Virtual channel setup method, device, electronic equipment and medium for on-chip network
By grouping the routers on the chip and building a minimum spanning tree, and adding dedicated virtual channels, the hardware overhead problem caused by routing deadlock in on-chip networks is solved, and more efficient data transmission is achieved.
Patent Information
- Application Number
- CN202611105289.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-23
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies for resolving routing deadlock in on-chip networks suffer from significant hardware overhead, including wasted hardware resources due to global constraint schemes.
Multiple routers on the core are grouped to construct a minimum spanning tree, and dedicated virtual channels are added to the minimum spanning tree to reduce the number and length of dedicated virtual channels and reduce hardware overhead.
By using router grouping and minimum spanning tree methods, the number and length of dedicated virtual channels are reduced, hardware overhead is lowered, and data transmission efficiency and performance are improved.
Smart Images

Figure CN123123255A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of semiconductor chip technology, and in particular to a method, apparatus, electronic device and medium for setting up virtual channels in an on-chip network. Background Technology
[0002] Chips are the core components of advanced chip architectures. Multiple chips are interconnected through packaging technologies such as silicon interposers and TSVs (Through Silicon Vias) to form a system-on-chip (SoC). Each chip has at least one router, and the routers of all chips form an on-chip network for transmitting data within or between chips. During data transmission, when a group of data packets occupies the transmission resources needed for the next step of another packet, a closed loop of waiting occurs, leading to routing deadlock.
[0003] In existing technologies, solutions to routing deadlock problems typically involve adding global constraints to the on-chip network to prevent infinite loops. These global constraints include: adding dedicated virtual channels to all routers, adding turn restrictions to all routers, and adding flow control to all nodes.
[0004] However, the above solution has the problem of high hardware overhead. Summary of the Invention
[0005] In view of the above problems, embodiments of this application are proposed to provide a method for setting up a virtual channel in a network-on-chip (NIC) to overcome or at least partially solve the above problems, thereby reducing the hardware overhead of setting up a virtual channel in a NIC.
[0006] Accordingly, embodiments of this application also provide a virtual channel setting device for a network-on-a-chip, an electronic device, and a storage medium to ensure the implementation and application of the above methods.
[0007] To address the aforementioned problems, this application discloses a method for setting up a virtual channel in a network-on-chip, the method comprising: Group the multiple routers on the core to obtain at least one router group; each router group includes: a border router and at least one internal router; construct a minimum spanning tree for each of the at least one router group; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the paths from the border router to the internal router, or the paths between internal routers; add dedicated virtual channels to the edges in the minimum spanning tree, and the dedicated virtual channels are used for data transmission between cores.
[0008] Optionally, the multiple routers on the chip are grouped to obtain at least one router group, including: Multiple border router groups are determined based on the multiple routers on the core, and each border router group includes at least one border router. For each border router group, the sum of the distances between each internal router on the core and the nearest border router of the internal router in the border router group is calculated as the path length of the border router group. Each internal router on the core includes routers outside the border router group. Each border router in the border router group with the smallest path length is designated as a border router in a router group, and the internal routers are assigned to the router group containing the target nearest border router of the internal router. The target nearest border router is the nearest border router in the border router group with the smallest path length.
[0009] Optionally, the router may be assigned to the router group containing the nearest border router, including: If there is one target nearest border router on an internal router, the internal router is assigned to the internal router group where the target nearest border router is located. After the assignment of each internal router with one target nearest border router on the core is completed, for each internal router with multiple target nearest border routers on the core, the number of internal routers already assigned in the router group where the multiple target nearest border routers are located is determined, and the internal router with multiple target nearest border routers is assigned to the router group with the smallest number of assigned internal routers.
[0010] Optionally, multiple border router groups are determined based on multiple routers on the chip, including: Obtain information about other chips on the system-on-a-chip (SoC) that communicate with the chip. This information includes the number and / or size of the other chips. Determine the number of border routers based on the chip size and the information about the other chips. Combine all routers corresponding to the number of border routers among the multiple routers of the chip to form multiple border router groups.
[0011] Optionally, dedicated virtual channels are added to the edges in the minimum spanning tree, including: Determine the layout orientation of multiple chips on the system-on-a-chip; if the routing directions supported by the chip include the layout orientation, select the edge consistent with the layout orientation from the minimum spanning tree as the target edge; add a dedicated virtual channel for bidirectional data transmission on the target edge.
[0012] Optionally, the layout orientation of multiple dies on the system-on-a-chip is determined, including: Identify the remaining cores that have data communication with the core, and determine the layout orientation between the remaining cores and the cores.
[0013] Optionally, dedicated virtual channels are added to the edges in the minimum spanning tree, including: Add a dedicated virtual channel to each edge of the minimum spanning tree; construct the channel dependency graph of the core particle according to the local routing algorithm of the core particle. The nodes in the channel dependency graph include dedicated virtual channels and general virtual channels of the core particle. The edges in the channel dependency graph represent the dependency relationship between nodes; for any dedicated virtual channel, if there is no cycle in the channel dependency graph after deleting the dedicated virtual channel, then delete the dedicated virtual channel.
[0014] Optionally, it also includes: The routing configuration information of the core is sent to each router in the core. The routing configuration information includes: the type of each router on the core, information of at least one router group corresponding to the border router group with the shortest path length, and information of the private virtual channel.
[0015] This application also discloses a virtual channel setting device for an on-chip network, including: The packet module is used to group multiple routers on the chip to obtain at least one router group; each router group includes: a border router and at least one internal router.
[0016] The building module is used to construct minimum spanning trees for at least one router group; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the paths from the border router to the internal router, or the paths between internal routers.
