Host interconnection access method, system, device, computer equipment and storage medium

By introducing switching nodes between the computing nodes and memory nodes, using CXL switching chips and memory expansion controllers to realize memory resource mapping between hosts, the problem of access difficulties between hosts is solved, and heterogeneous computing performance and CXL technology are improved flexibility and cost-effectiveness.

CN118519931BActive Publication Date: 2024-11-15INSPUR SUZHOU INTELLIGENT TECH CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411000729.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-15
Estimated Expiration
2044-07-24

AI Technical Summary

Technical Problem

In the prior art, there is a lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of CXL technology.

Method used

By introducing a switching node between the computing node and the memory node, decoding the access request using a preset decoder, and converting the CXL resources into memory resources through the CXL switching chip and the memory expansion controller, realizing memory resource mapping between hosts and interconnect access to avoid relying on processor bus and network cards.

Benefits of technology

It realizes fast access between hosts, reduces system memory costs and access costs, improves the performance of heterogeneous computing and the flexibility of CXL technology, and supports memory pooling and dynamic memory scheduling.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118519931B_ABST
    Figure CN118519931B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of servers, and discloses a host interconnection access method, system, device, computer equipment and storage medium, the method comprising: transmitting an access request to a memory node, decoding the access request using a preset decoder in the memory node, obtaining request information, and sending the request information to a second host; if the request information returned by the second host is obtained, sending the returned request information to the first host, indicating that the first host cannot access the second host; if a consent message is obtained from the second host, parsing the consent message to obtain consent information, sending the consent information to the first host, indicating that the first host uses the first memory in the first host to receive the memory resources mapped from the second memory in the second host, so as to access the second host. The present invention solves the problem that there is a lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of CXL.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of servers, and in particular to a host interconnection access method, system, device, computer equipment and storage medium. Background Art

[0002] Currently, the use of CXL (Compute Express Link, cache consistency protocol) technology in heterogeneous computing has expanded the memory capacity beyond the full memory capacity of the previous server with the same architecture. However, it is currently difficult to achieve interconnection access between hosts in memory access and management, which is not conducive to the data management of the entire system, and to a certain extent limits the performance of heterogeneous computing, affecting the flexibility and cost of using CXL. In addition, it is difficult for one host to quickly access the data in another host without causing any impact on the operating systems of both hosts, resulting in the inability to free up the host's memory bandwidth and CPU (Central Processing Unit) cycles, and unable to improve the performance of the application system on the host.

[0003] Therefore, the related technology lacks a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL. Summary of the invention

[0004] In view of this, the present invention provides a host interconnection access method, system, apparatus, computer device and storage medium to solve the problem that the lack of a fast access mechanism between hosts limits the performance of heterogeneous computing and affects the flexibility and cost of CXL.

[0005] In a first aspect, the present invention provides a host interconnection access system, the system comprising: a computing node, a switching node and a memory node; the computing node comprises a first host and a second host;

[0006] The first host is connected to the switching node and is used to send an access request to the switching node;

[0007] The switch node is connected to the memory node, and is used to transmit the access request to the memory node when the access request is obtained from the first host, and decode the access request using a preset decoder in the memory node to obtain the request information and the first target port number;

[0008] The switching node is connected to the second host and is used to send the request information to the second host through the output port corresponding to the first target port number;

[0009] The second host is used to return the request information to the switching node, or send a response message to the switching node;

[0010] The switching node is also used to send the returned request information to the first host, indicating that the first host cannot access the second host, or to send the response information in the response message to the first host, indicating that the first host uses the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts in the computing node.

[0011] The host interconnection access system provided in this embodiment decodes the access request of the first host, obtains access information, and sends the access information to the second host. If the second host returns the access information, it means that the second host cannot be accessed. If the second host returns a response message, the response message is parsed to obtain the response information, and the response information is sent to the first host, indicating that the first memory in the first host receives the memory resources mapped from the second memory in the second host to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the access cost of interconnection between hosts. It solves the problem of lack of a mechanism for fast access between hosts, limiting the performance of heterogeneous computing, and affecting the flexibility and cost of using CXL.

[0012] In some optional implementations, the switching node sends the acceptance information in the acceptance message to the first host, including:

[0013] Removing the message source identifier and the message destination identifier in the response message to obtain a processed message;

[0014] Obtaining a second target port number from the processed message, and using preset information in the processed message as response information;

[0015] The response information is sent to the first host through the output port corresponding to the second target port number.

[0016] In this embodiment, the response message is parsed to obtain the response information and the second target port number, so as to determine the output port for transmitting the response information by using the second target port number, and notify the first host that the second host can be accessed by accessing the memory.

[0017] In some optional implementations, the second host returns the request information to the switching node, including:

[0018] Acquire a first transmission path of the request information;

[0019] Returning the request information to the switching node via the first transmission path;

[0020] The second host sends a response message to the switching node, including:

[0021] determining a second transmission path different from the first transmission path;

[0022] The response message is sent to the switching node through the second transmission path.

[0023] In some optional implementations, the switching node is further configured to:

[0024] Get CXL resources from multiple hosts from a compute node;

[0025] The CXL resources are transmitted to the memory node, and the memory node transforms the CXL resources of multiple hosts to obtain memory resources, and maps the memory resources to the common extended memory of multiple hosts in the memory node.

[0026] In this embodiment, the switch node transmits the acquired CXL resources to the memory node, and the memory node converts the CXL resources of multiple hosts into memory resources, maps the memory resources to the common extended memory of multiple hosts in the memory node, and provides a memory resource basis for host interconnection access.

[0027] In some optional implementations, before sending the access request to the switching node, the first host is further configured to:

[0028] In the case where the first host does not have an address decoder, mapping the global integrated storage information in the physical address space to the preset address space, wherein the global integrated storage information is used to determine the first target port number, and the first target port number is used to determine the port for sending the request information to the second host;

[0029] Generate an access request according to the preset address space.

[0030] In this embodiment, when the first host does not have an address decoder, the physical address space of the first host is mapped to a preset address space, and an access request is generated based on the preset address space, so that the access request is subsequently decoded using the preset decoder to obtain the first target port number.

[0031] In some optional implementations, the second host is further configured to:

[0032] Determine the directly connected memory and extended memory of the second host;

[0033] A mappable second memory is determined in the direct-connected memory and the extended memory, wherein the second memory is used to map memory resources of the second memory to the first memory in the first host when the second host is accessed by the first host.

[0034] In this embodiment, the second host determines a mappable second memory in the direct memory and the extended memory, so that when accessed by the first host, the memory resources of the second memory are mapped to the first memory in the first host, thereby realizing host interconnection access.

[0035] In some optional implementations, the switching node includes: a CXL switching card, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables;

[0036] The computing node is connected to the CXL switch card via a first connector and a first cable, and is used to send CXL resources of multiple hosts in the computing node to the CXL switch card;

[0037] The CXL switch card is connected to the memory node via a second connector and a second cable, and is used to send the CXL resource to the memory node.

[0038] In some optional implementations, the memory node includes: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors;

[0039] The memory expansion controller is connected to the CXL switch card via the third connector and the second cable, and is used to obtain CXL resources from the CXL switch card;

[0040] The memory expansion controller is connected to the memory device and is used to convert the CXL resources into memory resources in the memory device.

[0041] In this embodiment, the CXL resources are converted into memory resources in the memory device through the memory expansion controller to realize memory expansion for the host in the computing node, and the converted memory resources can be accessed by the host, providing a basis for subsequent hosts to realize interconnection access through memory.

[0042] In some optional embodiments, the CXL switch card and the memory expansion controller are configured to be powered on before the host in the computing node issues a system reset instruction;

[0043] The host is configured to be powered on after the CXL switch card and memory expansion controller are powered on.

[0044] In this embodiment, the power-on sequence of the CXL switch board, the memory expansion controller and the host is set so that the host interconnection access system works normally, and the host memory resources can be expanded while the host interconnection access is realized.

