Out-of-band signal transmission method and system

Abstract

The invention discloses an out-of-band signal transmission method and system, and relates to the technical field of chip processing, nodes in a chip are divided into a plurality of groups of nodes, each group of nodes is managed by an independent routing information transfer station, the nodes are connected with the routing information transfer stations through channels, and the routing information transfer stations belong to the nodes. The routing information transfer stations are connected through an asynchronous processing unit, logic-level routing replaces physical-level full interconnection, the wiring density and the occupancy rate of a metal layer are remarkably reduced, the rear-end layout wiring convergence is improved, the first routing information transfer station accurately determines a target node for out-of-band signal transmission through a routing table, and the routing efficiency is improved. If it is determined that the target node set contains the nodes belonging to the second group of nodes, the out-of-band signal is sent to a second routing information transfer station through the asynchronous processing unit, the second routing information transfer station inquires the routing table to determine specific second node information and transmits the out-of-band signal to the second node, the transmission path is shortened, and the transmission efficiency is improved. And delay and extra power consumption overhead are reduced.

Description

Technical Field

[0001] This application relates to the field of chip processing technology, and in particular to an out-of-band signal transmission method and system. Background Technology

[0002] With the evolution of semiconductor process technology and breakthroughs in heterogeneous integration technology, modern System-on-Chip (SOC) is rapidly developing towards high complexity, high density interconnection, and multi-domain collaboration. Against this backdrop, how to achieve low-latency, high-reliability out-of-band (OOB) signal transmission and state synchronization between nodes, while taking into account the scalability and power efficiency of chip physical design, has become a key condition restricting the design of high-performance SOCs.

[0003] In related solutions, out-of-band signals are usually achieved by point-to-point hardwired connections to directly connect nodes. While this method can ensure the determinism of signal transmission in small-scale node scenarios, when the number of nodes increases dramatically, the number of ports and physical connections expand with the node scale, which can cause wiring congestion, waste of metal layer resources, and other issues. Furthermore, the static characteristics of hardwired connections cannot adapt to dynamic topology requirements, and firmware layer needs to frequently intervene in routing configuration, introducing additional latency and power consumption overhead. Summary of the Invention

[0004] This application provides an out-of-band signal transmission method and system to at least solve the problems in related technologies where point-to-point hard-wired connections cause wiring congestion and waste of metal layer resources as the number of nodes increases dramatically.

[0005] This application provides an out-of-band signal transmission method applied to a first routing information relay station in a chip. The chip includes at least two groups of nodes. The first routing information relay station manages the routing information of the first group of nodes. The first routing information relay station is connected to a second routing information relay station via an asynchronous processing unit. The second routing information relay station manages the routing information of the second group of nodes. The method includes: responding to receiving an out-of-band signal sent by a first node in the first group of nodes, querying the routing table corresponding to the first node to determine a target node set; when it is determined that the target node set contains nodes belonging to the second group of nodes, sending the out-of-band signal to the second routing information relay station via the asynchronous processing unit, so that the second routing relay station queries the routing table corresponding to the first node to determine the second node information belonging to the second group of nodes in the target node set, and sending the out-of-band signal to the second node.

[0006] This application also provides an out-of-band signal transmission method applied to a second routing information relay station in a chip. The chip includes at least two groups of nodes. A first routing information relay station is used to manage the routing information of a first group of nodes. The first routing information relay station is connected to a second routing information relay station through an asynchronous processing unit. The second routing information relay station is used to manage the routing information of a second group of nodes. The method includes: receiving an out-of-band signal sent by the first routing information relay station through the asynchronous processing unit; querying the routing table corresponding to the first node; determining the second node information belonging to the second group of nodes in the target node set; and sending the out-of-band signal to the second node. This application also provides an out-of-band signal transmission system, the system including a first routing information relay station and a second routing information relay station in a chip. The chip contains at least two groups of nodes. The first routing information relay station is used to manage the routing information of the first group of nodes. The first routing information relay station is connected to the second routing information relay station through an asynchronous processing unit. The second routing information relay station is used to manage the routing information of the second group of nodes. In response to receiving an out-of-band signal sent by a first node in the first group of nodes, the first routing information relay station queries the routing table corresponding to the first node to determine the target node set. When the first routing information relay station determines that the target node set contains nodes belonging to the second group of nodes, it sends the out-of-band signal to the second routing information relay station through the asynchronous processing unit. The second routing information relay station receives the out-of-band signal sent by the first routing information relay station through the asynchronous processing unit, queries the routing table corresponding to the first node to determine the second node information belonging to the second group of nodes in the target node set, and sends the out-of-band signal to the second node.