[0017] Add a module to add dedicated virtual channels on the edges in the minimum spanning tree. These dedicated virtual channels are used for data transmission between cores.
[0018] Optionally, the grouping module is also used for: Multiple border router groups are determined based on the multiple routers on the core, and each border router group includes at least one border router. For each border router group, the sum of the distances between each internal router on the core and the nearest border router of the internal router in the border router group is calculated as the path length of the border router group. Each internal router on the core includes routers outside the border router group. Each border router in the border router group with the smallest path length is designated as a border router in a router group, and the internal routers are assigned to the router group containing the target nearest border router of the internal router. The target nearest border router is the nearest border router in the border router group with the smallest path length.
[0019] Optionally, the grouping module is also used for: If there is one target nearest border router on an internal router, the internal router is assigned to the internal router group where the target nearest border router is located. After the assignment of each internal router with one target nearest border router on the core is completed, for each internal router with multiple target nearest border routers on the core, the number of internal routers already assigned in the router group where the multiple target nearest border routers are located is determined, and the internal router with multiple target nearest border routers is assigned to the router group with the smallest number of assigned internal routers.
[0020] Optionally, the grouping module is also used for: Obtain information about other chips on the system-on-a-chip (SoC) that communicate with the chip. This information includes the number and / or size of the other chips. Determine the number of border routers based on the chip size and the information about the other chips. Combine all routers corresponding to the number of border routers among the multiple routers of the chip to form multiple border router groups.
[0021] Optionally, adding modules is also used for: Determine the layout orientation of multiple chips on the system-on-a-chip; if the routing directions supported by the chip include the layout orientation, select the edge consistent with the layout orientation from the minimum spanning tree as the target edge; add a dedicated virtual channel for bidirectional data transmission on the target edge.
[0022] Optionally, adding modules is also used for: Identify the remaining cores that have data communication with the core, and determine the layout orientation between the remaining cores and the cores.
[0023] Optionally, adding modules is also used for: Add a dedicated virtual channel to each edge of the minimum spanning tree; construct the channel dependency graph of the core particle according to the local routing algorithm of the core particle. The nodes in the channel dependency graph include dedicated virtual channels and general virtual channels of the core particle. The edges in the channel dependency graph represent the dependency relationship between nodes; for any dedicated virtual channel, if there is no cycle in the channel dependency graph after deleting the dedicated virtual channel, then delete the dedicated virtual channel.
[0024] Optionally, it also includes: The distribution module is used to send the routing configuration information of the core to each router in the core. The routing configuration information includes: the type of each router on the core, information of at least one router group corresponding to the border router group with the shortest path length, and information of the private virtual channel.
[0025] This application also discloses an electronic device, including a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors. One or more programs include instructions for performing the following operations: Group the multiple routers on the core to obtain at least one router group; each router group includes: a border router and at least one internal router; construct a minimum spanning tree for each of the at least one router group; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the paths from the border router to the internal router, or the paths between internal routers; add dedicated virtual channels to the edges in the minimum spanning tree, and the dedicated virtual channels are used for data transmission between cores.
[0026] This application also discloses a readable storage medium, which, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to execute one or more on-chip network virtual channel setting methods of this application.
[0027] This application also discloses a computer program product, including a computer program for implementing one or more on-chip network virtual channel setting methods in this application embodiment.
[0028] The embodiments of this application have the following advantages: Dedicated virtual channels are added based on router groups and minimum spanning trees to minimize the number and length of dedicated virtual channels and reduce hardware overhead. Attached Figure Description
[0029] Figure 1 This is a schematic diagram of the structure of a SOC provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the steps of an embodiment of a virtual channel setting method for an on-chip network according to this application; Figure 3 This is a schematic diagram of a border router group and a router group provided in an embodiment of this application; Figures 4 to 6 This is a schematic diagram of the three spanning tree structures of router group RG1 provided in the embodiments of this application; Figure 7 This is a schematic diagram of the minimum spanning tree structure of router groups RG1 to RG4 provided in the embodiments of this application; Figure 8 and Figure 9 This is a schematic diagram of two dedicated virtual channels for router groups RG1 to RG4 provided in the embodiments of this application; Figure 10This is a structural block diagram of an embodiment of a virtual channel setting device for an on-chip network according to this application; Figure 11 This is a structural block diagram of an electronic device according to an exemplary embodiment. Detailed Implementation
[0030] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0031] This application applies to a chip system, which includes at least one chip and an inter-chip interconnect network. The at least one chip may include at least one type of chip, and the chips are connected to each other via the inter-chip interconnect network. The inter-chip interconnect network is an intermediate layer switching structure or a UCIe (Universal Chiplet Interconnect Express) switching chip, and typically includes at least one interconnected switching node.
[0032] Each of the aforementioned chip chips can be equipped with an internal Network On-Chip (NOC) and at least one border router. The NOC comprises multiple interconnected internal routers, and the border router connects the internal routers to the aforementioned switching nodes, thereby enabling interconnection between chip chips. The border router is used to forward data sent by the internal routers to the switching nodes, or vice versa.
[0033] Figure 1 This is a schematic diagram of the structure of a SOC provided in an embodiment of this application. (Refer to...) Figure 1 As shown, four types of chips are exemplarily provided, wherein chips 101 and 102 are computing chips, chip 103 is a low-power computing chip, chip 104 is an accelerator chip, and chip 105 is an I / O (Input / Output) interface and memory interface chip. Each chip 101 to 105 is equipped with an internal on-chip network 125, which consists of multiple interconnected internal routers 126. The internal on-chip network 125 is connected to an inter-chip interconnection network 128 via a boundary router 127 on the chip. The inter-chip interconnection network 128 includes four interconnected switching nodes 129 to 132.