[0045] In a second aspect, the present invention provides a host interconnection access method, the method is applied to a switching node, the method comprising:

[0046] When an access request is obtained from the first host, the access request is transmitted to the memory node, and the access request is decoded by using a preset decoder in the memory node to obtain request information and a first target port number;

[0047] Sending the request information to the second host through the output port corresponding to the first target port number;

[0048] When the request information returned by the second host is obtained, the returned request information is sent to the first host, indicating that the first host cannot access the second host;

[0049] When a response message is obtained from the second host, the response message is parsed to obtain response information and a second target port number;

[0050] The response information is sent to the first host through the output port corresponding to the second target port number, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0051] The host interconnection access method provided in this embodiment decodes the access request of the first host to obtain access information, and sends the access information to the second host. If the second host returns the access information, it means that the second host cannot be accessed. If the second host returns a response message, the response message is parsed to obtain the response information, and the response information is sent to the first host, indicating that the first memory in the first host receives the memory resources mapped from the second memory in the second host to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the access cost of interconnection between hosts. The problem of lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL is solved.

[0052] In a third aspect, the present invention provides a host interconnection access method, the method is applied to a first host, and the method includes:

[0053] Sending an access request to the switching node through an input port of the switching node, and the switching node sending request information in the access request to the second host through an output port corresponding to a first target port number, wherein the first target port number is obtained after the switching node decodes the access request using a preset decoder in the memory node;

[0054] In the case of obtaining request information returned by the second host from the switching node, determining that the second host cannot be accessed;

[0055] When the consent information is obtained from the switching node, the memory resources mapped from the second memory in the second host are received by the first memory in the first host to access the second host, wherein the memory resources are obtained by converting the CXL resources of multiple hosts by the memory node.

[0056] The host interconnection access method provided in this embodiment is that the first host sends an access request containing request information to the switching node, and the switching node sends the request information to the second host. If the second host returns the access information, it means that the second host cannot be accessed. If the first host receives the consent information, the memory resources mapped from the second memory in the second host are received through the first memory to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the access cost of interconnection between hosts. The problem of lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL is solved.

[0057] In a fourth aspect, the present invention provides a host interconnection access method, the method is applied to a second host, and the method comprises:

[0058] When request information is obtained from the switching node, verify whether the request information is legal;

[0059] If the request information is illegal, the request information is returned to the switching node, and the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host;

[0060] If the request information is legal, the request information is converted into a response message, and the response message is sent to the switching node. The switching node sends the response information in the response message to the first host, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0061] The host interconnection access method provided in this embodiment is that after the second host verifies the access information, if it is illegal, the access information is returned, indicating that the second host cannot be accessed. If it is legal, a response message is returned to instruct the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the cost of interconnection access between hosts. The problem of lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL is solved.

[0062] In a fifth aspect, the present invention provides a host interconnection access device, which is deployed in a switching node and includes:

[0063] A decoding module, configured to, when an access request is obtained from the first host, transmit the access request to the memory node, and decode the access request using a preset decoder in the memory node to obtain request information and a first target port number;

[0064] A first sending module, used for sending the request information to the second host through the output port corresponding to the first target port number;

[0065] A second sending module is used to send the returned request information to the first host when the request information returned by the second host is obtained, indicating that the first host cannot access the second host;

[0066] A parsing module, configured to parse the consent message when a consent message is obtained from the second host, and obtain the consent information and the second target port number;

[0067] The third sending module is used to send the consent information to the first host through the output port corresponding to the second target port number, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0068] In a sixth aspect, the present invention provides a host interconnection access device, which is deployed on a first host and includes:

[0069] a fourth sending module, configured to send an access request to the switching node through an ingress port of the switching node, and the switching node sends the request information in the access request to the second host through an egress port corresponding to a first target port number, wherein the first target port number is obtained after the switching node decodes the access request using a preset decoder in the memory node;

[0070] A determination module, configured to determine that the second host cannot be accessed when request information returned by the second host is obtained from the switching node;

[0071] The access module is used to receive memory resources mapped from the second memory in the second host by using the first memory in the first host to access the second host when obtaining the consent information from the switching node, wherein the memory resources are obtained by the memory node converting the CXL resources of multiple hosts.

[0072] In a seventh aspect, the present invention provides a host interconnection access device, which is deployed on a second host and includes:

[0073] A verification module, used to verify whether the request information is legal when the request information is obtained from the switching node;

[0074] An information return module, used for returning the request information to the switching node when the request information is illegal, and the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host;

[0075] The fifth sending module is used to convert the request information into a response message when the request information is legal, and send the response message to the switching node, which sends the response information in the response message to the first host, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0076] In an eighth aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the host interconnection access method of the second aspect or any corresponding embodiment thereof by executing the computer instructions, or executes the host interconnection access method of the third aspect or any corresponding embodiment thereof, or executes the host interconnection access method of the fourth aspect or any corresponding embodiment thereof.

[0077] In a ninth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to cause a computer to execute the host interconnection access method of the second aspect or any corresponding embodiment thereof, or to execute the host interconnection access method of the third aspect or any corresponding embodiment thereof, or to execute the host interconnection access method of the fourth aspect or any corresponding embodiment thereof.

[0078] In a tenth aspect, the present invention provides a computer program product, comprising computer instructions, the computer instructions being used to enable a computer to execute the host interconnection access method of the second aspect or any corresponding embodiment thereof, or to execute the host interconnection access method of the third aspect or any corresponding embodiment thereof, or to execute the host interconnection access method of the fourth aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the related technologies, the drawings required for use in the specific embodiments or the related technical descriptions will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0080] Figure 1 is a schematic diagram of realizing host interconnection access by using a UPI link according to an embodiment of the present invention;

[0081] Figure 2 is a schematic diagram of realizing host interconnection access by using a network card according to an embodiment of the present invention;

[0082] Figure 3 is a schematic diagram of a CXL application example according to an embodiment of the present invention;

[0083] Figure 4 is a node topology diagram of a server in a cabinet using CXL to expand memory according to an embodiment of the present invention;

[0084] Figure 5 is a topology diagram of multiple hosts connected to a CXL extended memory in a cabinet according to an embodiment of the present invention;

[0085] Figure 6 A topology diagram for realizing host interconnection access based on CXL Fabric according to an embodiment of the present invention;

[0086] Figure 7 is a schematic diagram of mapping a physical address space to a preset address space according to an embodiment of the present invention;

[0087] Figure 8 is a schematic flow chart of a host interconnection access method applied to a switching node according to an embodiment of the present invention;

[0088] Fig. 9 is a schematic flow chart of a host interconnection access method applied to a first host according to an embodiment of the present invention;

[0089] Fig.10is a flow chart of a host interconnection access method applied to a second host according to an embodiment of the present invention;

[0090] Fig.11 is a structural block diagram of a host interconnection access device deployed in a switching node according to an embodiment of the present invention;

[0091] Fig.12 is a structural block diagram of a host interconnection access device deployed on a first host according to an embodiment of the present invention;

[0092] Fig.13 is a structural block diagram of a host interconnection access device deployed on a second host according to an embodiment of the present invention;

[0093] Fig.14 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0094] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0095] Currently, there are two ways to achieve mutual access between hosts. The first way is as follows: Figure 1As shown in the figure, the HOSTs between multiple computing nodes, the CPU in each HOST is connected to multiple DIMMs (Dual In-line Memory Modules) through CXL switching chips and memory expansion controllers to expand the memory for the CPU, and each HOST reserves a corresponding UPI (Ultra Path Interconnect) CON (Control) to connect to the CPU. Multiple HOSTs are generally connected by cables (CABLE), and multiple HOSTs access each other through UPI links. However, using UPI links to access between HOSTs requires adding a high-speed connector on the hardware design side and using thicker UPI cables for interconnection, which has great limitations in structure and hardware layout, and is costly. Using UPI link to access between hosts strictly depends on the firmware design of the CPU platform. The BMC (Board Management Controller) must first access the CPU, and then the CPU accesses the CPU of the other host through the cable. Only then can the hosts communicate with each other. The links and protocols of this type of access are too complex, resulting in limited access rate and communication content. Figure 2 As shown, the CPU in each HOST is connected to multiple DIMMs (Dual In-line Memory Modules) through a CXL switching chip and a memory expansion controller to expand the memory for the CPU, and the CPU is connected to a PCIE (Peripheral Component Interconnect Express, a high-speed serial computer expansion bus standard) SLOT (slot), and then a network card, such as a NIC (Network Interface Card) network card, is inserted into the PCIE SLOT to interconnect and access between the HOSTs. However, using a network card to access between the HOSTs requires the use of an X8 or X16 PCIE external interface, and the mutual access is limited by factors such as the wiring length, network card performance, network cable type, and signal quality. Using a network card to access between the HOSTs has a large limitation in cost and structure. The server itself needs a network port for external management, so the need to design an additional network card increases a certain cost, and the network card needs to occupy a three-dimensional space in the chassis. In addition, the memory expansion of the above two methods is limited to the chassis, and it is impossible to truly pool memory in a cabinet or data center.

