An Internet verification platform for a system-on-a-chip
By employing a four-layer independently defined address information description file and dynamic parsing technology in the on-chip system's Internet verification platform, the inflexible address modeling and conflict control problems in existing Internet verification technologies are solved, achieving efficient and stable address resource management and data comparison.
Patent Information
- Application Number
- CN202611491230.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-23
- Publication Date
- 2026-11-17
AI Technical Summary
Existing Internet verification methods lack efficient and flexible full-chip address modeling, making it difficult to effectively constrain address resources and control conflicts, especially in multi-address space scenarios where data comparison is challenging.
An on-chip system-on-a-chip (SoC) interconnection verification platform is adopted. Through components such as a real-time parser, incentive agent builder, subscriber, and scoreboard, it decouples the address segment definition and mapping relationship using four independently defined address information description files, dynamically parses incentive agent architecture information, and realizes flexible and efficient full-chip address modeling and address resource management.
It improves verification and comparison efficiency, reduces maintenance costs, ensures the stability and iteration efficiency of the verification environment, effectively avoids address conflicts and uneven coverage, and improves data comparison efficiency in multi-address space and offset address scenarios.
Smart Images

Figure CN123115040A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of chip verification technology, and in particular to an interconnection verification platform for a system-on-a-chip. Background Technology
[0002] In the design and verification of ultra-large-scale system-on-chip (SoC), the network-on-chip (NoC) undertakes the core functions of data transmission and routing across the entire chip. However, the large number of routing nodes, complex address mapping relationships, numerous entry and exit points from IP cores to the network, and the fact that different entry and exit points often contain different protocol conversion logics result in extremely high verification complexity for the network.
[0003] In existing Internet verification methods, address information is usually hard-coded into the reference model of the verification environment or exists in the form of a simple configuration file, lacking an efficient, accurate, and flexible full-chip address modeling method.
[0004] Therefore, when multiple master devices initiate random access simultaneously, existing technologies cannot effectively constrain address resources and control conflicts. In multi-address space scenarios, the same physical address manifests as different address values in different address spaces, making it difficult for existing technologies to perform data comparison. Summary of the Invention
[0005] This invention provides an interconnection network verification platform for a system-on-a-chip, which enables flexible and efficient full-chip address modeling, reduces maintenance costs, and improves verification and comparison efficiency.
[0006] According to one aspect of the present invention, an interconnection verification platform for a system-on-a-chip is provided, comprising a real-time parser, an incentive agent builder, a subscriber, and a scoreboard; The real-time parser is used to parse the pre-built address information description file at simulation time 0, generate the address resource manager, parse the incentive agent architecture information description file, obtain the incentive agent configuration information, and call the incentive agent builder according to the incentive agent configuration information; An incentive agent builder is used to build and configure multiple incentive agents according to the incentive agent configuration information to form an incentive agent pool; wherein, the incentive agent pool includes master devices and slave devices; The master device is used to send address resource requests to the address resource manager, and to randomly access the address resources within the authorized range returned by the address resource manager, and send the first randomly accessed address to the subscriber; The device is used to send a second address to the subscriber upon receiving access from the internet under test; The subscriber is used to standardize the first and second addresses that need to be written to the scoreboard and then write them to the scoreboard. The scoring board is used to compare all written address pairs. If all address pairs pass the comparison, the test network verification is confirmed to be successful.
[0007] Optionally, the description of each address space in the address information description file includes: address segment information, mapping relationship information between slave devices and address segments, allowed access range information of master devices, and address space index information; The address segment information is defined independently of the mapping relationship information between the slave device and the address segment and the allowed access range information of the master device. The mapping relationship information between the slave device and the address segment is used to establish the association between the slave device and the defined address segment.
[0008] Optionally, the address segment information includes the address segment name, starting address, address size, address segment type, access attributes, and random access switch status; The mapping relationship information between slave devices and address segments includes the slave device instance name and the address segment list corresponding to each slave device; the address segment list uses wildcards and / or set operators to describe the set of address segments; The allowed access range information of the master device includes the master device instance name, the list of allowed address ranges for each master device, and the random access switch status; the address range list uses wildcards and / or set operators to describe the set of address ranges, or uses preset keywords to represent all address ranges under this address space; The address space index information includes unique identifiers for multiple address spaces, as well as address remapping rules used to describe the address change patterns of the same address segment across different address spaces; the address change patterns include addition offsets, subtraction offsets, and / or high-order address string replacements.