[0034] The core 101 includes cores 106 to 109, each connected to an internal router 126. Cores 106 to 109 are responsible for either processing data sent by their respective internal routers 126 or sending the processed data back to their corresponding internal routers 126. The core 101 also has two border routers 127, each connected to one of the two internal routers 126 within the core 101. Each border router 127 is also connected to a switching node 129.
[0035] The core 102 includes cores 110 to 113, each connected to an internal router 126. Cores 110 to 113 are responsible for either processing data sent by their respective internal routers 126 or sending the processed data back to their corresponding internal routers 126. Two border routers 127 are configured on the core 102. Each border router 127 is connected to one of the two internal routers 126 within the core 102. Each border router 127 is also connected to a switching node 130.
[0036] The core 103 includes LLC (Last Level Cache) fragments 114 to 117, each connected to an internal router 126. Each LLC fragment 114 to 117 is responsible for either caching data sent by its corresponding internal router 126 or sending cached data to the internal router 126. The core 103 also has two border routers 127, each connected to one of the two internal routers 126 within the core 103. Each border router 127 is also connected to a switching node 131.
[0037] The core 104 includes accelerators 118 to 121, each connected to an internal router 126. Accelerators 118 to 121 are responsible for accelerating data sent by their respective internal routers 126, or for sending the accelerated data to the internal routers 126. Two border routers 127 are configured on the core 104. Each border router 127 is connected to one of the two internal routers 126 within the core 104. Each border router 127 is also connected to a switching node 132.
[0038] The chip 105 includes a DDR (Double Data Rate Synchronous Dynamic Random Access Memory) controller 122, a PCIe (Peripheral Component Interconnect Express) / CXL (Compute Express Link) interface 123, and other I / O (Input / Output) interfaces 124. The DDR controller 123, PCIe / CXL interface 123, and other I / O interfaces 123 are each connected to an internal router 126. The DDR controller 123 is used to write data sent by the corresponding internal router 126 into the DDR memory, or to send data from the DDR memory to the corresponding internal router 126. The PCIe / CXL interface 123 is used to send data sent by the corresponding internal router 126 outwards, or to send externally received data to the corresponding internal router 126. The other I / O interfaces 123 are used to output data sent by the corresponding internal router 126, or to send externally received data to the corresponding internal router 126.
[0039] The data interaction process between the aforementioned core particles may include: Data is transmitted within the sending chip's internal on-chip network to the sending chip's border router. The sending chip's border router forwards the data to connected switching nodes. These switching nodes then forward the data via at least one intermediate switching node to a switching node connected to the receiving chip's border router. The switching node connected to the receiving chip's border router forwards the data to the receiving chip's border router. The receiving chip's border router then transmits the data via the receiving chip's on-chip network to the functional units on the receiving chip. The functional units of the receiving chip are related to the type of receiving chip and are used to process the received data. For example, the functional unit of a computing chip is a core, the functional unit of a low-density computing chip is an LLC fragment, the functional unit of an accelerator chip is an accelerator, and the functional units of an I / O interface and memory interface chip are a DDR controller, a PCIe / CXL interface, and other I / O interfaces.
[0040] For example, the caching process of the data calculated by the core of core 101 may include the following steps: the core of core 101 sends the calculated data to the border router 127 of core 101 through the internal router of core 101. The border router 127 of core 101 can send the data to the switching node 129. The switching node 129 forwards the data to the switching node 131. The switching node 131 sends the data to the border router 127 of core 103. The border router 127 of core 103 sends the data to the internal router of core 103, which then forwards it to the LLC fragment 114.
[0041] It should be noted that, Figure 1 The type of chip, the number of internal routers 126, and the number of switching nodes in the inter-chip interconnection network are merely examples. This application embodiment does not limit the type of chip, the number of internal routers included in the chip, or the number of switching nodes included in the inter-chip interconnection network.
[0042] To address the routing deadlock issue in SOCs, technologies such as virtual channel isolation, deadlock-free fault-tolerant routing, modular turn restrictions, and remote flow control can be employed.
[0043] The aforementioned virtual channel isolation technology assigns different data packets to different virtual channels to achieve packet isolation and avoid deadlock. Common virtual channel isolation technologies use at least two virtual channels: one for transmitting data packets within and entering the core, and another for transmitting data packets leaving the core. This approach requires setting up two virtual channels, resulting in significant hardware overhead.
[0044] The aforementioned deadlock-free fault-tolerant routing technique employs two not entirely isolated virtual networks, VN0 and VN1. Data packet transmission within these virtual networks follows these rules: First, data packets cannot enter VN0 from VN1; second, data packets on VN0 cannot travel from the intermediate layer to the core layer; third, data packets on VN1 cannot be transmitted from the core layer to the intermediate layer. This scheme involves two virtual networks and also suffers from significant hardware overhead.
[0045] Modular turning restriction technology abstracts the parts of the core system other than the designed core into a virtual node, and then introduces turning restrictions to the border routers under the new network topology. A deadlock-free routing design is then performed through a traversal algorithm. This approach suffers from high algorithmic complexity, and the turning restrictions lead to poor data transmission performance.
[0046] Remote flow control technology implements a flow control mechanism by adding hardware modules such as flow control buffers and flow control communication channels to ensure that incoming data packets do not cause deadlock in the core system. This approach requires modifying the router's microarchitecture and adding control signal communication channels, which also increases hardware overhead.
[0047] To address the aforementioned issue of high hardware overhead, this application's embodiments group multiple routers on the core chip and construct a minimum spanning tree for each router group. Based on the minimum spanning tree, dedicated virtual channels are added at certain locations. This reduces the number of dedicated virtual channels, lowers hardware overhead, and avoids redundant dedicated virtual channels. Therefore, compared to virtual channel isolation technology and deadlock-free fault-tolerant routing technology, this application achieves higher virtual channel utilization. Furthermore, the algorithm complexity of this application is low, and it does not require turn restrictions, thus improving data transmission performance.