[0096] The current memory expansion is based on CXL technology. The memory channel of CXL is the same as the PCIE channel, but the protocol is different. Adding remote direct media access technology, using UIO (Unordered Input / Output) information and adding GIM (global integrated memory) functions, combined with the processor's own directly connected memory, can initially achieve access between multiple hosts through CXL memory channels while memory expansion. Fabric is a highly interconnected network architecture that can connect a large number of nodes, such as servers, storage devices, accelerators, etc., and provide high-bandwidth, low-latency communication. The Fabric architecture is used to meet the communication needs of large-scale data centers and high-performance computing environments, and usually includes switches, routers and other network devices to form a flexible and scalable communication infrastructure. Figure 3 A CXL application example is shown. Figure 3 As shown in the figure, different hosts are connected together through the CXL switch chip (CXL SWITCH) that supports Fabric. The CPU in each host is connected to multiple DIMMs (Dual In-line Memory Modules) that support the GIM function through a memory expansion controller to achieve memory expansion for the CPU. The CPU is connected to a local memory (NATIVE DIMM) that supports the GIM function. Each host can allow other hosts to access its own expanded DIMMs and direct-connected DIMMs through the GIM function. The host can freely choose whether to map out all or part of its memory.

[0097] Based on the above content, this application forms all hosts into computing nodes, uses CXL switching chips (CXL SWITCH chips) that support Fabric to form switching nodes, and then uses memory expansion controllers (MXC chips) to manage DIMMs that support GIM functions to form memory nodes. The computing nodes and memory nodes are placed separately in the cabinet, and the computing nodes are connected to the memory nodes in the form of BOARD-to-BOARD through MCIO (Multi-Channel I / O, multi-channel input and output interface). The CXL resources will be connected to the switching nodes, and the CXL SWITCH chips will be used for expansion. Then, the MXC chips in the memory nodes will be used to convert the CXL resources into standard memory resources or directly connect the CXL resources to the MXC chips to convert them into memory resources. When the devices in the computing nodes and memory nodes have GIM functions, the CXL resources of multiple HOSTs can be interacted in the CXL SWITCH, and the memory resources converted after the interaction can map the information in the memory resources in the system, so when the system calls resources, the information between the HOSTs is shared and can be accessed by other HOSTs. The switch node uses the Fabric architecture for internal interconnection and management, so that the memory resources of a single CPU DOMAIN can be accessed by other CPUs. The memory resources of the DOMAIN can map data into non-DOMAIN memory. According to the feedback process mechanism, non-DOMAIN memory can also access data to the DOMAIN memory, realizing the mutual access mechanism between CPUs in different HOSTs through memory, and realizing the interconnection between HOSTs under a certain access process according to the protocol. It can solve the high-cost and limited access problems of the high-speed channel IPB (inner processor bus) between HOSTs that depends on the CPU itself, and solve the problems of direct access between HOSTs and HOSTs on multiple computing nodes that depends on the network card, occupies PCIE SLOT and chassis space, and has low flexibility and high cost in CXL use. In order to achieve interconnected access for multiple HOSTs, it not only improves the access efficiency and flexibility of the entire server and even the data center, but also realizes memory pooling in an overall sense, and brings out the effect of the dynamic call characteristics of memory.

[0098] In this embodiment, a host interconnection access system is provided, the system comprising: a computing node, a switching node and a memory node; the computing node comprises a first host and a second host;

[0099] The first host is connected to the switching node and is used to send an access request to the switching node;

[0100] The switch node is connected to the memory node, and is used to transmit the access request to the memory node when the access request is obtained from the first host, and decode the access request using a preset decoder in the memory node to obtain the request information and the first target port number;

[0101] The switching node is connected to the second host and is used to send the request information to the second host through the output port corresponding to the first target port number;

[0102] The second host is used to return the request information to the switching node, or send a response message to the switching node;

[0103] The switching node is also used to send the returned request information to the first host, indicating that the first host cannot access the second host, or to send the response information in the response message to the first host, indicating that the first host uses the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts in the computing node.

[0104] Specifically, Figure 4 This is a node topology diagram of the server using CXL to expand memory in the cabinet, such as Figure 4 As shown in the figure, the computing node can be a dual-way node or a single-way node, that is, it can contain 1 central processing unit or 2 central processing units. The computing node is connected to the switching node through multiple MCIOs. The computing node connects to the CXL resources through MCIOs and transmits the CXL resources to the switching node through cables (CABLE). The CXL resources of multiple HOSTs are aggregated in the switching node. The CXL switching board of the switching node uses the CXL switching chip (CXL SWITCH) as the core to exchange resources. Data is exchanged and distributed in the switching node, and the physical link of the CXL Fabric is generated in the switching node. The switching node is connected to the memory node through multiple MCIOs. The CXL switching chip transmits the CXL resources to the memory node through MCIO and CABLE. The core of the memory node is the memory extension controller (MXC chip), which converts the CXL resources into memory resources in the memory device (DEVICE) that supports the GIM function. The memory resources converted from the CXL resources are remotely mapped through the GIM function and mapped to the common extended memory of the HOST, which provides a basis for multiple HOSTs to access each other.

[0105] Host interconnect access systems such as: Figure 5The system shown includes a computing node, a switching node and a memory node. The computing node includes a first host and a second host. The first host has two central processing units, and the second host has two central processing units. Each central processing unit is connected to multiple DIMMs as local memory. The computing node is connected to the switching node via multiple MCIOs.

[0106] like Figure 6 As shown, in this embodiment, a preset decoder is set in the dynamic random access memory (DRAM) of the memory node to decode the access request (DICODE). When the first host needs to access the second host, an access request is generated. The access request is, for example, a UIO request. The UIO allows unordered data transmission on the CXL link, allowing data packets to be sent and received out of order as long as they can eventually be reassembled correctly. The first host sends the access request to the switching node through the MCIO of the computing node and the MCIO of the switching node, and then the access request is transmitted to the CXL switching chip of the switching node through the physical interface and the ingress port (IngressPort) of the switching node. Correspondingly, at the firmware level, the UIO information is sent to the Fabric address range, pointing to the GIM space. When the switching node obtains the access request from the first host, that is, when the UIO request information of the first host arrives at the Ingress Port of the CXL SWITCH chip, as shown Figure 6 As shown, the switch node transmits the access request to the memory node memory expansion controller, and uses the preset decoder of the DRAM in the memory node to decode (DECODE) the access request. The decoded content includes the request information and the first target port number. The first target port number is: the Destination Port ID (Destination Port Identifier, DPID) of which specific path in the Fabric is passed.

[0107] The switching node obtains the specified path through the egress port corresponding to the first target port number, and sends the request information to the second host through the specified path. Figure 6 As shown, the designated path is the physical interface connected to the memory node, the output port corresponding to the first target port number, the physical interface connected to the switch node and the computing node, the MCIO of the switch node, and the MCIO of the computing node. The request information (UIO REQUEST) is sent to the second host through the designated path.

[0108] After receiving the request information, namely the UIO information, the second host will check the legality of the request information. If the request information is illegal, the second host will not allow the first host to access it and will return the illegal request information. If the request information is legal, the second host will convert the UIO information into a response message through the ingress port of the switching node. The response message is, for example, a PBR (Port Based Routing) message.

[0109] When the switch node obtains the request information returned by the second host, it can be confirmed that the request information is illegal and the first host cannot access the second host. The switch node sends the returned request information to the first host, indicating that the first host cannot access the second host. When the switch node obtains the acceptance message, the CXL SWITCH chip in the switch node parses the acceptance message to obtain the acceptance information and the second target port number. Figure 6 As shown, the switching node determines the corresponding output port according to the second target port number, determines the path for transmitting the consent information, determines the physical interface connecting the switching node to the memory node, the corresponding output port for the second target port number, the physical interface connecting the switching node to the computing node, the MCIO of the switching node, and the MCIO of the computing node, and sends the consent information to the first host through the path for transmitting the consent information, indicating that the first host can access the second host by accessing the memory.