[0009] Optionally, the real-time parser is specifically used for: At simulation time 0, the pre-built address information description file is parsed; Load the address segment information in the address information description file into the address segment attribute data structure in the address resource manager; Based on the mapping relationship information between slave devices and address segments in the address information description file, analyze and generate a slave device address awareness range list in the address resource manager; Based on the allowed access range information of the master device in the address information description file, generate the master device address range access truth table in the address resource manager; The address space index information in the address information description file is loaded into the address space mapping data structure in the address resource manager.
[0010] Optionally, the address resource manager is used for: The system receives an address resource request from the master device, and allocates an authorized range of address resources to the master device based on the master device address segment access truth table, address segment attribute data structure, and global access control policy in the address resource manager.
[0011] Optionally, the address resource manager is also used for: The system receives actual access address information from the master device and updates the usage status of the corresponding address segment in the address segment attribute data structure based on the actual access address information, so that it can be used in the next address resource allocation.
[0012] Optionally, the address resource manager is a global singleton, which is shared among the real-time parser, incentive agent builder, master device, slave device, subscriber and scoreboard in the on-chip system's interconnection verification platform through a static acquisition interface; The address resource manager also provides an application programming interface for performing any of the following control operations during simulation: enabling or disabling random access to a specified address range, enabling or disabling random access to a specified master device, modifying the access permissions of a specified master device to a specified address range, and configuring the access conditions of the master device.
[0013] Optionally, the subscriber is specifically used for: The system receives a first address sent by the master device. When the source addressing space of the master device is different from the target addressing space of the slave device, the system converts the first address from the source addressing space address of the master device to the target addressing space address according to the addressing space mapping data structure in the address resource manager, so as to predict the receiving address of the slave device. Based on the receiving address of the slave device and / or the connectivity of the interconnection network under test, a drop prediction is made for the transmission transactions corresponding to the first address; When the prediction result indicates that the transmission transaction will be discarded by the Internet, the first address will not be written into the scoring board; Otherwise, the first address is converted into a scoreboard address structure and then written to the scoreboard.
[0014] Optionally, the subscriber is specifically used for: Receive the second address sent by the slave device, call the address resource manager, query the base address of the address segment corresponding to the slave device in the slave device address awareness range list, and complete the second address according to the base address; In cross-address space scenarios, based on the address space mapping data structure in the address resource manager, the completed second address is converted into a global address under the source address space, and the global address is converted into a scoreboard address structure and then written into the scoreboard.
[0015] Optionally, the incentive agent pool also includes a monitor; the monitor is deployed at the routing nodes of interest in the interconnected network under test.
[0016] The technical solution of this invention adopts a four-layer independent definition structure for the address information description file, decoupling the address segment definition from the mapping relationship between slave or master devices. When the interconnection network design changes, only the description of the corresponding layer needs to be modified, without affecting other parts, thus ensuring the stability of the verification environment. By dynamically parsing and loading the address information description file and the stimulus agent architecture information description file at simulation time 0, simulation can be started without recompiling the verification environment when the design changes, which greatly improves the verification iteration efficiency compared to traditional full-chip incremental compilation. The address resource manager performs global management of random access address resources of multiple master devices, effectively avoiding address conflicts and uneven coverage in random testing. The subscriber uses forward remapping to predict the slave device's received address and performs drop prediction, and uses base address completion and reverse remapping to restore the offset address to the global address, improving the data comparison efficiency in multi-address space and offset address scenarios.
[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a schematic diagram of the structure of an interconnection network verification platform for a system-on-a-chip according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of another system-on-a-chip interconnection verification platform provided according to an embodiment of the present invention. Detailed Implementation
[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0022] Figure 1 This is a schematic diagram of a system-on-a-chip (SoC) interconnection verification platform provided in an embodiment of the present invention. This embodiment is applicable to the verification of SoC interconnections.
[0023] like Figure 1 As shown, the on-chip system's interconnection verification platform includes a real-time parser 110, an incentive agent builder 120, a subscriber 130, and a scoreboard 140.