[0048] The following example illustrates an implementation of a virtual channel configuration method for a network-on-a-chip.
[0049] Reference Figure 2 As shown, the virtual channel configuration method for the above-mentioned on-chip network may specifically include the following steps: 201. Group the multiple routers on the chip to obtain at least one router group; each router group includes: a border router and at least one internal router.
[0050] 202. Construct minimum spanning trees for at least one router group; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the paths from the border router to the internal router, or the paths between internal routers.
[0051] 203. Add dedicated virtual channels to the edges in the minimum spanning tree. These dedicated virtual channels are used for data transmission between cores.
[0052] Each router group consists of border routers and internal routers, both of which are routers on a single core. Border routers are used for data transmission between cores, while internal routers are used for data transmission within a single core. The process of determining a router group can be understood as dividing the multiple routers on a single core into blocks.
[0053] In one possible implementation, 201 can be achieved through the following steps: First, the area where all routers of the chip are located is divided into multiple areas, and the routers included in each area constitute a router group. The number of routers included in each router group is a preset packet size, or close to the preset packet size. Then, the routers located at the edge of each area are selected as border routers, and the remaining routers in the area are the internal routers of the corresponding router group.
[0054] In another possible implementation, step 201 includes: 2011. Multiple border router groups are determined based on multiple routers on the chip, and each border router group includes at least one border router.
[0055] 2012. For each border router group, calculate the sum of the distances between each internal router on the core and the nearest border router in the border router group, and use this as the path length of the border router group. Each internal router on the core includes routers outside the border router group.
[0056] 2013. Each border router in the border router group with the shortest path length is treated as a border router in a router group, and the internal routers are assigned to the router group where the target nearest border router is located. The target nearest border router is the nearest border router in the border router group with the shortest path length.
[0057] As can be seen, the steps described above from 2011 to 2013 reduced line length in two ways: firstly, router groups were divided based on the border router group with the shortest path length to reduce the overall routing line length of the core; secondly, internal routers were assigned to the router group containing the nearest border router to minimize the data transmission path for that internal router. These two aspects improve data transmission speed and performance.
[0058] In one embodiment, step 2011 includes randomly selecting at least one router from a plurality of routers in the core to form at least one border router in a border router group. Multiple random selections can thus yield multiple border router groups, with each group containing at least one different border router. For example, if the plurality of routers in the core are R1 to R100, a first random selection of 10 routers R1, R4, R20, R11, R45, R28, R19, R47, R23, and R89 forms one border router group; a second random selection of 10 routers R27, R4, R63, R89, R12, R51, R76, R33, R95, and R18 forms another border router group.
[0059] In another embodiment, the implementation process of step 1011 includes: combining all the routers of the core in a permutation and combination manner to obtain all possible combinations, and each possible combination is a border router group.
[0060] In both of the above embodiments, the number of border routers included in each border router group is not limited, so as to generate border router groups with multiple border routers. Multiple border router groups are generated for each number of border routers.
[0061] In another embodiment, the number of border routers included in each border router group is predetermined. The number of border routers can be set randomly or dynamically. In the scheme of dynamically setting the number of border routers, step 2011 includes: Step 20111: Obtain information about the remaining chips on the system-on-a-chip that communicate with the chip. The information about the remaining chips includes: the number of the remaining chips and / or the size of the remaining chips.
[0062] Step 20112: Determine the number of border routers based on the size of the core chip and the information of the remaining core chips.
[0063] Step 20113: Combine all routers corresponding to the number of border routers in the multiple routers of the core into multiple border router groups.
[0064] Through the above steps 20111 to 20113, the number of border routers can be dynamically adapted, improving the rationality of the number of border routers. On the one hand, it can avoid the low data transmission speed and congestion caused by too few border routers, and on the other hand, it can avoid the large hardware overhead caused by too many border routers.
[0065] The number of border routers is directly related to the size of the core chip and the number and size of the remaining core chips. Given a fixed number and size of the remaining core chips, the number of border routers increases with increasing core chip size; similarly, given a fixed core chip size and number of remaining core chips, the number of border routers increases with increasing size of the remaining core chips; and so on. This allows for dynamic adjustment of the size of the border router group, and consequently, the number of router groups, further reducing the probability of data transmission congestion and improving data transmission speed and performance.
[0066] Based on the aforementioned number of border routers, the process of determining each border router group includes: combining all the routers in the core according to a permutation and combination method to obtain all possible combinations corresponding to the number of border routers, with each possible combination constituting a border router group, and each border router group including the aforementioned number of border routers.
[0067] For any border router group, when calculating the path length of the border router group in step 2012, it is necessary to calculate the distance between each internal router on the core and the nearest border router of the internal router in the border router group. The nearest border router of each internal router in a border router group is the border router in the border router group that is closest to the internal router.
[0068] The path length of the aforementioned border router group can be calculated using the following formula:
[0069] Here, BR stands for Border Router Group, which corresponds to a border router layout. IR includes all internal routers outside the border router group BR, and r is any internal router in IR. It is the distance between internal router r and the nearest border router in the border router group BR. This is the path length of the Border Router Group (BR).