[0110] It should be noted that, taking a dual-host system as an example, the system includes a first host and a second host. This HOST refers to an OS domain, and each domain has more than one CPU processor. Therefore, the dual-CPU system is the smallest system that supports CXL Fabric for host interconnection access. The switch node uses the Fabric architecture for internal interconnection and management. All the memory in this system supports the GIM function. It supports mapping its own memory resources to the memory of other HOST DOMAINs (domains) through GIM under the management of CXL Fabric. Therefore, the DOMAIN CXL memory of any CPU can receive the memory resources mapped from the DOMAIN CXL memory of other CPUs, realizing the access of multiple HOSTs through memory interconnection. Among them, HOST DOMAIN refers to the CXL memory expanded under a single CPU, which becomes the HOST DOMAIN CXL memory for this CPU.

[0111] Based on the above content, the step of the first host accessing the second host by accessing the memory includes: using the first memory in the first host to receive the memory resource mapped from the second memory in the second host, the first memory is for example: the DOMAIN CXL memory under any CPU in the first host, and the second memory is for example: the DOMAIN CXL memory of the CPU in the second host. Accessing the second host is achieved through the memory resource.

[0112] The host interconnection access system provided in this embodiment decodes the access request of the first host, obtains access information, and sends the access information to the second host. If the second host returns the access information, it means that the second host cannot be accessed. If the second host returns a response message, the response message is parsed to obtain the response information, and the response information is sent to the first host, indicating that the first memory in the first host receives the memory resources mapped from the second memory in the second host to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the access cost of interconnection between hosts. It solves the problem of lack of a mechanism for fast access between hosts, limiting the performance of heterogeneous computing, and affecting the flexibility and cost of using CXL.

[0113] In some optional implementations, the switching node sends the acceptance information in the acceptance message to the first host, including:

[0114] Removing the message source identifier and the message destination identifier in the response message to obtain a processed message;

[0115] Obtaining a second target port number from the processed message, and using preset information in the processed message as response information;

[0116] The response information is sent to the first host through the output port corresponding to the second target port number.

[0117] Specifically, the response message is, for example, a PBR message. After the second host receives the request information, if the request information is legal, the second host converts the request information into a PBR message through the ingress port and sends the PBR message to the switching node.

[0118] When the PBR message arrives at the SWITCH chip of the switching node, the SWITCH chip will verify the message source identifier (Source PBR ID) and the message destination identifier (Destination PBR ID) in the message and obtain the processed message.

[0119] The switching node obtains the second destination port number from the processed message, and uses preset information in the processed message as the response information. The preset information is, for example, information of a Transaction Layer Packet in the processed message.

[0120] The switching node sends the response information to the first host through an egress port corresponding to the second target port number.

[0121] In this embodiment, the response message is parsed to obtain the response information and the second target port number, so as to determine the output port for transmitting the response information by using the second target port number, and notify the first host that the second host can be accessed by accessing the memory.

[0122] In some optional implementations, the second host returns the request information to the switching node, including:

[0123] Acquire a first transmission path of the request information;

[0124] Returning the request information to the switching node via the first transmission path;

[0125] The second host sends a response message to the switching node, including:

[0126] determining a second transmission path different from the first transmission path;

[0127] The response message is sent to the switching node through the second transmission path.

[0128] Specifically, after receiving the request information, namely the UIO information, the second host will perform a validity check on the request information. If the UIO information is illegal, the second host will return the illegal request information along the original transmission path that the request information was sent. Therefore, if the request information is illegal, the second host obtains the first transmission path used when the request information is transmitted to the second host, and returns the request information to the switching node through the first transmission path;

[0129] If the request information is legitimate, the second host converts the UIO information into a response message through the ingress port of the switching node, such as a PBR (Port Based Routing) message. This process should use a second transmission path different from the first transmission path used when the request information is transmitted to the second host. Therefore, if the request information is legitimate, the second host determines a second transmission path different from the first transmission path, and sends the response message to the switching node through the second transmission path.

[0130] In some optional implementations, the switching node is further configured to:

[0131] Get CXL resources from multiple hosts from a compute node;

[0132] The CXL resources are transmitted to the memory node, and the memory node transforms the CXL resources of multiple hosts to obtain memory resources, and maps the memory resources to the common extended memory of multiple hosts in the memory node.

[0133] Specifically, Figure 5 As shown in the figure, the computing node connects CXL resources to the switch node through MCIO, and connects the CXL resources to the switch node through the CABLE outside the chassis. The switch node obtains the CXL resources of multiple hosts from the computing node. The switch node uses the CXL switch chip as the core to exchange resources, and exchanges and distributes data in the switch node. The physical link of the CXL Fabric is generated in the switch node. The CXL resources of multiple hosts are gathered in the switch node. After passing through the CXL switch chip, the CXL resources are transmitted to the memory node in the form of MCIO and CABLE connection. The memory node is also the memory in a broad sense. The capacity of the memory in the memory node is calculated in TB (terabyte). The MXC chip in the memory node converts the CXL resources into standard memory resources through the GIM function, and the extended memory resources can be remotely mapped to the common extended memory of multiple hosts in the memory node. When the system calls resources, the information between the hosts is shared and can be accessed by other hosts, which lays the foundation for multiple hosts to access each other.

[0134] In this embodiment, the switch node transmits the acquired CXL resources to the memory node, and the memory node converts the CXL resources of multiple hosts into memory resources, maps the memory resources to the common extended memory of multiple hosts in the memory node, and provides a memory resource basis for host interconnection access.

[0135] In some optional implementations, before sending the access request to the switching node, the first host is further configured to:

[0136] In the case where the first host does not have an address decoder, mapping the global integrated storage information in the physical address space to the preset address space, wherein the global integrated storage information is used to determine the first target port number, and the first target port number is used to determine the port for sending the request information to the second host;

[0137] Generate an access request according to the preset address space.

[0138] Specifically, in this embodiment, CXL allows the HOST or device to implement an address decoder through a private solution, that is, to obtain a DPID (destination port identifier) ​​by decoding from the physical address space, so that there is no need to use a preset decoder in the memory node for decoding.

[0139] The GIM function can only complete transactions through UIO, so the access request is a piece of UIO information. CXL allows the host or other devices to implement an address decoder through a private solution, that is, to obtain the first target port number DPID (target port identifier) ​​by decoding from the physical address space. If the host has an address decoder, that is, to implement the address decoder through a private solution, the first target port number DPID can be decoded from any field in the UIO information. Therefore, when the preset decoder of the DRAM in the memory node is used to decode the access request, the preset decoder needs to use the proprietary algorithm of the HOST for decoding.

[0140] If the first host does not have an address decoder, that is, the host does not implement the address decoder through a private solution, when the preset decoder of the DRAM in the memory node is used to decode (DECODE) the access request, the preset decoder will only decode the first target port number DPID from the field corresponding to the preset address space. The preset address space is the Fabric Address Space. In order for the preset decoder to decode the DPID, the first host needs to map the GIM (global integrated memory) information in the physical address space to the preset address space. The GIM information records the memory resource information. Therefore, the first host maps the GIM information in the physical address space to the global integrated memory field in the Fabric address space, such as Figure 7 As shown, Figure 7The left side shows the UIO information after the mapping is completed, including: local other fields (LOCAL OTHER), local memory mapping read and write fields (LOCAL MMIO), global integrated storage fields (GLOBAL INTERGRATED MEMORY), global Fabric-Attached Memory fields (Global Fabric-Attached Memory), unused fields (UNUSED), local CXL memory port X fields (LOCAL CXL MEM PORT X), ..., local CXL memory port 0 fields (LOCAL CXLMEM PORT 0), local memory (direct connection) fields (Local Memory (direct)), where X is an integer greater than 0. The unused field, the local CXL memory port X field, ..., the local CXL memory port 0 field, and the local memory (direct connection) field belong to the local memory address space. The global integrated storage field and the global fabric connection memory field belong to the fabric address space. In this embodiment, the fabric address space is divided into multiple segments. The global integrated storage space field includes segments N to M-1, and the global fabric connection memory space field includes segments 0 to N-1. The specific values ​​of M and N are set according to requirements.