[0007] The out-of-band signal transmission system designed in this application divides the nodes within the chip into multiple groups. Each group of nodes is managed by an independent routing information relay station. Nodes are connected to their respective routing information relay stations through channels, and the routing information relay stations are connected through asynchronous processing units. This achieves the replacement of physical-level full interconnection with logical-level routing, significantly reducing wiring density and metal layer occupancy, and improving back-end layout and routing convergence. Furthermore, the first routing information relay station accurately determines the target node for out-of-band signal transmission through the routing table. If the target node set contains a node belonging to the second group of nodes, the out-of-band signal is sent to the second routing information relay station through the asynchronous processing unit. The second routing information relay station queries the routing table to determine the specific second node information and transmits the out-of-band signal to the second node, thereby shortening the transmission path and reducing latency and additional power consumption. Attached Figure Description

[0008] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic diagram of the port of an out-of-band signal transmission system according to an embodiment of the present invention; Figure 2 This is a structural example diagram of a chip proposed according to an embodiment of the present invention; Figure 3 This is a structural example diagram of a signal processing unit according to an embodiment of the present invention; Figure 4 This is an example diagram illustrating the processing logic of the arbitration unit setting the allocation number according to an embodiment of the present invention; Figure 5 This is a structural example diagram of a distribution unit according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the interaction between ports of an out-of-band signal transmission system proposed according to an embodiment of the present invention; Figure 7 This is a flowchart illustrating an out-of-band signal transmission method according to an embodiment of the present invention; Figure 8 This is a structural block diagram of an electronic device according to an embodiment of this invention. Detailed Implementation

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0011] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0012] Modern System-on-a-Chip (SoC) is rapidly evolving towards higher complexity, higher density interconnects, and multi-domain collaboration. Especially in typical application scenarios such as Multi-Processor System on Chip (MPSOC), which is a complex chip that integrates multiple different or similar processor cores (such as CPU, GPU, DSP, NPU, etc.) and peripherals, storage, and dedicated accelerators on the basis of traditional SOC, the core is to achieve multi-task parallel processing and performance acceleration in specific scenarios through heterogeneous or homogeneous multi-core architecture, which is a key technical solution to support high-performance and multi-functional electronic devices, high-speed interconnect controllers (such as PCIe Switch chips (PCIe Switch (PCI Express Switch chip is a special integrated circuit used to expand and manage PCIe bus topology, whose core function is to establish a flexible high-speed data path between multiple PCIe devices (such as graphics cards, NVMe SSDs, network cards, etc.) and the host CPU, realize the interconnection between devices, data forwarding and bandwidth allocation, and is a key component for building multi-device PCIe systems), power management units (PMU), security monitoring modules, and distributed memory architectures, the number of nodes inside the SOC system grows exponentially, and the communication topology between nodes also exhibits dynamic and non-deterministic characteristics.

[0013] Against this backdrop, achieving low-latency, high-reliability out-of-band (OOB) signal transmission and state synchronization between nodes, while simultaneously considering the scalability and power efficiency of chip physical design, has become a key challenge restricting high-performance SoC design. In traditional solutions, out-of-band signals are usually directly connected between nodes using point-to-point hardwired connections. While this method can guarantee deterministic signal transmission in small-scale node scenarios, its inherent defects become increasingly significant as the number of nodes increases: for example, the number of ports and connections expands quadratically with the node scale, and the static characteristics of hardwired connections cannot adapt to dynamic topology requirements (such as hot-swappable device access, node state anomaly isolation, etc.), requiring frequent firmware intervention in routing configuration, introducing additional latency and power consumption overhead. Furthermore, in high-frequency scenarios, long-distance hardwired connections are prone to clock skew and timing violations across clock domains, forcing designers to reduce the operating frequency or increase the number of buffer stages, severely limiting the upper limit of system performance.

[0014] To address the aforementioned issues, embodiments of this application design an out-of-band signal transmission system. Through an innovative hierarchical design paradigm, the out-of-band signal communication mechanism of a multi-node SOC is reconstructed. Specifically, this application divides the physical layout on-chip, introduces a centralized routing information relay station and an asynchronous processing unit, and combines cross-clock domain synchronization, time arbitration, and route-driven distribution technologies to achieve unified processing and flexible routing of out-of-band signals.

[0015] Specifically, such as Figure 1 As shown, an embodiment of this application provides a vehicle upgrade system. The system includes a first routing information relay station 1 and a second routing information relay station 2 within a chip. The chip contains at least two sets of nodes. The first routing information relay station 1 manages the routing information of the first set of nodes. The first routing information relay station 1 is connected to the second routing information relay station 2 via an asynchronous processing unit. The second routing information relay station 2 manages the routing information of the second set of nodes. In response to receiving an out-of-band signal sent by a first node in the first set of nodes, the first routing information relay station 1 queries the routing table corresponding to the first node to determine the target node set. When the first routing information relay station 1 determines that the target node set includes nodes belonging to the second set of nodes, it sends the out-of-band signal to the second routing information relay station 2 via the asynchronous processing unit. The second routing information relay station 2 receives the out-of-band signal sent by the first routing information relay station 1 via the asynchronous processing unit, queries the routing table corresponding to the first node, determines the information of the second node belonging to the second set of nodes in the target node set, and sends the out-of-band signal to the second node.