[0024] Optionally, a system-on-a-chip (SoC) is an integrated circuit system that integrates multiple functional modules onto a single chip. In the embodiments described in this invention, SoC can refer to a very large-scale chip containing a large-scale interconnect network, such as a cloud-based artificial intelligence training or inference chip.
[0025] Optionally, the interconnection network is a subsystem in the system-on-a-chip used to connect various IP cores and realize data transmission and routing. It consists of multiple routing nodes and transmission links. IP cores are reusable functional modules in the chip.
[0026] Optionally, the real-time parser 110 is used to parse the pre-built address information description file at simulation time 0, generate an address resource manager, parse the incentive agent architecture information description file, obtain incentive agent configuration information, and call the incentive agent builder according to the incentive agent configuration information.
[0027] Optionally, the address information description file is a pre-built text file that uses a four-level table structure to describe all the address information of the entire chip.
[0028] Optionally, the description of each address space in the address information description file includes: address segment information, mapping relationship information between slave devices and address segments, allowed access range information of master devices, and address space index information; wherein, the address segment information is defined independently of the mapping relationship information between slave devices and address segments and the allowed access range information of master devices, and the mapping relationship information between slave devices and address segments is used to establish the association between slave devices and defined address segments.
[0029] Optionally, the address information description file adopts a multi-level table structure. Each address space description includes independently defined address segment information, mapping information between slave devices and address segments, allowed access range information for master devices, and address space index information. The address segment information is defined independently of the mapping information between slave devices and address segments and the allowed access range information for master devices. That is, the basic attributes of all address segments across the entire chip are fully defined in the first layer, and then the associations between slave devices, master devices, and the defined address segments are established in the second and third layers respectively. This allows for changes to the interconnection design, requiring only modifications to the descriptions of the corresponding layers. For example, when an address segment is split into two, only the address segment definition needs to be modified in the first layer, and the mapping relationships adjusted accordingly in the second and third layers. No changes to the definitions of other address segments or the configurations of other slave or master devices are required, thus ensuring the stability of the verification environment.
[0030] Optionally, the address segment information includes the address segment name, starting address, address size, address segment type, access attributes, and random access switch status; The mapping relationship information between slave devices and address segments includes the slave device instance name and the address segment list corresponding to each slave device; the address segment list uses wildcards and / or set operators to describe the set of address segments; The allowed access range information of the master device includes the master device instance name, the list of allowed address ranges for each master device, and the random access switch status; the address range list uses wildcards and / or set operators to describe the set of address ranges, or uses preset keywords to represent all address ranges under this address space; The address space index information includes unique identifiers for multiple address spaces, as well as address remapping rules used to describe the address change patterns of the same address segment across different address spaces; the address change patterns include addition offsets, subtraction offsets, and / or high-order address string replacements.
[0031] Optionally, the address segment information is the first level of the address information description file, which is used to independently define the basic attributes of all address segments of the entire chip. The description of each address segment includes the address segment name, starting address, address size, address segment type, access attributes, and random access switch status.
[0032] Optionally, the address segment name is a unique identifier for the address segment. In an address information description file containing multiple address spaces, all address segment names are globally unique; the start address is the starting address value of the address segment; the address size is the size of the address segment, which, combined with the start address, determines the end address of the address segment; the address segment type is the type identifier of the address segment; the access attributes are the access permission attributes of the address segment, such as readable and writable, read-only, write-only, etc.; the random access switch status is used to describe whether the address segment is allowed to be accessed by random testing. When the switch status is off, no random stimulus from any master device will access the address segment.
[0033] Optionally, the address segment information may also include a description of invalid or reserved address segments. By explicitly defining reserved address segments in address modeling, it is possible to actively verify in random tests whether access to reserved addresses is correctly blocked or reported by the Internet, thereby improving the completeness of the verification.
[0034] Optionally, the mapping relationship information between slave devices and address segments is the second layer of the address information description file, which describes which address segments each slave device or monitor can perceive. Each record includes the slave device instance name and the address segment list corresponding to each slave device, wherein the address segment list corresponding to the slave device is the set of address segments that the slave device can perceive.