[0141] In addition, the first host can configure CXL as MMIO (Memory-Mapped I / O, memory-mapped read and write) type, that is, the cache type needs to be set to Uncacheable. In the above case, the first host can generate a UIO message as an access request. When the UIO message arrives at the Ingress Port of the CXL Switch, the UIO message can be decoded by the preset decoder of the DRAM in the memory node to obtain the DPID of the target side. The preset decoder is: address decoder (Fabric Address Segment Table, FAST).

[0142] In this embodiment, when the first host does not have an address decoder, the physical address space of the first host is mapped to a preset address space, and an access request is generated based on the preset address space, so that the access request is subsequently decoded using the preset decoder to obtain the first target port number.

[0143] In some optional implementations, the second host is further configured to:

[0144] Determine the directly connected memory and extended memory of the second host;

[0145] A mappable second memory is determined in the direct-connected memory and the extended memory, wherein the second memory is used to map memory resources of the second memory to the first memory in the first host when the second host is accessed by the first host.

[0146] Specifically, in the host interconnection access system, different hosts (HOSTs) are connected together using the CXL switching chip in the switching node. The direct-connected memory in the HOST and the extended memory directly extended through CXL can be accessed by other HOSTs through GIM. The HOST can freely choose whether to map out all or part of its memory. The process of the second host determining the mappable memory is as follows: first determine all the memories of the second host, including: direct-connected memory and extended memory. Among all the memories of the second host, set the mappable second memory. When the second host is accessed by the first host, the memory resources of the second memory are mapped to the first memory in the first host. The first memory is, for example, the DOMAIN CXL memory under any CPU in the first host.

[0147] In this embodiment, the second host determines a mappable second memory in the direct memory and the extended memory, so that when accessed by the first host, the memory resources of the second memory are mapped to the first memory in the first host, thereby realizing host interconnection access.

[0148] In some optional implementations, the switching node includes: a CXL switching card, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables;

[0149] The computing node is connected to the CXL switch card via a first connector and a first cable, and is used to send CXL resources of multiple hosts in the computing node to the CXL switch card;

[0150] The CXL switch card is connected to the memory node via a second connector and a second cable, and is used to send the CXL resource to the memory node.

[0151] Specifically, combined Figure 5This embodiment is described. The switching node includes: a CXL switching board, a first preset number of first connectors, and a first preset number of first cables. The first preset number represents a plurality of connectors, and there is no specific quantity limit here. The first connector is an MCIO connector that connects the CXL switching board in the switching node to the computing node, and the first cable is a physical cable that connects the CXL switching board in the switching node to the computing node. In addition, the switching board includes multiple UP interfaces (uplink interfaces), for example: UP0, UP1, ..., and the UP interface is connected to the MCIO of the computing node through the first connector and the first cable. In addition, the connection between the first connector and the UP interface is a physical link of the CXL Fabric, for example: the link between the first MCIO on the left side of the first connector and UP4, the link between the first MCIO on the right side and UP3, ..., for details, see Figure 5 , I will not go into details here.

[0152] The computing node is connected to the CXL switch card through a first connector and a first cable, and is used to send CXL resources of multiple hosts in the computing node to the CXL switch card.

[0153] The switching node further includes: a second preset number of second connectors and a second preset number of second cables, where the second preset number represents a plurality, and no specific quantity limitation is imposed here, the second connector is an MCIO connector connecting the CXL switching board in the switching node to the memory node, and the second cable is a physical cable connecting the CXL switching board in the switching node to the memory node. In addition, the switching board includes a plurality of DP interfaces (downstream interfaces), and the DP interface is connected to the MCIO of the memory node through the second connector and the second cable.

[0154] The CXL switch card is connected to the memory node via a second connector and a second cable, and is used to send the CXL resource to the memory node.

[0155] In some optional implementations, the memory node includes: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors;

[0156] The memory expansion controller is connected to the CXL switch card via the third connector and the second cable, and is used to obtain CXL resources from the CXL switch card;

[0157] The memory expansion controller is connected to the memory device and is used to convert the CXL resources into memory resources in the memory device.

[0158] Specifically, combined Figure 5This embodiment is described. The memory node includes: a memory expansion controller (MXC chip) and a second preset number of third connectors, the second preset number represents a plurality, and no specific number is limited here, and the third connector is an MCIO connector connecting the memory expansion controller in the memory node to the CXL switch board of the memory node. Figure 5 There are multiple MXC chips in the system, which together constitute the MXC memory expansion board.

[0159] The memory expansion controller is connected to the CXL switch card through the third connector and the second cable to obtain CXL resources from the CXL switch card. The CXL switch card will be routed according to the CXL Fabric supported by the SWITCH chip, and the upstream and downstream numbers are consistent. The downstream port of the CXL switch card in the memory node is connected to the MXC chip on the MXC memory expansion card through the second connector, the second cable, and the third connector to transmit the CXL resources to the MXC chip.

[0160] like Figure 5 As shown, the memory node also includes: a second preset number of memory devices, the memory device is, for example: DIMM ×4 supporting the GIM function, ×4 means that the bit width of the DIMM is 4 times 64 bits. The memory expansion controller is connected to the memory device, manages the memory device, and converts the CXL resources into standard memory resources in the memory device.

[0161] In this embodiment, the CXL resources are converted into memory resources in the memory device through the memory expansion controller to realize memory expansion for the host in the computing node, and the converted memory resources can be accessed by the host, providing a basis for subsequent hosts to realize interconnection access through memory.

[0162] In some optional embodiments, the CXL switch card and the memory expansion controller are configured to be powered on before the host in the computing node issues a system reset instruction;

[0163] The host is configured to be powered on after the CXL switch card and memory expansion controller are powered on.

[0164] Specifically, in the host interconnection access system, the system requires that the CXL switch board (CXL SWITCH board) and the memory expansion controller (MXC memory expansion board) be powered on first, and then the HOST is powered on. It is required that the power-on of the CXL SWITCH board and the MXC memory expansion board be completed before the HOST issues a system reset. Therefore, the CXL switch board and the memory expansion controller are configured to be powered on before the host in the computing node issues a system reset instruction; the host is configured to be powered on after the CXL switch board and the memory expansion controller are powered on.

[0165] In this way, logical training will be performed on the CXL path, so that the system will work normally, memory resources can be expanded, and HOST TO HOST access can be carried out between HOSTs.

[0166] In this embodiment, the power-on sequence of the CXL switch board, the memory expansion controller and the host is set so that the host interconnection access system works normally, and the host memory resources can be expanded while the host interconnection access is realized.

[0167] According to an embodiment of the present invention, a host interconnection access method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0168] In this embodiment, a host interconnection access method is provided. Figure 8 is a flow chart of a host interconnection access method according to an embodiment of the present invention, the method is applied to a switching node, such as Figure 8 As shown, the process includes the following steps.

[0169] Step S801, when an access request is obtained from the first host, the access request is transmitted to the memory node, and the access request is decoded by using a preset decoder in the memory node to obtain request information and a first target port number.

[0170] Specifically, Figure 4 This is a node topology diagram of the server using CXL to expand memory in the cabinet, such as Figure 4As shown in the figure, the computing node can be a dual-way node or a single-way node, that is, it can contain 1 central processing unit or 2 central processing units. The computing node is connected to the switching node through multiple MCIOs. The computing node connects to the CXL resources through MCIOs and transmits the CXL resources to the switching node through cables (CABLE). The CXL resources of multiple HOSTs are aggregated in the switching node. The CXL switching board of the switching node uses the CXL switching chip (CXL SWITCH) as the core to exchange resources. Data is exchanged and distributed in the switching node, and the physical link of the CXL Fabric is generated in the switching node. The switching node is connected to the memory node through multiple MCIOs. The CXL switching chip transmits the CXL resources to the memory node through MCIO and CABLE. The core of the memory node is the memory extension controller (MXC chip), which converts the CXL resources into memory resources in the memory device (DEVICE) that supports the GIM function. The memory resources converted from the CXL resources are remotely mapped through the GIM function and mapped to the common extended memory of the HOST, which provides a basis for multiple HOSTs to access each other.