[0016] This application does not limit the type of chip; any chip containing multiple nodes with variable node relationships is acceptable, such as multiprocessor system chips, power management chips, security monitoring chips, and PCIe switch chips. Taking a PCIe switch chip as an example, the nodes in the chip are the corresponding upstream port node (USP) and downstream port node (DSP), used to connect PCIe master devices, such as... Figure 2As shown, all nodes are divided into left and right groups according to the on-chip layout. The nodes on the left are the first group of nodes, managed by the first routing information relay station 1, and the nodes on the right are the second group of nodes, managed by the second routing information relay station 2. Each routing information relay station can transmit out-of-band information to each node in the corresponding group through dual channels. The first channel is responsible for receiving out-of-band signals sent by nodes within the same group, and the second channel is responsible for receiving out-of-band signals sent by nodes across groups. When a node or channel has no task processing for a long time, i.e., it is in an idle state, the power supply of the node or the transmission link of the channel can be dynamically turned off, retaining only the necessary wake-up mechanism, and based on the signal type (such as emergency interrupt, periodic state)... (State update) Implement differentiated power consumption strategy to achieve Pareto optimality of performance and energy efficiency. The number of node groups and routing information relay stations can be designed according to the actual chip layout. Based on the division of nodes into several groups within the chip, the corresponding number of routing information relay stations can be designed. This is just an example. In this embodiment, considering that the node layout within the chip is relatively dispersed and the nodes are often located at the edge of the chip, if the traditional method is used for hard connection, timing problems will occur. Moreover, due to the large physical distance, the nodes are in a synchronous and asynchronous relationship. Therefore, the first routing information relay station 1 and the second routing information relay station 2 can be connected through an asynchronous processing unit.

[0017] Furthermore, each channel can transmit multiple types of out-of-band signals in parallel. This means each channel includes multiple parallel signal transmission management modules, containing, but not limited to, out-of-band information such as interrupts, power management, and device credit values. Out-of-band interrupts are emergency request signals sent by the device to the master controller (e.g., CPU). Their core function is to break the conventional waiting pattern and prioritize response. Functionally, when a device has an urgent need (e.g., receiving new data or encountering an error), it doesn't need to wait for the master controller to "actively query"; it directly sends an interrupt signal through the out-of-band channel, forcing the master controller to pause its current task and prioritize processing the device's request. Out-of-band power management signals are control information used between the master controller and the device to synchronize "power consumption status." It is about "adjusting power consumption on demand to balance performance and energy saving." Its functions include: the main controller sending power commands (such as "enter low-power sleep" or "wake up and resume work") to devices via out-of-band channels, or devices reporting their current power consumption status to the main controller (such as "ready for sleep" or "requires higher power supply"), avoiding the consumption of resources by transmitting power commands in the main data channel; and the out-of-band device credit value signal, which is a "data reception capacity quota" agreed upon between devices. The corresponding out-of-band signal is used to synchronize this "quota," with the core purpose of "avoiding excessive data transmission and preventing receiver overload." Its function is: the receiving device informs the sending device of its current "credit value" (e.g., it can still receive 10 data frames) via the out-of-band channel; the sending end can only send data according to this "quota." After transmission, the credit value decreases, and the receiving end replenishes the credit value via the out-of-band signal after processing the data, forming dynamic flow control.

[0018] In this embodiment of the invention, the first node in the first group of nodes can initiate an out-of-band signal transmission request. The out-of-band signal may include data and source node information. Upon receiving the out-of-band signal from the first node in the first group of nodes, the first routing information relay station 1 can query the routing table corresponding to the first node based on the source node information to determine the target node set and whether the target node set contains nodes from the second group of nodes. Fixed routing relationships between nodes can be pre-configured according to chip usage. For example, if there are 100 nodes, with 50 nodes in the first group and 50 in the second group, the corresponding routing table size can be 100*100 bits. The source node information includes... The corresponding routing table read address is obtained. Each entry read from this address indicates the target node set of the first node. For the first routing information relay station 1, bits [49:0] in its routing table represent the routing relationship between the first group of nodes, and bits [99:50] represent the node routing relationship of the second group of nodes. Any valid bit in the high 50 bits (e.g., a bit of 1) indicates that the out-of-band signal needs to be transmitted across groups (i.e., the target node set includes nodes of the second group). The out-of-band signal can then be sent to the second routing information relay station 2 through the asynchronous processing unit. The asynchronous processing unit can use a handshake protocol or asynchronous first-in-first-out (FIFO) protocol. First-Out (FIFO) enables reliable data transmission between relay stations, overcoming the problem of asynchronous transmission due to physical dispersion of nodes. If there are valid bits in the lower 50 bits, the specific valid bits can be queried. For example, if bits 0, 1, 7, and 59 of the routing information are 1 and other bits are 0, it indicates that the target node set includes nodes 1_1, 1_2, 1_8, and 2_60. It can be specifically determined that nodes 1_1, 1_2, and 1_8 belong to the in-group out-of-band signal transmission. Then, the port information of the nodes to be transmitted in the group can be queried, and the out-of-band signal can be sent to the above-mentioned in-group nodes. During the out-of-band signal transmission process, there is no back pressure, ensuring that the signal is ultimately accurately distributed.