[0070] For example, Figure 3 This is a schematic diagram of a border router group and a router group provided in an embodiment of this application. (Refer to...) Figure 3 As shown, routers R2, R8, R9, and R15 form a border router group. The internal routers include: R1, R3, R4, R5, R6, R7, R10, R11, R12, R13, and R14. The nearest border router for internal routers R1, R3, and R6 is border router R2; for internal routers R5, R10, and R13, it is border router R9; for internal routers R4, R7, and R12, it is border router R8; and for internal routers R11, R14, and R16, it is border router R15. Therefore, the path length for the corresponding border router group [R2, R8, R9, R15] is:
[0071] The path lengths corresponding to the multiple border router groups of the kernel are calculated according to the above process, for example... Figure 3The boundary router group [R2, R8, R9, R15] has the shortest path length. Therefore, boundary routers R2, R8, R9, and R15 are designated as boundary routers in the four router groups. After determining the boundary routers for each router group, the internal routers on the core can be assigned to the router group containing the nearest boundary router in the boundary router group [R2, R8, R9, R15]. Specifically, internal routers R1, R3, and R6 are assigned to router group RG1 containing boundary router group R2; internal routers R5, R10, and R13 are assigned to router group RG2 containing boundary router group R9; internal routers R4, R7, and R12 are assigned to router group RG3 containing boundary router group R8; and internal routers R11, R14, and R16 are assigned to router group RG4 containing boundary router group R15.
[0072] In some possible implementations, step 2013 above includes: Step 20131: If there is a target nearest border router for the internal router, assign the internal router to the internal router group where the target nearest border router of the internal router is located.
[0073] Step 20132: After the internal routers with one target nearest border router on the core are partitioned, for each internal router with multiple target nearest border routers on the core, determine the number of internal routers already partitioned in the router group to which the multiple target nearest border routers of the internal router belong, and partition the internal router with multiple target nearest border routers into the router group with the smallest number of partitioned internal routers.
[0074] Steps 20131 to 20132 above can minimize the path length while ensuring that the number of internal routers in all router groups is as close as possible or even the same, thus ensuring the transmission balance of different boundary routers and helping to improve data transmission speed.
[0075] For example, if the border router group with the shortest path length includes R2, R8, R9, and R15, and there are internal routers R17 and R18, with distances of D1, D2, D1, and D3 between internal router R17 and R2, R8, R9, and R15 respectively, and distances of D4, D2, D4, and D3 between internal router R18 and R2, R8, R9, and R15 respectively, where D1 is less than D2 and D3, and D4 is less than D2 and D3, then internal router R17 has two nearest border routers R2 and R9, and internal router R18 has two nearest border routers R2 and R9. In this case, internal routers R1, R3, R4, R5, R6, R7, R10, R11, R12, R13, and R14, which each have a target nearest border router, are first assigned to the router group containing the corresponding target nearest border router. Next, the number of internal routers already assigned in the router groups of border routers R2 and R9 is obtained. Since the number of internal routers already assigned in router group RG1 of border router R2 and router group RG2 of border router R9 is both 3, internal router R17 is randomly assigned to router group RG1 of border router R2, and the number of internal routers already assigned in router group RG1 is 4. At this time, the number of internal routers already assigned in router group RG1 of border router R2 (4) is greater than the number of internal routers already assigned in router group RG2 of border router R9 (3), therefore internal router R18 is assigned to router group RG2 of border router R9.
[0076] After performing step 201 and its sub-steps, step 202 is performed to generate a minimum spanning tree for each router group. The minimum spanning tree is a spanning tree rooted at the border router, covering all routers in the router group, free of loops, and with the minimum total weight of all edges. The output traffic of each node in the spanning tree is the sum of the input traffic of all its child nodes, and the weight of each edge is directly proportional to its length. The length of an edge is the distance between the two nodes it connects.
[0077] Reference Figure 3 As shown, router groups RG1 to RG4 each need to generate a minimum spanning tree. The following example... Figure 3 Taking router group RG1 as an example, this explains the process of calculating the total weight of the edges of several spanning trees.
[0078] Figure 4 This is a schematic diagram of the spanning tree structure of router group RG1 provided in an embodiment of this application. (Refer to...) Figure 4As shown, the root node is R2, and the child nodes of the root node R2 are R1 and R6. The child node of R1 is R3. The distance between R2 and R1 is 1, the distance between R2 and R6 is 1, and the distance between R1 and R3 is 2. Therefore, the edge...<R2,R1> and edge<R2,R6> The weights of all edges are 1.<R1,R3> The weight is 2. Figure 4 The total weight of all edges in the spanning tree shown is 4.
[0079] Figure 5 This is a schematic diagram of another spanning tree structure for router group RG1 provided in an embodiment of this application. (Refer to...) Figure 5 As shown, the root node is R2, the child node of the root node R2 is R1, and the child nodes of R1 are R3 and R6. The distance between R2 and R1 is 1, the distance between R1 and R3 is 2, and the distance between R1 and R6 is 1.5. Therefore, the edge...<R2,R1> and edge<R1,R3> ,<R1,R6> The weights are 1, 2, and 1.5 respectively. Figure 5 The total weight of all edges in the spanning tree shown is 4.5.
[0080] Figure 6 This is a schematic diagram of another spanning tree structure for router group RG1 provided in an embodiment of this application. (Refer to...) Figure 6 As shown, the root node is R2, the child node of the root node R2 is R1, the child node of R1 is R6, and the child node of R6 is R3. The distance between R2 and R1 is 1, the distance between R1 and R3 is 2, and the distance between R3 and R6 is 1.5.<R2,R1> and edge<R1,R3> ,<R3,R6> The weights are 1, 2, and 1.5 respectively. Figure 6 The total weight of all edges in the spanning tree shown is 4.5.
[0081] Reference Figure 7 As shown, in the spanning tree of router group RG1, the root node R2 corresponds to three child nodes R1, R3, and R6. The distances between R2 and R1, R2 and R3, and R2 and R6 are all 1. Therefore, the edges...<R2,R1> and edge<R2,R3> ,side<R2,R6> The weights are all 1. Figure 7 The total weight of all edges in the spanning tree shown is 3.