[0171] like Figure 6 As shown, in this embodiment, a preset decoder is set in the dynamic random access memory (DRAM) of the memory node to decode the access request (DICODE). When the first host needs to access the second host, an access request is generated. The access request is, for example, a UIO request. The UIO allows unordered data transmission on the CXL link, allowing data packets to be sent and received out of order as long as they can eventually be reassembled correctly. The first host sends the access request to the switching node through the MCIO of the computing node and the MCIO of the switching node, and then the access request is transmitted to the CXL switching chip of the switching node through the physical interface and the ingress port (IngressPort) of the switching node. Correspondingly, at the firmware level, the UIO information is sent to the Fabric address range, pointing to the GIM space. When the switching node obtains the access request from the first host, that is, when the UIO request information of the first host arrives at the Ingress Port of the CXL SWITCH chip, as shown Figure 6 As shown, the switch node transmits the access request to the memory node memory expansion controller, and uses the preset decoder of the DRAM in the memory node to decode (DECODE) the access request. The decoded content includes the request information and the first target port number. The first target port number is: the Destination Port ID (Destination Port Identifier, DPID) of which specific path in the Fabric is passed.

[0172] Step S802: Send the request information to the second host through the output port corresponding to the first target port number.

[0173] Specifically, the specified path is obtained through the egress port corresponding to the first target port number, and the request information is sent to the second host through the specified path. Figure 6 As shown, the designated path is the physical interface connected to the memory node, the output port corresponding to the first target port number, the physical interface connected to the switch node and the computing node, the MCIO of the switch node, and the MCIO of the computing node. The request information (UIO REQUEST) is sent to the second host through the designated path.

[0174] Step S803: When the request information returned by the second host is obtained, the returned request information is sent to the first host, indicating that the first host cannot access the second host.

[0175] Specifically, after receiving the request information, namely the UIO information, the second host will perform a validity check on the request information. If the request information is illegal, the second host will not allow the first host to access it and will return the illegal request information.

[0176] Therefore, the switching node obtains the request information returned by the second host and can confirm that the request information is illegal and the first host cannot access the second host. The switching node sends the returned request information to the first host, indicating that the first host cannot access the second host.

[0177] Step S804: When a response message is obtained from the second host, the response message is parsed to obtain response information and a second target port number.

[0178] Specifically, if the request information is legal, the second host converts the UIO information into a response message through the ingress port of the switching node, such as a PBR (Port Based Routing) message. When the PBR message reaches the CXL SWITCH chip in the switching node, the CXL SWITCH chip parses the response message to obtain the response information and the second target port number.

[0179] Step S805, sending a response message to the first host through the output port corresponding to the second target port number, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0180] Specifically, Figure 6As shown, the switching node determines the corresponding output port according to the second target port number, determines the path for transmitting the consent information, determines the physical interface connecting the switching node to the memory node, the corresponding output port for the second target port number, the physical interface connecting the switching node to the computing node, the MCIO of the switching node, and the MCIO of the computing node, and sends the consent information to the first host through the path for transmitting the consent information, indicating that the first host can access the second host by accessing the memory.

[0181] It should be noted that, taking a dual-host system as an example, the system includes a first host and a second host. This HOST refers to an OS domain, and each domain has more than one CPU processor. Therefore, the dual-CPU system is the smallest system that supports CXL Fabric for host interconnection access. The switch node uses the Fabric architecture for internal interconnection and management. All the memory in this system supports the GIM function. It supports mapping its own memory resources to the memory of other HOST DOMAINs (domains) through GIM under the management of CXL Fabric. Therefore, the DOMAIN CXL memory of any CPU can receive the memory resources mapped from the DOMAIN CXL memory of other CPUs, realizing the access of multiple HOSTs through memory interconnection. Among them, HOST DOMAIN refers to the CXL memory expanded under a single CPU, which becomes the HOST DOMAIN CXL memory for this CPU.

[0182] Based on the above content, the step of the first host accessing the second host by accessing the memory includes: using the first memory in the first host to receive the memory resource mapped from the second memory in the second host, the first memory is for example: the DOMAIN CXL memory under any CPU in the first host, and the second memory is for example: the DOMAIN CXL memory of the CPU in the second host. Accessing the second host is achieved through the memory resource.

[0183] The host interconnection access method provided in this embodiment decodes the access request of the first host to obtain access information, and sends the access information to the second host. If the second host returns the access information, it means that the second host cannot be accessed. If the second host returns a response message, the response message is parsed to obtain the response information, and the response information is sent to the first host, indicating that the first memory in the first host receives the memory resources mapped from the second memory in the second host to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the access cost of interconnection between hosts. The problem of lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL is solved.

[0184] In this embodiment, another host interconnection access method is provided. Fig. 9 is a flow chart of a host interconnection access method according to an embodiment of the present invention, the method is applied to a switching node, such as Fig. 9 As shown, the process includes the following steps.

[0185] Step S901, sending an access request to the switching node through the input port of the switching node, and the switching node sending the request information in the access request to the second host through the output port corresponding to the first target port number, wherein the first target port number is obtained after the switching node decodes the access request using a preset decoder in the memory node.

[0186] Specifically, when the first host needs to access the second host, an access request is generated, such as a UIO request. The first host sends the access request to the switch node through the MCIO of the computing node and the MCIO of the switch node, and then transmits the access request to the CXL switch chip of the switch node through the physical interface and the ingress port of the switch node. Correspondingly, at the firmware level, the UIO information is sent to the Fabric address range, pointing to the GIM space.

[0187] The switching node transmits the access request to the memory node memory expansion controller, and uses the preset decoder of the DRAM in the memory node to decode (DECODE) the access request. The decoded content includes the request information and the first target port number, and obtains the designated path through the egress port (Egress Port) corresponding to the first target port number, and sends the request information to the second host from the designated path.

[0188] Step S902: When request information returned by the second host is obtained from the switching node, it is determined that the second host cannot be accessed.

[0189] Specifically, after receiving the request information, namely the UIO information, the second host will perform a validity check on the request information. If the UIO information is not legal, the second host will return the illegal request information.

[0190] Therefore, if the switching node obtains the request information returned by the second host, it indicates that the request information is illegal and the first host cannot access the second host. The switching node sends the returned request information to the first host, indicating that the first host cannot access the second host. When the first host receives the request information returned by the second host, it can be determined that the second host cannot be accessed.

[0191] Step S903, when the consent information is obtained from the switch node, the memory resources mapped from the second memory in the second host are received by the first memory in the first host to access the second host, wherein the memory resources are obtained by converting the CXL resources of multiple hosts by the memory node.

[0192] Specifically, if the request information is legal, the second host converts the UIO information into a response message through the ingress port of the switch node, such as a PBR message. When the PBR message reaches the CXL SWITCH chip in the switch node, the CXL SWITCH chip parses the response message to obtain the response information and the second target port number.

[0193] The switching node determines the corresponding output port according to the second target port number, determines the path for transmitting the acceptance information, and sends the acceptance information to the first host through the path for transmitting the acceptance information, indicating that the first host can access the second host by accessing the memory.

[0194] The mechanism of multiple hosts realizing interconnected access through memory is shown in the embodiment corresponding to step S405, which will not be described here in detail. Based on the above mechanism, the step of the first host accessing the second host through memory access includes: using the first memory in the first host to receive the memory resources mapped from the second memory in the second host, the first memory is for example: the DOMAIN CXL memory under any CPU in the first host, and the second memory is for example: the DOMAIN CXL memory of the CPU in the second host. Access to the second host is realized through memory resources.

[0195] The host interconnection access method provided in this embodiment is that the first host sends an access request containing request information to the switching node, and the switching node sends the request information to the second host. If the second host returns the access information, it means that the second host cannot be accessed. If the first host receives the consent information, the memory resources mapped from the second memory in the second host are received through the first memory to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the access cost of interconnection between hosts. The problem of lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL is solved.

[0196] In this embodiment, a host interconnection access method is provided. Fig.10 is a flow chart of a host interconnection access method according to an embodiment of the present invention, the method is applied to a second host, such as Fig.10 As shown, the process includes the following steps.

[0197] Step S1001: When request information is obtained from a switching node, verify whether the request information is legal.

[0198] Specifically, after receiving the access request sent by the first host, the switch node transmits the access request to the memory node memory expansion controller, and uses the preset decoder of the DRAM in the memory node to decode the access request (DECODE), and the decoded content includes the request information and the first target port number. Through the egress port (Egress Port) corresponding to the first target port number, the specified path is obtained, and the request information is sent to the second host from the specified path.