[0019] After receiving the out-of-band signal sent by the first routing information relay station 1 through the asynchronous processing unit, the second routing information relay station 2 can query the routing table corresponding to the first node to determine the specific second node information belonging to the second group of nodes in the target node set. For example, if the routing table is searched and it is determined that node 2_60 belongs to the second group of nodes, then the second node can be determined to be node 2_60, and then the out-of-band signal can be sent to node 2_60. This is just an example.

[0020] The out-of-band signal transmission system designed in this invention divides the nodes within the chip into multiple groups of nodes. Each group of nodes is managed by an independent routing information relay station. Nodes are connected to their respective routing information relay stations through channels, and the routing information relay stations are connected through asynchronous processing units. This achieves logical-level routing instead of physical-level full interconnection, significantly reducing wiring density and metal layer occupancy, and improving back-end layout and routing convergence. Furthermore, the first routing information relay station 1 accurately determines the target node for out-of-band signal transmission through a routing table. If the target node set contains a node belonging to the second group of nodes, the out-of-band signal is sent to the second routing information relay station 2 through an asynchronous processing unit. The second routing information relay station 2 queries the routing table to determine the specific second node information and transmits the out-of-band signal to the second node, thereby shortening the transmission path and reducing latency and additional power consumption.

[0021] In one optional implementation, before querying the routing table corresponding to the first node, the first routing information relay station 1 may also perform cross-clock domain synchronization processing on the out-of-band signal; and convert the synchronized out-of-band signal into an out-of-band signal of a preset transmission standard type.

[0022] like Figure 2 As shown, the routing information relay station (including the first and second routing information relay stations 2) designed in this application embodiment includes at least a signal processing unit, an arbitration unit, a routing information management unit, and a distribution unit. The signal processing unit has two internal paths: a sending path and a receiving path. The sending path is responsible for cross-clock domain processing of out-of-band signals sent by nodes (from the node clock to the routing information relay station clock), preprocessing of different signal types, and handshake control with the routing information relay station. Since handshake processing is required during transmission, the signals can be uniformly converted into out-of-band signals of a preset transmission standard type. Taking level-based signals as an example, level-based signals can represent signal meaning simply through continuous high and low levels. The transmission protocol logic is simple, hardware requirements are low, and it is compatible with handshake interaction, ensuring transmission reliability. Specifically, as shown... Figure 3As shown, the signal processing unit includes an asynchronous processing unit, a type conversion unit, and a transmitting-side handshake control. The out-of-band signal transmission request sent by the first node first enters the signal processing unit. The asynchronous processing unit can perform cross-clock domain processing on the out-of-band signal sent by the first node. The method of cross-clock domain processing is not limited; it can be achieved through two-stage flip-flops or an asynchronous FIFO to eliminate metastability risks. This is just one example. The type conversion unit then performs standardized preprocessing on the synchronized out-of-band signal, i.e., uniformly converting it into a level-based signal before transmission (i.e., implementing data format shaping). The transmitting-side handshake control of the transmission path can be used to inform the first routing information relay station 1 that it has completed the process. After processing the out-of-band signal prompts, the first routing information relay station 1 can achieve reliable transmission through the asynchronous processing unit, solving the problem of synchronous and asynchronous transmission caused by the physical dispersion of nodes, unifying the management of clock domain isolation, optimizing the complexity of back-end layout and wiring, and reducing long-distance hardwires through the centralized routing information relay station architecture. Combined with the cross-clock domain unified processing mechanism, it significantly optimizes the timing convergence in high-frequency scenarios. Among them, optimizing the timing convergence in high-frequency scenarios refers to, in the process of integrated circuit design, for scenarios such as high-frequency clock signals or high-speed data transmission, using a series of technical means and methods to ensure that the circuit meets the predetermined time constraints, ensuring that the signal can accurately reach the target location within the specified time, thereby achieving the normal operation of the system.

[0023] This application eliminates metastability propagation and ensures signal transmission reliability by performing cross-clock domain synchronization processing on out-of-band signals. It also converts the synchronized out-of-band signals into out-of-band signals of a preset transmission standard type, thereby achieving a unified signal transmission format and ensuring the reliability of the converted signals during transmission.

[0024] In one optional implementation, before querying the routing table corresponding to the first node, the first routing information relay station 1 may add the request representing the first node to a sorting queue; when the sorting queue schedules the request for the first node, the step of querying the routing table corresponding to the first node is executed.

[0025] In this embodiment, the signal processing unit can send the processed out-of-band signal to the arbitration unit. The arbitration unit can allocate arbitration rights based on a first-in-first-out strategy. The first node to obtain arbitration rights can send out-of-band signals (including data and source node information) to the distribution unit. The arbitration unit can monitor node request signals REQ_0 to REQ_N in real time. REQ_N represents the request signal for N+1 nodes to send out-of-band signals. If REQ_x is detected to change from low level to high level, it indicates that the x-th node has initiated a request. The request representing the x-th node (i.e., the first node) can be added to the sorting queue. When the sorting queue reaches the request of the first node, the arbitration right can be sent to the first node.

[0026] This invention ensures the timing correctness of out-of-band signal transmission through a first-in-first-out (FIFO) strategy.