[0082] Therefore, the minimum spanning tree of router group RG1 is Figure 7 The spanning tree is shown below. Similarly, the minimum spanning tree for router groups RG2 to RG4 can be obtained as follows: Figure 7As shown. The minimum spanning tree of router group RG2 includes the root node R9 and its three child nodes R5, R10, and R13. The minimum spanning tree of router group RG3 includes the root node R8 and its three child nodes R4, R7, and R12. The minimum spanning tree of router group RG4 includes the root node R15 and its three child nodes R11, R14, and R16.
[0083] After obtaining the minimum spanning tree for each router group through step 202, step 203 can be performed to add dedicated virtual channels to the edges of the minimum spanning tree.
[0084] In some possible implementations, step 203 includes: Step 2031: Determine the layout orientation of multiple chips on the system-on-a-chip.
[0085] Step 2032: If the routing directions supported by the core include the layout direction, then select the edge that is consistent with the layout direction from the minimum spanning tree as the target edge.
[0086] Step 2033: Add a dedicated virtual channel for bidirectional data transmission on the target edge.
[0087] By using steps 2031 to 2033, the added dedicated virtual channel is consistent with the layout direction and routing direction. As long as there is only one change of direction in the routing path, routing success and data transmission success can be guaranteed.
[0088] Here, the layout direction refers to the arrangement direction of multiple core particles, such as... Figure 1 As shown, the layout direction of cores 101 and 102 is X, and the layout direction of cores 101 and 103 is Y. When the routing directions supported by the cores include both the X and Y directions, it can be... Figure 7 The edge in the X or Y direction is taken as the target edge. Figure 7 The example shown illustrates a target edge in the X direction: edge<R2,R1> ,side<R2,R3> ,side<R8,R7> ,side<R9,R10> ,side<R15,R14> and edge<R15,R16> Therefore, a dedicated virtual channel was added, such as Figure 8 The bold lines in the X direction are shown. Alternatively, as shown... Figure 9 Add a thicker line in the Y direction as shown.
[0089] Furthermore, when both the layout direction and the supported routing directions of the core are multiple, the number of edges in that direction in the minimum spanning tree of each core can be determined, and the edge in the direction with the fewest edges can be added as the target edge to a dedicated virtual channel. For example, if the layout direction and the supported routing directions of the core include directions X and Y, and the number of edges in direction X in the minimum spanning tree of each core is less than the number of edges in direction Y in the minimum spanning tree of each core, then direction X can be used as the target edge to add a dedicated virtual channel.
[0090] If the system-on-a-chip (SoC) only includes chip 101 and chip 102, and the layout direction is X-axis, then when the routing directions supported by the chip include both X and Y directions, only the edge in the X direction can be selected as the target edge, such as... Figure 8 Add a thicker line in the X direction as shown.
[0091] If the system-on-a-chip (SoC) only includes chip 101 and chip 103, and the layout direction is Y, then when the routing directions supported by the chip include both X and Y directions, only the edge in the Y direction can be selected as the target edge, such as... Figure 9 Add a thicker line in the Y direction as shown.
[0092] In some possible implementations, step 2031 includes: Step 20311: Identify the remaining cores that have data communication with the core, and determine the layout orientation between the remaining cores and the cores.
[0093] In step 20311, the system-on-a-chip (SoC) is divided into regions according to data communication relationships. Each region includes at least one chip with data communication capabilities. Dedicated virtual channels can be added to the chips within each region based on the layout orientation of that region. This allows for more accurate addition of dedicated virtual channels when the SoC has multiple layout orientations.
[0094] In some possible implementations, step 203 includes: Step 2034: Add a dedicated virtual channel to each edge of the minimum spanning tree.
[0095] Step 2035: Construct the channel dependency graph of the core particle according to the local routing algorithm of the core particle. The nodes in the channel dependency graph include dedicated virtual channels and general virtual channels of the core particle. The edges in the channel dependency graph represent the dependency relationship between nodes. Step 2036: For any dedicated virtual channel, if there are no loops in the channel dependency graph after deleting the dedicated virtual channel, then delete the dedicated virtual channel.
[0096] By performing steps 2034 to 2036 above, dedicated virtual channels are deleted under the condition that no loops exist in the local routing algorithm. In this way, redundant dedicated virtual channels can be eliminated as much as possible, reducing the number of dedicated virtual channels and lowering hardware overhead.
[0097] In this system, general virtual channels are existing virtual channels on the core, while dedicated virtual channels are newly added virtual channels in addition to general virtual channels. When there are no loops in the channel dependency graph, it indicates no deadlock; when there are loops, it indicates a deadlock. If deleting a dedicated virtual channel does not result in any loops in the channel dependency graph, then that dedicated virtual channel is redundant and can be deleted. If deleting a dedicated virtual channel results in a loop in the channel dependency graph, then that dedicated virtual channel is critical to resolving the deadlock and cannot be deleted. Therefore, by deleting all redundant dedicated virtual channels, a channel dependency graph with minimal channels and no loops can be obtained.
[0098] In some possible implementations, after step 203, the above method further includes: Step 204: Send the routing configuration information of the core to each router in the core. The routing configuration information includes: the type of each router on the core, information of at least one router group corresponding to the border router group with the shortest path length, and information of the private virtual channel.
[0099] The router type is used to indicate whether the router is a border router, and it is usually set in accordance with the router's identifier.
[0100] The information for a router group includes: the router group identifier, the identifiers of the border routers included in the router group, and the identifiers of the internal routers included in the router group.