[0199] After receiving the request information, the second host will verify the legitimacy of the request information.

[0200] Step S1002: if the request information is illegal, the request information is returned to the switching node, and the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host.

[0201] Specifically, if the request information is illegal, the second host does not allow the first host to access it and will return the illegal request information.

[0202] If the switching node obtains the request information returned by the second host, indicating that the request information is illegal and the first host cannot access the second host, the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host.

[0203] Step S1003, if the request information is legal, convert the request information into a response message, send the response message to the switching node, and the switching node sends the response information in the response message to the first host, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0204] Specifically, if the request information is legal, the second host converts the UIO information into a response message through the ingress port of the switch node, such as a PBR message. When the PBR message reaches the CXL SWITCH chip in the switch node, the CXL SWITCH chip parses the response message to obtain the response information and the second target port number.

[0205] The switching node determines the corresponding output port according to the second target port number, determines the path for transmitting the acceptance information, and sends the acceptance information to the first host through the path for transmitting the acceptance information, indicating that the first host can access the second host by accessing the memory.

[0206] The mechanism for multiple hosts to achieve interconnected access through memory refers to the embodiment corresponding to step S405, which will not be repeated here. Based on the above mechanism, the step of the first host accessing the second host by accessing the memory includes: using the first memory in the first host to receive the memory resources mapped from the second memory in the second host, the first memory is for example: the DOMAIN CXL memory under any CPU in the first host, and the second memory is for example: the DOMAIN CXL memory of the CPU in the second host. Access to the second host is achieved through memory resources. The step of the first host accessing the second host by accessing the memory includes: using the first memory in the first host to receive the memory resources mapped from the second memory in the second host, the first memory is for example: the DOMAIN CXL memory under any CPU in the first host, and the second memory is for example: the DOMAIN CXL memory of the CPU in the second host. In the above process, the second host needs to map the memory resources in the second memory to the first memory, so that the first host can access the second host through memory resources. It should be noted that, Figure 6 As shown, the second host includes a local memory, and the memory in the local memory that does not support GIM and the memory resources that only allow local access to the memory cannot be mapped to the second memory in the second host.

[0207] The host interconnection access method provided in this embodiment is that after the second host verifies the access information, if it is illegal, the access information is returned, indicating that the second host cannot be accessed. If it is legal, a response message is returned to instruct the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host. The hosts do not rely on the internal bus of the processor to access each other, do not occupy other channels, and the performance of the host system will not be reduced due to the access process. There is no need to rely on the network card for interconnection access, which reduces the cost of system memory and the cost of interconnection access between hosts. The problem of lack of a fast access mechanism between hosts, which limits the performance of heterogeneous computing and affects the flexibility and cost of using CXL is solved.

[0208] In some optional implementations, the memory expansion controller (CXL chip) monitors the operating parameters of the extended memory in real time. Therefore, through the CXL chip, the health status of the extended memory can be obtained, the memory operating status can be judged, and whether the memory needs to be powered off and replaced. Specifically, it includes:

[0209] Acquire parameter information of a set of operating parameters of each extended memory from the CXL chip to which each extended memory belongs, wherein the operating status information of each extended memory includes parameter information of a set of operating parameters of each extended memory;

[0210] A mailbox event log corresponding to each extended memory is obtained from a CXL chip to which each extended memory belongs, wherein the operating status information of each extended memory includes an error record associated with the operating status of each extended memory extracted from the mailbox event log corresponding to each extended memory.

[0211] The status reference information includes but is not limited to: parameter information of a set of operating parameters of each extended memory, the operating status information of each extended memory includes parameter information of a set of operating parameters of each extended memory, which may include but is not limited to status information such as voltage, temperature, power consumption, etc.; error records associated with the operating status of each extended memory extracted from the mailbox event record log corresponding to each extended memory.

[0212] The baseboard management controller of the memory resource pool can obtain parameter information of a set of operating parameters of each extended memory from the CXL chip to which each extended memory belongs, and obtain a mailbox event record log corresponding to each extended memory from the CXL chip to which each extended memory belongs. For the obtained mailbox event record log corresponding to each extended memory, an error record associated with the operating status of each extended memory can be extracted therefrom. The status reference information corresponding to each extended memory includes: parameter information of a set of operating parameters of each extended memory, and an error record associated with the operating status of each extended memory.

[0213] For example, the BMC of the memory resource pool can collect the memory data transmission status obtained by the MXC end in the following ways: the BMC of the memory resource pool can interact with the MXC end through I2C commands to obtain the voltage, temperature, power consumption and other status information of the memory, and judge its running health status through thresholds; the BMC of the memory resource pool can interact with the MXC end through I2C commands to obtain the Mailbox Event Records Log of the memory, identify the Fatal Error, Failure Error and Warning Error therein, and use them to judge the running health status of the memory.

[0214] The alarm information identified and reported by the host BIOS can be: the host BIOS can identify and report memory errors such as CE (Correctable Error), UCE (UnCorrectable Error), ScrubFail (cleaning failure), Configuration Error, etc., and inform the host BMC of abnormal memory alarm information through the IPMI command AddSel method. The BMC of the memory resource pool then obtains memory status information (From BIOS) from each host BMC through the network through the Redfish command.

[0215] The memory pool BMC integrates the above three memory health states to judge the memory operation status and determine whether it is necessary to power off and replace the memory.

[0216] In this embodiment, a host interconnection access device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0217] This embodiment provides a host interconnection access device, which is deployed in a switching node, such as Fig.11 As shown, including:

[0218] The decoding module 1101 is used for, when an access request is obtained from the first host, transmitting the access request to the memory node, decoding the access request using a preset decoder in the memory node, and obtaining request information and a first target port number;

[0219] The first sending module 1102 is used to send the request information to the second host through the output port corresponding to the first target port number;

[0220] The second sending module 1103 is used to send the returned request information to the first host when the request information returned by the second host is obtained, indicating that the first host cannot access the second host;

[0221] The parsing module 1104 is used to parse the acceptance message when the acceptance message is obtained from the second host to obtain the acceptance information and the second target port number;

[0222] The third sending module 1105 is used to send the consent information to the first host through the output port corresponding to the second target port number, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0223] This embodiment provides a host interconnection access device, which is deployed on a first host, such as Fig.12 As shown, including:

[0224] The fourth sending module 1201 is used to send an access request to the switching node through the input port of the switching node, and the switching node sends the request information in the access request to the second host through the output port corresponding to the first target port number, wherein the first target port number is obtained after the switching node decodes the access request using a preset decoder in the memory node;

[0225] A determination module 1202, configured to determine that the second host cannot be accessed when request information returned by the second host is obtained from the switching node;

[0226] The access module 1203 is used to receive memory resources mapped from the second memory in the second host by using the first memory in the first host to access the second host when the consent information is obtained from the switch node, wherein the memory resources are obtained by converting CXL resources of multiple hosts by the memory node.

[0227] This embodiment provides a host interconnection access device, which is deployed on a second host, such as Fig.13 As shown, including:

[0228] The verification module 1301 is used to verify whether the request information is legal when the request information is obtained from the switching node;

[0229] The information return module 1302 is used to return the request information to the switching node when the request information is illegal, and the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host;

[0230] The fifth sending module 1303 is used to convert the request information into a response message when the request information is legal, and send the response message to the switching node, which sends the response information in the response message to the first host, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory resources are obtained after the memory node converts the CXL resources of multiple hosts.

[0231] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0232] The host interconnection access device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, including a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0233] The embodiment of the present invention also provides a computer device having the above Fig.11 or Fig.12 or Fig.13 The host interconnect access device is shown.

[0234] See also Fig.14 , Fig.14 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Fig.14 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Fig.14 A processor 10 is taken as an example.

[0235] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0236] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0237] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0238] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0239] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0240] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0241] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0242] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A host interconnection access system, characterized in that: The system includes: a computing node, a switching node and a memory node; the computing node includes a first host and a second host; the switching node includes: a CXL switching board, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors and a second preset number of second cables; the memory node includes: a memory expansion controller, a second preset number of memory devices and a second preset number of third connectors; The first host is connected to the switching node and is used to send an access request to the switching node; The switch node is connected to the memory node, and is used for, when an access request is obtained from the first host, transmitting the access request to the memory node, and decoding the access request using a preset decoder in the memory node to obtain request information and a first target port number; The switching node is connected to the second host and is used to send the request information to the second host through the output port corresponding to the first target port number; The second host is used to return the request information to the switching node, or send a response message to the switching node; The switching node is further used to send the returned request information to the first host, indicating that the first host cannot access the second host, or to send the consent information in the consent message to the first host, indicating that the first host uses the first memory in the first host to receive the memory resources mapped from the second memory in the second host, so as to access the second host, wherein the memory node is used to convert the CXL resources of multiple hosts into memory resources in the memory device through the memory expansion controller, and the second memory is used to map the memory resources to the first memory.