[0027] Furthermore, the arbitration unit adds the request representing the first node to the sorting queue, including: incrementing the counter record value of the current group to obtain the latest counter record value; using the latest counter record value as the allocation number of the request representing the first node; adding the request representing the first node to the sorting queue based on the allocation number; and resetting the counter after the latest counter record value is the maximum counter value, creating a new group, and returning to the step of incrementing the counter record value of the current group.

[0028] In this embodiment, the arbitration unit can determine the scheduling order of requests from each node using a counter. During the initialization phase, the arbitration unit can reset the global counter T=0 and clear all node tag registers T_x=0, where x represents the node number. For example, x=1_2 indicates that node 1_2 initiated the request. When a request from a subsequent node enters the arbitration unit, the counter can be incremented by one, serving as the arbitration batch number (i.e., allocation number) for that node, facilitating subsequent scheduling. When the request signal of the first node changes from low to high, it is determined that the request of the first node is valid. The counter value of the current group is then incremented by one to obtain the latest counter value, which is used as the allocation number representing the request of the first node. Based on the determined allocation number, the request representing the first node is added to the sorting queue. At this point, it can also be determined whether the latest counter value has reached the maximum counter limit. Figure 4 As shown, taking 16 as an example of the maximum counter value, when the latest recorded value of the counter is 16, after setting 16 as the allocation number of the node, the counter can be reset and a new group can be created to recount subsequent node requests to prevent the counter from overflowing. For example, after node 1_3 sends a request, the request of node 1_3 can be placed in the newly set group, and the reset counter can be incremented by one to obtain the latest recorded value of the counter, which can be used as the allocation number of node 1_3. This is just an example.

[0029] In this embodiment, the latest recorded value of the counter can also be divided by the maximum limit of the counter, i.e., (T+1) mod(S), where S represents the maximum limit of the counter and T+1 represents the latest recorded value of the counter. The quotient and remainder are taken, with the quotient indicating which group to schedule and the remainder indicating the allocation number. For example, if the initial recorded value of the counter is 0, and after node 1_2 initiates its first request, the latest recorded value of the counter is 1. Dividing it by 16, the quotient is 0 and the remainder is 1, so the allocation number Batch_ID of node 1_2 is 2. After node 1_3 initiates its 17th request, the latest recorded value of the counter is 17. Dividing it by 16, the quotient is 1 and the remainder is 1, so node 1_3 can be allocated to the newly set group to start a new arbitration cycle, and its allocation number Batch_ID is 1. This is just an example, and the newly created group is still ranked after the previously created group.

[0030] This application uses an in-group counter to generate allocation numbers for node requests, avoiding request sorting conflicts. Furthermore, the design of resetting the counter after reaching its maximum value and creating a new lease avoids counter overflow and adapts to the long-term expansion of chip node scale.

[0031] Furthermore, when the first node includes multiple nodes from the first group of nodes, the multiple nodes are added to the sorting queue according to a preset priority, so that when the allocation number is polled, each node is scheduled in turn according to the preset priority.

[0032] In this embodiment of the application, when the first node is detected to include multiple nodes from the first group of nodes (i.e., multiple nodes send out-of-band signal transmission requests in parallel), the multiple nodes can be managed uniformly by the aforementioned determined allocation number. Furthermore, the multiple nodes can be added to a sorting queue according to a preset priority. Subsequently, after polling the allocation number, each node can be scheduled sequentially according to the preset priority, for example... Figure 4 As shown, nodes 1_6 and 1_10 concurrently request out-of-band signal transmission. Their allocation number Batch_ID is set to 8. When allocation number 8 is polled, the out-of-band signals corresponding to nodes 1_6 and 1_10 are processed. At this time, since node 1_6 has a higher priority, node 1_6 can be given priority to obtain the arbitration right. This realizes that based on the time order strategy, the request of the node with higher priority is given priority. After the handshake request for data transmission initiated by the node, the data and valid signals must be maintained before obtaining the arbitration right. This is only an example.

[0033] When multiple nodes in the first group of nodes simultaneously initiate out-of-band signal requests, this application can manage them using the same allocation number, simplifying the queue logic for concurrent requests from multiple nodes. It can also schedule nodes sequentially according to node priority, accurately matching the functional importance differences of out-of-band signals from different nodes, and ensuring the timeliness of requests from critical nodes.

[0034] The arbitration unit can send the out-of-band signal (including data and source node information) corresponding to the first node that has obtained the arbitration right to the distribution unit. The distribution unit can query the routing table based on the source node information in the out-of-band signal to determine the target node set. If the target node set includes nodes within the group, the out-of-band signal can be directly transmitted to the signal processing unit in the first routing information relay station 1, so that the signal processing unit can perform clock domain processing and type conversion on the out-of-band signal and send it to the corresponding node within the group. If the target node set includes nodes across groups, the distribution unit can transmit the out-of-band signal to the distribution unit in the second routing information relay station 2 through the asynchronous processing unit. Before sending the out-of-band signal to the second routing information relay station 2 or to the signal processing unit within the group through the asynchronous processing unit, the distribution unit can also perform buffer processing on the out-of-band signal. After detecting that the previous transmission task has been completed, the next out-of-band signal to be transmitted is read from the data buffer and sent to the signal processing unit within the group and / or sent to the second routing information relay station 2 through the asynchronous processing unit.

