainboard; The second connecting cable is used to transmit logic control signals between the first mainboard and the second mainboard.

4. The memory system according to claim 3, It is characterized in that Also includes: A miscellaneous signal transmission interface integrated in the first mainboard; A miscellaneous signal transmission interface integrated in the first adapter board; The connecting cable is also used to establish a physical link between the miscellaneous signal transmission interface of the first mainboard and the miscellaneous signal transmission interface of the first adapter board; wherein the connection between the miscellaneous signal transmission interface of the first mainboard and the miscellaneous signal transmission interface of the first adapter board is used to transmit miscellaneous signals.

5. The memory system according to claim 4, It is characterized in that The miscellaneous signals include a clock signal and further include: The chip of the first adapter board is also used to obtain the clock signal.

6. The memory system according to claim 5, It is characterized in that The miscellaneous signals include a chip enable signal and further include: A field programmable gate array module integrated in the first mainboard; The field programmable logic gate array module of the first mainboard is used to send the chip enable signal; The chip of the first adapter board is also used to start working after obtaining the chip enable signal, and convert the internal dedicated bus signal of the processor into the double rate memory signal.

7. The memory system according to claim 6, It is characterized in that The miscellaneous signals include a power-on completion signal, and also include: The memory module of the first adapter board is used to send a power-on completion signal after power-on is completed, and transmit the power-on completion signal to the field programmable logic gate array module of the first main board; and is also used to perform in-situ detection after obtaining the double-rate memory signal to generate a corresponding detection signal.

8. The memory system according to claim 7, It is characterized in that The miscellaneous signal includes the detection signal and further includes: A baseboard management controller module integrated in the first mainboard; The baseboard management controller module of the first mainboard is used to obtain the detection signal and determine the system working state based on the detection signal, and initialize the chip of the first adapter board when it is determined that the system working state is normal.

9. The memory system according to claim 8, It is characterized in that The miscellaneous signals include a reset signal, and further include: a field programmable logic gate array module of the first mainboard, further configured to send the reset signal when initializing the chip of the first adapter board; The memory module of the first adapter board is further used to provide corresponding memory resources to the processor of the first mainboard for access and use after obtaining the reset signal.

10. The memory system according to claim 9, It is characterized in that Also includes: A second adapter board, a high-speed interconnect channel interface, a miscellaneous signal transmission interface, a chip and a memory module integrated in the second adapter board; The connection cable is further used to establish a physical link between the high-speed interconnection channel interface of the second mainboard and the high-speed interconnection channel interface of the second adapter board, and to establish a physical link between the miscellaneous signal transmission interface of the second mainboard and the miscellaneous signal transmission interface of the second adapter board, so as to establish a communication connection between the second mainboard and the second adapter board; The connection between the high-speed interconnection channel interface of the second main board and the high-speed interconnection channel interface of the second adapter board is used to transmit a high-speed channel interconnection signal; The connecting cable is also used to convert the high-speed channel interconnection signal into a processor internal dedicated bus signal for physical link transmission based on the high-speed serial computer expansion bus standard; The chip of the second adapter board is used to obtain the converted internal dedicated bus signal of the processor and convert the internal dedicated bus signal of the processor into a double-rate memory signal so that the memory resources of the memory module of the second adapter board can be accessed and used by the processor of the second mainboard.

11. The memory system according to claim 10, It is characterized in that The first main board is communicatively connected to the plurality of the first adapter boards; the second main board is communicatively connected to the plurality of the second adapter boards.

12. The memory system according to claim 11, It is characterized in that The memory module of the first adapter board is the first memory module or the second memory module; the memory module of the second adapter board is the first memory module or the second memory module.

13. The memory system according to claim 1, It is characterized in that The first mainboard is integrated with a plurality of processors. The processors are interconnected to transmit internal dedicated bus signals of the processors, the first mainboard is communicatively connected with the plurality of the first adapter boards, and further comprises: Each processor of the first mainboard is used to call the memory resources of the memory modules of the first adapter boards.