2. The system according to claim 1, characterized in that The switching node sends the response information in the response message to the first host, including: Removing the message source identifier and the message destination identifier in the acceptance message to obtain a processed message; Acquire a second target port number from the processed message, and use the preset information in the processed message as the consent information; The response information is sent to the first host through the output port corresponding to the second target port number.

3. The system according to claim 1, characterized in that The second host returns the request information to the switching node, including: Acquire a first transmission path of the request information; Returning the request information to the switching node via the first transmission path; The second host sends a response message to the switching node, including: determining a second transmission path different from the first transmission path; The response message is sent to the switching node through the second transmission path.

4. The system according to claim 1, characterized in that The switching node is also used for: Acquire CXL resources of the plurality of hosts from a computing node; The CXL resources are transmitted to a memory node, and the memory node transforms the CXL resources of the multiple hosts to obtain the memory resources, and maps the memory resources to a common extended memory of the multiple hosts in the memory node.

5. The system according to claim 1, characterized in that Before sending the access request to the switching node, the first host is further configured to: In the case where the first host does not have an address decoder, mapping the global integrated storage information in the physical address space to a preset address space, wherein the global integrated storage information is used to determine the first target port number, and the first target port number is used to determine the port for sending the request information to the second host; The access request is generated according to the preset address space.

6. The system according to claim 1, characterized in that The second host is also used for: Determine the directly connected memory and extended memory of the second host; In the direct-connected memory and the extended memory, a mappable second memory is determined, wherein the second memory is used to map memory resources of the second memory to the first memory in the first host when the second host is accessed by the first host.

7. The system according to claim 1, characterized in that The computing node is connected to the CXL switch card through the first connector and the first cable, and is used to send CXL resources of multiple hosts in the computing node to the CXL switch card; The CXL switch card is connected to the memory node through the second connector and the second cable, and is used to send the CXL resource to the memory node.

8. The system according to claim 7, characterized in that The memory expansion controller is connected to the CXL switch card through the third connector and the second cable, and is used to obtain the CXL resource from the CXL switch card; The memory expansion controller is connected to the memory device and is used to convert the CXL resources into memory resources in the memory device.

9. The system according to claim 8, characterized in that The CXL switch card and the memory expansion controller are configured to be powered on before the host in the computing node issues a system reset instruction; The host is configured to be powered on after the CXL switch card and the memory expansion controller are powered on.

10. A host interconnection access method, characterized in that: The method is applied to a switching node, and the method comprises: In the case of obtaining an access request from the first host, the access request is transmitted to the memory node, and the access request is decoded by using a preset decoder in the memory node to obtain request information and a first target port number, wherein the switch node includes: a CXL switch card, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables; the memory node includes: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors; Sending the request information to the second host through the output port corresponding to the first target port number; In the case of obtaining the request information returned by the second host, sending the returned request information to the first host, indicating that the first host cannot access the second host; When a response message is obtained from the second host, the response message is parsed to obtain response information and a second target port number; The response information is sent to the first host through the output port corresponding to the second target port number, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host, so as to access the second host, wherein the memory node is used to convert the CXL resources of multiple hosts into memory resources in the memory device through the memory expansion controller, and the second memory is used to map the memory resources to the first memory.

11. A host interconnection access method, characterized in that: The method is applied to a first host, and the method comprises: An access request is sent to the switching node through an input port of the switching node, and the switching node sends the request information in the access request to the second host through an output port corresponding to a first target port number, wherein the first target port number is obtained after the switching node decodes the access request using a preset decoder in a memory node, the switching node comprises: a CXL switching board, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables; the memory node comprises: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors; In the case of obtaining request information returned by the second host from the switching node, determining that the second host cannot be accessed; When the consent information is obtained from the switching node, the memory resources mapped from the second memory in the second host are received by using the first memory in the first host to access the second host, wherein the memory node is used to convert the CXL resources of multiple hosts into memory resources in the memory device through the memory expansion controller, and the second memory is used to map the memory resources to the first memory.

12. A host interconnection access method, characterized in that: The method is applied to a second host, and the method comprises: In the case of obtaining request information from the switching node, verifying whether the request information is legal, wherein the switching node includes: a CXL switching board, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables; If the request information is illegal, the request information is returned to the switching node, and the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host; If the request information is legal, the request information is converted into a response message, and the response message is sent to the switching node. The switching node sends the response information in the response message to the first host, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host to access the second host, wherein the memory node includes: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors, and the memory node is used to convert CXL resources of multiple hosts into memory resources in memory devices through the memory expansion controller, and the second memory is used to map memory resources to the first memory.

13. A host interconnection access device, characterized in that: The device is deployed in a switching node, and includes: A decoding module, configured to, when an access request is obtained from a first host, transmit the access request to a memory node, decode the access request using a preset decoder in the memory node, and obtain request information and a first target port number, wherein the switch node comprises: a CXL switch card, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables; the memory node comprises: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors; A first sending module, used for sending the request information to the second host through the output port corresponding to the first target port number; A second sending module is used for, when obtaining the request information returned by the second host, sending the returned request information to the first host, indicating that the first host cannot access the second host; A parsing module, configured to parse the response message to obtain the response information and the second target port number when the response message is obtained from the second host; a third sending module, configured to send the consent information to the first host through the output port corresponding to the second target port number, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host, so as to access the second host, wherein the memory node is configured to convert the CXL resources of multiple hosts into memory resources in the memory device through the memory expansion controller, and the second memory is configured to map the memory resources to the first memory.

14. A host interconnection access device, characterized in that: The device is deployed on a first host, and includes: a fourth sending module, configured to send an access request to the switching node through an input port of the switching node, and the switching node sends the request information in the access request to the second host through an output port corresponding to a first target port number, wherein the first target port number is obtained after the switching node decodes the access request using a preset decoder in a memory node, the switching node comprising: a CXL switching board, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables; the memory node comprising: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors; a determination module, configured to determine that the second host cannot be accessed when request information returned by the second host is obtained from the switching node; An access module is used to, when obtaining the consent information from the switching node, use the first memory in the first host to receive the memory resources mapped from the second memory in the second host, so as to access the second host, wherein the memory node is used to convert the CXL resources of multiple hosts into memory resources in the memory device through the memory expansion controller, and the second memory is used to map the memory resources to the first memory.

15. A host interconnection access device, characterized in that: The device is deployed on a second host, and the device includes: A verification module, configured to verify whether the request information is legal when the request information is obtained from the switching node, wherein the switching node comprises: a CXL switching board, a first preset number of first connectors, a first preset number of first cables, a second preset number of second connectors, and a second preset number of second cables; An information return module, used for returning the request information to the switching node when the request information is illegal, and the switching node sends the returned request information to the first host, indicating that the first host cannot access the second host; a fifth sending module, configured to convert the request information into a response message if the request information is legal, and send the response message to the switching node, and the switching node sends the response information in the response message to the first host, instructing the first host to use the first memory in the first host to receive the memory resources mapped from the second memory in the second host, so as to access the second host, wherein the memory node includes: a memory expansion controller, a second preset number of memory devices, and a second preset number of third connectors, and the memory node is used to convert the CXL resources of multiple hosts into memory resources in the memory device through the memory expansion controller, and the second memory is used to map the memory resources to the first memory.

16. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the host interconnection access method according to any one of claims 10 to 12 by executing the computer instructions.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the host interconnection access method according to any one of claims 10 to 12.

Citation Information

Patent Citations

  • Memory extension method and device, equipment and storage medium

    CN114020655A

  • Memory optimization method and system for software and hardware collaborative design under heterogeneous memory situation

    CN117827464A

  • Memory resource access system and method, storage medium and electronic equipment

    CN118193408A

  • Memory access method, system and device

    CN118276773A