[0035] like Figure 5 As shown, the distribution unit includes at least a data distribution module, a data caching module, and a routing information query / arbitration module. The routing information query / arbitration module of the distribution unit can query the routing table based on the source node information in the out-of-band signal output by the arbitration unit to determine the target node set. The queried routing information is used as the mask of the data distribution module. The source node information can be source node encoding, which is just an example. The data caching module can cache the determined target node set (routing information) and the out-of-band signal. After the data distribution module completes the previous transmission task, it reads the next out-of-band signal to be transmitted from the data caching module and sends it to the signal processing unit in the group and / or sends it to the second routing information relay station 2 through the asynchronous processing unit, which is just an example. The data caching is set up in this application because the signal processing unit will perform asynchronous logic processing. Therefore, the data distribution unit needs to have a certain caching capacity to offset the processing delay caused by the cross-clock synchronization process and ensure that the system operates stably at the GHz level. The depth of the data caching module can be calculated based on the difference between the working clock of the first routing information relay station 1 and the clock of the node.

[0036] In one alternative implementation, when the data distribution unit determines that the target node set includes the fourth node in the first group of nodes, it processes the out-of-band signal accordingly based on the signal requirements of the fourth node and sends it to the fourth node.

[0037] In this embodiment of the application, when the data distribution unit determines that the target node set includes the fourth node in the first group of nodes, it can determine the specific node information of the fourth node and send the node information of the fourth node to the signal processing unit corresponding to the fourth node. At this time, the signal processing unit can use the receiving path to process and perform logical operations on the out-of-band signal sent to this node by the distribution unit, such as... Figure 3 As shown, the signal processing unit also includes a receiving-side handshake control and a logic operation unit. The receiving-side handshake control is used to inform the distribution unit that the out-of-band signal has been received. The logic operation unit can first restore the out-of-band signal to a signal type (level type or pulse type) related to the fourth node. Then, it can perform customized preset logic operation processing on the out-of-band signal according to the signal type and the signal requirements of the fourth node, such as simple "AND, OR, NOT" or setting up a complex state machine to generate the control information required by the fourth node. Finally, it can synchronize the generated control signal from the clock domain of the relay station to the clock domain of the fourth node. The order of processing is not limited. Clock domain synchronization processing can be performed first, followed by customized logic operation. This is just an example.

[0038] This application recovers the signal type through a logic operation unit and supports freely customizable logic operations. It supports differentiated processing of multiple types of out-of-band signals according to actual needs, thereby improving compatibility and flexibility.

[0039] In one alternative implementation, the distribution unit displays a routing configuration error message when it determines that the target node set contains the first node.

[0040] In this embodiment of the application, the corresponding bit of the local node must be 0 when initializing the routing table, because the state information of the source node does not need to be propagated to this node. That is, the bit on the diagonal of the routing table is forcibly set to 0, indicating that the routing path from the source node to itself is prohibited. In the subsequent process of the distribution unit querying the routing table to determine the target node set, it can be determined whether the bit of the source node in the target node set is 1, that is, the first node sends a signal to the first node. If the bit of the source node in the target node set is 1, it indicates that there may be a configuration error in the routing information. At this time, an interruption can be reported, indicating that the routing configuration error information is displayed.

[0041] This invention incorporates an error correction mechanism, which reports an interruption when the local node forwards to itself, thus avoiding redundant transmission caused by routing configuration errors.

[0042] In one optional implementation, the routing information relay station also includes a routing information management unit. The routing information management unit can be directly connected to the status enable information of each node. When a device is inserted, the corresponding node is enabled. When a device is unplugged, the enable of the corresponding node is masked, and the distribution unit will no longer send data to the node corresponding to that device. When a device is unplugged, if this application still broadcasts the information related to that node according to the original routing table, it will cause a certain amount of power waste, because at this time the node no longer has any connected devices and is in an disabled state. If it is still necessary to notify each routing information relay station through software, a large delay will occur, because the speed of firmware configuration is certainly not as fast as the speed of hardware self-adjustment. The routing information management unit responds to receiving the status enable signal sent by the third node in the first group of nodes. Based on the status enable signal, it changes the status information of the third node recorded in the routing table, so as to determine whether to distribute data to the third node according to the status information of the third node in the routing table.

[0043] The routing information management unit is responsible for managing the routing table, which can be dynamically updated based on node enable signals (such as hot-plugging) and the node status (enabled / disabled) can be adjusted in real time through a dedicated path. By responding to changes in the enable status of nodes in real time through hardware (such as hot-swapping) and combining the initial relationship of the routing table with firmware configuration, dynamic topology adjustment and low-latency signal transmission are achieved. Specifically, after receiving the enable signal sent by the third node in the first group of nodes, the routing information management unit can change the status information of the third node recorded in the routing table based on the enable signal. For example, the status of the third node in the routing table can be marked as disabled (0) or enabled (1), thereby achieving real-time response to topology changes. The routing information management unit continuously monitors the enable status of nodes and closes the transmission channels of nodes that have been powered off. While reducing power consumption, it ensures the sub-microsecond transmission delay of key signals. This method not only reduces the power consumption inside the chip, but also improves the transmission efficiency of out-of-band signals. In particular, the entry corresponding to the routing node in the routing table can still be maintained. Only the data sent by other nodes to this node is masked, because the device will not continue to send data, causing power consumption. It is also necessary to ensure that the node can still notify the associated nodes after it is powered on.