[0101] The information for a Private Virtual Channel (PVC) includes the PVC identifier and the identifiers of the two routers connected by the PVC. The two routers connected by the PVC can be a border router and an internal router, or two internal routers.
[0102] The aforementioned routing configuration information is output to each router in the core system, enabling them to route data packets between cores according to this configuration information. This reduces the number of dedicated virtual channels while ensuring that the core system's routing is free of deadlocks. It should be noted that, for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, as some steps may be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of this application.
[0103] Reference Figure 10 This diagram illustrates a structural block diagram of an embodiment of a virtual channel setting device for an on-chip network according to this application. The virtual channel setting device 300 for an on-chip network may specifically include the following modules: The packet module 301 is used to group multiple routers on the chip to obtain at least one router group; each router group includes: a border router and at least one internal router.
[0104] Module 302 is used to construct minimum spanning trees for at least one router group; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the paths from the border routers to the internal routers, or the paths between the internal routers.
[0105] Add module 303 to add dedicated virtual channels on the edges in the minimum spanning tree. These dedicated virtual channels are used for data transmission between cores.
[0106] Optionally, the grouping module 301 is also used for: Multiple border router groups are determined based on the multiple routers on the core, and each border router group includes at least one border router. For each border router group, the sum of the distances between each internal router on the core and the nearest border router of the internal router in the border router group is calculated as the path length of the border router group. Each internal router on the core includes routers outside the border router group. Each border router in the border router group with the smallest path length is designated as a border router in a router group, and the internal routers are assigned to the router group containing the target nearest border router of the internal router. The target nearest border router is the nearest border router in the border router group with the smallest path length.
[0107] Optionally, the grouping module 301 is also used for: If there is one target nearest border router on an internal router, the internal router is assigned to the internal router group where the target nearest border router is located. After the assignment of each internal router with one target nearest border router on the core is completed, for each internal router with multiple target nearest border routers on the core, the number of internal routers already assigned in the router group where the multiple target nearest border routers are located is determined, and the internal router with multiple target nearest border routers is assigned to the router group with the smallest number of assigned internal routers.
[0108] Optionally, the grouping module 301 is also used for: Obtain information about other chips on the system-on-a-chip (SoC) that communicate with the chip. This information includes the number and / or size of the other chips. Determine the number of border routers based on the chip size and the information about the other chips. Combine all routers corresponding to the number of border routers among the multiple routers of the chip to form multiple border router groups.
[0109] Optionally, module 303 is also used for: Determine the layout orientation of multiple chips on the system-on-a-chip; if the routing directions supported by the chip include the layout orientation, select the edge consistent with the layout orientation from the minimum spanning tree as the target edge; add a dedicated virtual channel for bidirectional data transmission on the target edge.
[0110] Optionally, module 303 is also used for: Identify the remaining cores that have data communication with the core, and determine the layout orientation between the remaining cores and the cores.
[0111] Optionally, module 303 is also used for: Add a dedicated virtual channel to each edge of the minimum spanning tree; construct the channel dependency graph of the core particle according to the local routing algorithm of the core particle. The nodes in the channel dependency graph include dedicated virtual channels and general virtual channels of the core particle. The edges in the channel dependency graph represent the dependency relationship between nodes; for any dedicated virtual channel, if there is no cycle in the channel dependency graph after deleting the dedicated virtual channel, then delete the dedicated virtual channel.
[0112] Optionally, it also includes: The distribution module 304 is used to send the routing configuration information of the core to each router in the core. The routing configuration information includes: the type of each router on the core, information of at least one router group corresponding to the border router group with the shortest path length, and information of the private virtual channel.
[0113] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.
[0114] Figure 11 This is a structural block diagram illustrating an electronic device 400 according to an exemplary embodiment. For example, the electronic device 400 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.
[0115] Reference Figure 11 The electronic device 400 may include one or more of the following components: processing component 402, memory 404, power supply component 406, multimedia component 408, audio component 410, input / output (I / O) interface 412, sensor component 414, and communication component 416.
[0116] Processing component 402 typically controls the overall operation of electronic device 400, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 402 may include one or more processors 420 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 402 may include one or more modules to facilitate interaction between processing component 402 and other components. For example, processing component 402 may include a multimedia module to facilitate interaction between multimedia component 408 and processing component 402.
[0117] Memory 404 is configured to store various types of data to support the operation of device 400. Examples of this data include instructions for any application or method operating on electronic device 400, contact data, phonebook data, messages, pictures, videos, etc. Memory 404 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
[0118] Power component 404 provides power to various components of electronic device 400. Power component 404 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 400.
[0119] Multimedia component 408 includes a screen that provides an output interface between electronic device 400 and a user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 408 includes a front-facing camera and / or a rear-facing camera. When electronic device 400 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.
[0120] Audio component 410 is configured to output and / or input audio signals. For example, audio component 410 includes a microphone (MIC) configured to receive external audio signals when electronic device 400 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 404 or transmitted via communication component 416. In some embodiments, audio component 410 also includes a speaker for outputting audio signals.
[0121] I / O interface 412 provides an interface between processing component 402 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.
[0122] Sensor assembly 414 includes one or more sensors for providing state assessments of various aspects of electronic device 400. For example, sensor assembly 414 may detect the on / off state of device 400, the relative positioning of components such as the display and keypad of electronic device 400, changes in position of electronic device 400 or a component of electronic device 400, the presence or absence of user contact with electronic device 400, orientation or acceleration / deceleration of electronic device 400, and temperature changes of electronic device 400. Sensor assembly 414 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 414 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 414 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.
[0123] Communication component 416 is configured to facilitate wired or wireless communication between electronic device 400 and other devices. Electronic device 400 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 414 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 414 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.
[0124] In an exemplary embodiment, the electronic device 400 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.