[0035] Optionally, the definition of the address range list supports wildcards (*) and set operators for flexible and efficient description. The wildcard (*) can match all address ranges or match address range names with a specific prefix. Set operators include addition and subtraction operators. The addition operator (+) can represent the union of multiple address ranges, such as REGION_A+REGION_B representing the set of address ranges A and B. The subtraction operator (-) can exclude a specified address range from the set of address ranges, such as *-REGION_C representing all address ranges except address range C.
[0036] The advantage of this setup is that, by using wildcards and set operators, complex sets of address ranges can be described with concise expressions, avoiding the need to list a large number of address range names one by one.
[0037] Optionally, the allowed access range information for the master device is the third level of the address information description file, which describes which address ranges each master device is allowed to access.
[0038] Optionally, the list of address ranges that the master device is allowed to access is a set of address ranges that the master device is allowed to access. The address range list also supports wildcards and set operators. The random access switch status indicates whether the master device participates in random testing. When the switch status is off, the master device will not generate random stimuli.
[0039] Optionally, the address range list of the master device supports the use of preset keywords, such as REGION_ALL, to represent all address ranges under this address space. Using preset keywords can avoid listing them one by one when the master device needs to access all address ranges, simplifying the expression.
[0040] Optionally, the address space index information is the fourth layer of the address information description file, used to support the description of multiple address spaces in the whole chip and the address mapping relationship between different address spaces.
[0041] Optionally, the unique identifier of multiple address spaces means that each address space has a corresponding unique index, which is used to distinguish different address spaces; the address remapping rule can be used to describe the address change pattern of the same address segment between different address spaces. Specifically, a specified index column can be added to the address segment information section to describe the address change pattern of the current address segment in the specified address space.
[0042] Optionally, the address change pattern, i.e. the remapping rule, can support mapping methods such as addition offset, subtraction offset, and high-order address string replacement; where addition offset is adding an offset to the entire address; subtraction offset is subtracting an offset from the entire address; and high-order address string replacement is replacing the high-order bits of the address with a new prefix string.
[0043] Optionally, address space index information and remapping rules can accurately describe the address value differences of the same physical address in different address spaces, providing a basis for address translation and data comparison in cross-address space scenarios.
[0044] Optionally, the incentive agent configuration information describes which incentive agents need to be built in the verification environment, as well as the protocol type, physical interface parameters, and working mode (master device / slave device / monitor) of each incentive agent. The real-time parser 110 can call the incentive agent builder 120 to complete the dynamic construction of the incentive agents based on the incentive agent configuration information.
[0045] Optionally, the incentive agent architecture information description file can also be in plain text format. Modifications to this file do not require recompilation. When it is necessary to add or remove incentive agents or modify their configuration, simply modify this file and rerun the simulation.
[0046] Optionally, the incentive agent builder 120 is used to build and configure multiple incentive agents according to the incentive agent configuration information to form an incentive agent pool; wherein the incentive agent pool includes master devices and slave devices.
[0047] Optionally, the incentive agent pool may also include monitors, which are deployed at the routing nodes of interest in the interconnected network under test.
[0048] Optionally, the incentive agent builder 120 is a component in the verification platform responsible for dynamically building incentive agents. It has a built-in multi-protocol incentive element builder, which can call the matching builder according to the protocol type specified in the incentive agent configuration information to complete the construction and physical attribute configuration of the corresponding protocol incentive agent.
[0049] Optionally, since the configuration information comes from the dynamic parsing of external text files, modifying the number, type, or parameters of the stimulus agents does not require recompiling the verification environment. The iteration speed is fast, which enables the test platform environment that requires a large number of stimulus elements to converge and stabilize quickly.
[0050] Optionally, the incentive agent pool is a collection of all configured incentive agents, which may include master devices and slave devices.
[0051] Optionally, the master device is an incentive agent in the incentive agent pool used to simulate the behavior of a real master device; the slave device is an incentive agent in the incentive agent pool used to simulate the behavior of a real slave device; and the monitor is an incentive agent in the incentive agent pool used to passively observe the transmission behavior of the Internet.
[0052] Optionally, subscriber 130 is used to standardize the first address and the second address that need to be written to the scoreboard and write them to the scoreboard.
[0053] Optionally, a scoring board 140 is used to compare all written address pairs. If all address pairs pass the comparison, the test network verification is confirmed to be successful.