[0044] In one optional implementation, after receiving the out-of-band signal sent by the first routing information relay station 1, the distribution unit in the second routing information relay station 2 can query the routing table to determine the second node information belonging to the second group of nodes in the target node set. Then, the distribution unit can send the out-of-band signal to the signal processing unit corresponding to the second node. The signal processing unit restores the out-of-band information sent by the first routing information relay station 1 to an out-of-band signal of the type related to the second node; synchronizes the out-of-band signal to the clock domain corresponding to the second node; and performs customized preset logic operation processing on the processed out-of-band signal to generate the control information required by the second node, and sends the control information to the second node.

[0045] The signal processing unit of the second routing information relay station 2 in this embodiment restores the out-of-band signal to a signal type related to the second node in the clock domain. Then, it can perform customized preset logic operations on the out-of-band signal according to the signal type and the signal requirements of the second node, such as simple "AND, OR, NOT" or setting up a complex state machine to generate the control information required by the second node. Finally, the generated control signal can be synchronized from the clock domain of the relay station to the clock domain of the second node. This is just an example.

[0046] This application combines signal processing, arbitration, distribution, and inter-node routing relationship control to achieve out-of-band signal processing and transmission between multiple nodes on the chip. It has low logic processing latency, can operate at high frequencies, is friendly to back-end layout and routing, and reduces timing issues that may arise in high-frequency scenarios.

[0047] This invention also provides an out-of-band signal transmission method, which is applied to, for example... Figure 1 The first routing information relay station 1 and the second routing information relay station 2 shown are, as follows: Figure 6 As shown, the first routing information relay station 1 is used to execute steps S101 to S102, and the second routing information relay station 2 is used to execute step S201.

[0048] Step S101: In response to receiving an out-of-band signal sent by the first node in the first group of nodes, query the routing table corresponding to the first node to determine the target node set.

[0049] In step S102, when it is determined that the target node set contains nodes belonging to the second group of nodes, the out-of-band signal is sent to the second routing information relay station through the asynchronous processing unit.

[0050] Step S201: Receive the out-of-band signal sent by the first routing information relay station through the asynchronous processing unit, query the routing table corresponding to the first node, determine the second node information belonging to the second group of nodes in the target node set, and send the out-of-band signal to the second node.

[0051] In specific embodiments, such as Figure 7As shown, the first node can send an out-of-band signal transmission request to the first routing information relay station. The signal processing unit in the first routing information relay station can perform cross-clock domain processing and signal type conversion on the out-of-band signal, converting the out-of-band signal into a level-based out-of-band signal. After processing, the signal processing unit sends the out-of-band signal to the arbitration unit. The arbitration unit schedules the requests of each node based on a first-in-first-out strategy. After determining that the first node has obtained the arbitration right, it can send the out-of-band signal of the first node to the distribution unit. The distribution unit queries the routing table based on the source node information to determine the target node set, and determines whether there are any nodes in the target node set that cross the relay station, that is, whether there are any nodes in the target node set that belong to the second group of nodes. If there are nodes in the target node set that belong to the second group of nodes... Out-of-band signals can be sent to a second routing information relay station. The distribution unit of the second routing information relay station determines the specific second node, and then sends the out-of-band signal to the corresponding signal processing unit. The signal processing unit performs clock domain and type recovery processing on the out-of-band signal and performs logical operations, and finally sends the processed data to the second node. If there is a node belonging to the first group of nodes in the target node set, the out-of-band signal can be sent directly to the signal processing unit, so that the signal processing unit can send the data to the corresponding node after performing data recovery processing and logical operations. If the out-of-band signal is a node enable signal, the routing information management unit can change the routing status information of the node in the routing table based on the node enable signal. For details, please refer to the above embodiment, which will not be repeated here.

[0052] This application aims to address the resource waste inherent in traditional hard-wired methods for small-data-volume, high-time-sensitivity out-of-band signals when there are many nodes with unpredictable relationships. It supports flexible configuration of routing relationships between nodes through a combination of hardware and software. By dividing the layout on-chip (layout is the process of planning the specific locations and connecting wires (copper foil) of components in the circuit schematic according to electrical rules, mechanical dimensions, and heat dissipation requirements on the physical circuit board), and combining signal processing, arbitration, distribution, and routing relationship control between nodes, it enables out-of-band signal processing and transmission between multiple on-chip nodes. This results in low logic processing latency, operation at high frequencies, and is friendly to backend layout and routing, reducing timing issues that may arise in high-frequency scenarios. Simultaneously, dynamically adjusting routing relationships can reduce power consumption to some extent. It is suitable for on-chip out-of-band signal transmission scenarios with many nodes and unpredictable node relationships, such as multi-processor system chips, power management, security monitoring, memory management, and PCIe switch chips.