14. A memory resource adjustment method, It is characterized in that Applied to a memory system according to any one of claims 1 to 13, the memory system comprising a first mainboard, a first adapter board, a second mainboard, and a second adapter board, the method comprising: Determining current load information of the memory system, and determining target memory resources to be provided according to the load information; Based on the target memory resources, determine whether to control the first mainboard to be connected to the first adapter board, and / or determine whether to control the second mainboard to be connected to the second adapter board, and / or determine the number of the first adapter boards connected to the first mainboard, and / or determine the number of the second adapter boards connected to the second mainboard.

15. The method according to claim 14, It is characterized in that The determining, based on the target memory resource, whether to control the first mainboard to be connected to the first adapter board, and / or, determining whether to control the second mainboard to be connected to the second adapter board, and / or, determining the number of the first adapter boards connected to the first mainboard, and / or, determining the number of the second adapter boards connected to the second mainboard, includes: If the target memory resources do not exceed the memory resources provided by the first mainboard and the second mainboard, the first mainboard is controlled to be disconnected from the first adapter board, and the second mainboard is controlled to be disconnected from the second adapter board, Therefore, only the first mainboard and the second mainboard are responsible for providing memory resources.

16. The method according to claim 14, It is characterized in that The determining, based on the target memory resource, whether to control the first mainboard to be connected to the first adapter board, and / or, determining whether to control the second mainboard to be connected to the second adapter board, and / or, determining the number of the first adapter boards connected to the first mainboard, and / or, determining the number of the second adapter boards connected to the second mainboard, includes: If the target memory resources exceed the memory resources provided by the first mainboard and the second mainboard, and do not exceed the preset first memory resource usage threshold, the first mainboard is controlled to be connected to the first adapter board, and the second mainboard is controlled to be disconnected from the second adapter board, so that the first mainboard, the second mainboard and at least one of the first adapter boards are responsible for providing memory resources.

17. The method according to claim 14, It is characterized in that The determining, based on the target memory resource, whether to control the first mainboard to be connected to the first adapter board, and / or, determining whether to control the second mainboard to be connected to the second adapter board, and / or, determining the number of the first adapter boards connected to the first mainboard, and / or, determining the number of the second adapter boards connected to the second mainboard, includes: If the target memory resources exceed the preset first memory resource usage threshold and do not exceed the preset second memory resource usage threshold, the first mainboard is controlled to be connected to the first adapter board, and the second mainboard is controlled to be connected to the second adapter board, so that the first mainboard, the second mainboard, at least one of the first adapter boards and at least one of the second adapter boards are responsible for providing memory resources.

18. The method according to claim 14, It is characterized in that The method further comprises: If the target memory resource exceeds the preset second memory resource usage threshold, the memory module of the first adapter board and / or the second adapter board is configured as a second memory module.

19. A memory resource adjustment device, It is characterized in that Applied to a memory system according to any one of claims 1 to 13, the memory system comprising a first mainboard, a first adapter board, a second mainboard and a second adapter board, the device comprising: A first determination module, used to determine current load information of the memory system, and determine target memory resources to be provided according to the load information; a second determination module, configured to determine, based on the target memory resource, whether to control the first mainboard to be connected to the first adapter board, and / or to determine whether to control the second mainboard to be connected to the second adapter board, and / or to determine whether to control the second mainboard to be connected to the second adapter board The number of the first adapter boards connected to the first mainboard, and / or the number of the second adapter boards connected to the second mainboard are determined.

20. An electronic device, It is characterized in that include: A processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the steps of a memory resource adjustment method as described in any one of claims 14 to 18 are implemented.

21. A computer non-volatile readable storage medium, It is characterized in that The computer non-volatile readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a memory resource adjustment method as described in any one of claims 14-18 are implemented.

Citation Information

Patent Citations

  • Extension structure of memory module slot

    CN101196763A

  • A remote memory expansion management system

    CN109684257A

  • Resource sharing device, resource management device, and resource management method

    CN115586964A

  • Memory allocation system and server

    CN115904714A

  • Memory system, memory resource adjusting method and device, electronic equipment and medium

    CN116225177A

Cited By

  • Test system and method, electronic equipment, storage medium and product

    CN120234200A