[0125] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 404 including instructions, which can be executed by a processor 420 of an electronic device 400 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.
[0126] A non-transitory computer-readable storage medium, when the instructions in the storage medium are executed by the processor of a terminal, enables the terminal to execute the virtual channel setup method of the on-chip network described above.
[0127] A computer product includes a computer program for implementing the virtual channel setup method of the above-described on-chip network.
[0128] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.
[0129] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus, or computer program products. Therefore, embodiments of this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of this application can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0130] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0131] These computer program instructions may also be stored in a computer-understandable memory that can direct a computer or other programmable data processing terminal device to operate in a predictive manner, such that the instructions stored in the computer-understandable memory produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0133] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.
[0134] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device 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 terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes the element.
[0135] The foregoing has provided a detailed description of a virtual channel setting method and apparatus for an on-chip network, an electronic device, and a storage medium provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for setting a virtual channel of a network-on-chip, characterized by, include: Multiple routers on the chip are grouped to obtain at least one router group; Each router group includes: a border router and at least one internal router; For each of the at least one router group, a minimum spanning tree is constructed; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the path lengths from the border routers to the internal routers, or the paths between the internal routers. A dedicated virtual channel is added to the edges in the minimum spanning tree. This dedicated virtual channel is used for data transmission between cores.
2. The method of claim 1, wherein, Grouping multiple routers on the chip yields at least one router group, including: Multiple border router groups are determined based on multiple routers on the core, and each border router group includes at least one border router. For each of the border router groups, the sum of the distances between each internal router on the core and the nearest border router of the internal router in the border router group is calculated as the path length of the border router group. Each internal router on the core includes routers outside the border router group. Each border router in the border router group with the shortest path length is considered as a border router in the same router group. The internal router is then assigned to the router group containing the target nearest border router of the internal router, where the target nearest border router is the nearest border router in the border router group with the shortest path length.
3. The method of claim 2, wherein, Assigning the router to the router group containing the nearest border router includes: If there is a target nearest border router for the internal router, the internal router is assigned to the internal router group where the target nearest border router of the internal router is located; After the internal routers with one target nearest border router on the core are partitioned, for each internal router with multiple target nearest border routers on the core, the number of internal routers already partitioned in the router group to which the multiple target nearest border routers of the internal router are located is determined, and the internal router with multiple target nearest border routers is partitioned into the router group with the smallest number of partitioned internal routers.
4. The method of claim 2, wherein, Multiple border router groups are determined based on the multiple routers on the core, including: Obtain information about other chips on the system-on-a-chip (SoC) where the chip is located that communicate with the chip, including: the number of other chips and / or the size of other chips; The number of border routers is determined based on the size of the core and the information of the remaining cores; The number of routers corresponding to the number of border routers in the multiple routers of the core are combined to form multiple border router groups.
5. The method according to any one of claims 1 to 4, characterized in that, Add dedicated virtual channels to the edges in the minimum spanning tree, including: Determine the layout orientation of multiple chips on the system-on-a-chip; If the routing direction supported by the core includes the layout direction, then the edge that is consistent with the layout direction is selected from the minimum spanning tree as the target edge; A dedicated virtual channel capable of bidirectional data transmission is added to the target side.
6. The method of claim 5, wherein, Determine the layout orientation of multiple chips on the system-on-a-chip, including: Identify the remaining cores that have data communication with the core, and determine the layout orientation between the remaining cores and the core.
7. The method according to any one of claims 1 to 4, characterized in that, Add dedicated virtual channels to the edges in the minimum spanning tree, including: Add a dedicated virtual channel to each edge of the minimum spanning tree; The channel dependency graph of the core is constructed according to the local routing algorithm of the core. The nodes in the channel dependency graph include the dedicated virtual channel and the general virtual channel of the core. The edges in the channel dependency graph represent the dependency relationship between the nodes. For any of the dedicated virtual channels, if the channel dependency graph does not contain any loops after deleting the dedicated virtual channel, then the dedicated virtual channel is deleted.
8. The method according to any one of claims 1 to 4, characterized in that, Also includes: The routing configuration information of the core is sent to each router of the core. The routing configuration information includes: the type of each router on the core, information of at least one router group corresponding to the border router group with the shortest path length, and information of the dedicated virtual channel.
9. A virtual channel setting device for an on-chip network, characterized in that, include: The grouping module is used to group multiple routers on the chip to obtain at least one router group; Each router group includes: a border router and at least one internal router; A construction module is used to construct a minimum spanning tree for each of the at least one router group; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the path lengths from the border routers to the internal routers, or the paths between the internal routers. An add module is used to add dedicated virtual channels on the edges in the minimum spanning tree, the dedicated virtual channels being used for inter-core data transmission.
10. An electronic device, comprising: It includes memory and one or more programs, wherein one or more programs are stored in memory and configured to be executed by one or more processors. The one or more programs contain instructions for performing the following operations: Multiple routers on the chip are grouped to obtain at least one router group; each router group includes: a border router and at least one internal router; For each of the at least one router group, a minimum spanning tree is constructed; the root node of the minimum spanning tree is the border router; the child nodes of the minimum spanning tree are the internal routers; the edges of the minimum spanning tree are the path lengths from the border routers to the internal routers, or the paths between the internal routers. A dedicated virtual channel is added to the edges in the minimum spanning tree. This dedicated virtual channel is used for data transmission between cores.
11. A readable storage medium, characterized by, When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device is able to perform the virtual channel setup method for an on-chip network as described in one or more of claims 1 to 8.
12. A computer program product, characterised in that, Includes a computer program for implementing the virtual channel setup method for a network-on-chip as described in one or more of claims 1 to 8.