[0053] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0054] Embodiments of this application also provide an electronic device, such as... Figure 8 As shown, it includes a memory 10 and a processor 20. The memory 10 stores a computer program, and the processor 20 is configured to run the computer program to perform the steps in any of the out-of-band signal transmission method embodiments described above.

[0055] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the out-of-band signal transmission method embodiments described above when running.

[0056] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0057] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described out-of-band signal transmission method embodiments.

[0058] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in any of the out-of-band signal transmission method embodiments described above.

[0059] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0060] The out-of-band signal transmission method and system provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A method for out-of-band signal transmission, characterized in that, A first routing information relay station is applied in a chip, the chip containing at least two sets of nodes. The first routing information relay station manages the routing information of a first set of nodes. The first routing information relay station is connected to a second routing information relay station via an asynchronous processing unit. The second routing information relay station manages the routing information of a second set of nodes. The method includes: In response to receiving an out-of-band signal sent by the first node in the first group of nodes, the routing table corresponding to the first node is queried to determine the target node set; When it is determined that the target node set contains nodes belonging to the second group of nodes, the out-of-band signal is sent to the second routing information relay station through the asynchronous processing unit, so that the second routing relay station queries the routing table corresponding to the first node, determines the second node information belonging to the second group of nodes in the target node set, and sends the out-of-band signal to the second node.

2. The method according to claim 1, characterized in that, Before querying the routing table corresponding to the first node, the method further includes: The out-of-band signal is synchronized across clock domains. The out-of-band signal after synchronous processing is converted into an out-of-band signal of a preset transmission standard type.

3. The method according to claim 1 or 2, characterized in that, Before querying the routing table corresponding to the first node, the method further includes: Add the request representing the first node to the sorting queue; When the sorting queue schedules a request to the first node, the step of querying the routing table corresponding to the first node is executed.

4. The method according to claim 3, characterized in that, Adding the request representing the first node to the sorting queue includes: Increment the counter value of the current group by one to get the latest counter value; The latest recorded value of the counter is used as the allocation number representing the request of the first node; Based on the allocation number, the request representing the first node is added to the sorting queue; After the latest recorded value of the counter is the maximum value of the counter, the counter is reset, a new group is created, and the process returns to the step of incrementing the counter record value of the current group by one.

5. The method according to claim 4, characterized in that, The method further includes: When the first node includes multiple nodes from the first group of nodes, the multiple nodes are added to a sorting queue according to a preset priority, so that when the allocation number is polled, each node is scheduled in sequence according to the preset priority.

6. The method according to claim 1, characterized in that, The method further includes: In response to receiving a status enable signal from the third node in the first group of nodes; Based on the state enable signal, the state information of the third node recorded in the routing table is changed, so as to determine whether to distribute data to the third node according to the state information of the third node in the routing table.

7. The method according to claim 1, characterized in that, The method further includes: When it is determined that the target node set includes the fourth node in the first group of nodes, the out-of-band signal is processed accordingly based on the signal requirements of the fourth node and sent to the fourth node. And / or, when it is determined that the first node is included in the target node set, a routing configuration error message is displayed.

8. An out-of-band signal transmission method, characterized in that, A second routing information relay station is applied in a chip, the chip containing at least two sets of nodes. A first routing information relay station manages the routing information of a first set of nodes. The first routing information relay station is connected to the second routing information relay station via an asynchronous processing unit. The second routing information relay station manages the routing information of a second set of nodes. The method includes: The system receives out-of-band signals sent by the relay station through the asynchronous processing unit, queries the routing table corresponding to the first node, and determines the information of the second node belonging to the second group of nodes in the target node set. The out-of-band signal is sent to the second node.

9. The method according to claim 8, characterized in that, Sending the out-of-band signal to the second node includes: Restore the out-of-band information sent by the first routing information relay station to an out-of-band signal of the type related to the second node; The out-of-band signal is synchronized to the clock domain corresponding to the second node, and the processed out-of-band signal is subjected to customized preset logic operations to generate the control information required by the second node, and the control information is sent to the second node.

10. An out-of-band signal transmission system, characterized in that, The system includes a first routing information relay station and a second routing information relay station in a chip. The chip contains at least two sets of nodes. The first routing information relay station manages the routing information of a first set of nodes. The first routing information relay station is connected to the second routing information relay station via an asynchronous processing unit. The second routing information relay station manages the routing information of a second set of nodes. The first routing information relay station responds by receiving an out-of-band signal sent by the first node in the first group of nodes, queries the routing table corresponding to the first node, and determines the target node set; When the first routing information relay station determines that the target node set contains nodes belonging to the second group of nodes, it sends the out-of-band signal to the second routing information relay station through the asynchronous processing unit. The second routing information relay station receives the out-of-band signal sent by the first routing information relay station through the asynchronous processing unit, queries the routing table corresponding to the first node, determines the second node information belonging to the second group of nodes in the target node set, and sends the out-of-band signal to the second node.

Images