[0054] Optionally, the subscriber 130 is the core component responsible for data preprocessing in the verification platform of this invention. Before the data is written to the scoring board, the addresses of the master device and the slave device can be preprocessed to ensure that the data at both ends can be compared under a unified benchmark.
[0055] Optionally, the scoreboard 140 is a component in the verification platform responsible for data comparison. The scoreboard 140 can receive the scoreboard address structure from the subscriber 130 and compare the sender address and receiver address belonging to the same transmission transaction into address pairs.
[0056] Figure 2 This is a schematic diagram of the structure of another on-chip system interconnection verification platform provided in an embodiment of the present invention. (See diagram below.) Figure 2 As shown, the on-chip system's interconnection verification platform also includes an address resource manager 150 configured by a real-time resolver 110, and a master device 160 and a slave device 170 constructed by an incentive agent builder 120; wherein, there can be multiple master devices 160 and slave devices 170, which are managed in batches in the incentive agent pool.
[0057] Specifically, the real-time parser can be used for: At simulation time 0, the pre-built address information description file is parsed; Load the address segment information in the address information description file into the address segment attribute data structure in the address resource manager; Based on the mapping relationship information between slave devices and address segments in the address information description file, analyze and generate a slave device address awareness range list in the address resource manager; Based on the allowed access range information of the master device in the address information description file, generate the master device address range access truth table in the address resource manager; The address space index information in the address information description file is loaded into the address space mapping data structure in the address resource manager.
[0058] Optionally, the address segment attribute data structure is stored in the form of a dynamic array or an associative array. Each element corresponds to an address segment and includes fields such as the start address, end address, access attributes, and random access switch status of the address segment. It also includes a runtime-maintained usage status field, which records which master devices are currently accessing the address segment and the overlap of accesses, for reference when allocating address resources.
[0059] Optionally, for each slave device, the real-time parser 110 resolves all address ranges that the slave device can actually perceive based on its address range list, and calculates the base address and address range of each address range, ultimately forming a slave device address perception range list. The slave device address perception range list is stored in the form of an associative array, indexed by the slave device instance name, and each element contains information such as the base address, start address, and end address of all address ranges corresponding to that slave device.
[0060] Optionally, the master device address segment access truth table is a two-dimensional Boolean matrix, with rows corresponding to master devices and columns corresponding to address segments. The value of each matrix element indicates whether the master device is allowed to access that address segment. When generating the truth table, the real-time parser 110 needs to parse the list of master device address segments, expand it into a specific set of address segments, and set the values of the corresponding elements in the truth table accordingly. After the truth table is generated, the address resource manager can quickly query whether a master device is allowed to access a certain address segment when allocating address resources.
[0061] Optionally, the address space mapping data structure stores the unique identifier of each address space, as well as the address remapping rules for the same address segment across different address spaces. Each remapping rule records the source address space, the target address space, the address segment name, and the remapping method. The address space mapping data structure provides a rule-based basis for forward and reverse remapping in cross-address space scenarios.
[0062] Optionally, the master device 160 is used to send an address resource request to the address resource manager 150, and to perform random access within the authorized address resource range returned by the address resource manager 150, and send the first randomly accessed address to the subscriber 130.
[0063] Optionally, the slave device 170 is configured to send a second address to the subscriber 130 upon receiving access from the internet under test.
[0064] Optionally, the first address is the address sent by the master device 160 when it initiates a random access, which is the address under the source addressing space where the master device is located; the second address is the address reported by the slave device 170 after receiving the access, which is usually the offset address after removing the base address under the addressing space where the slave device is located.
[0065] Optionally, the address resource manager 150 is configured to: receive an address resource request sent by the master device 160, allocate an allowed range of address resources to the master device 160 and return authorization based on the master device address segment access truth table, address segment attribute data structure and global access control policy in the address resource manager 150.
[0066] Optionally, after receiving the address resource request sent by the master device 160, the address resource manager 150 queries the allowed address ranges of the master device according to the master device address range access truth table; according to the address range attribute data structure, it checks the random access switch status and current usage status of each allowed address range to determine whether random access is enabled for each allowed address range and whether it is occupied by other master devices, and the occupation status; based on the current global access control policy and the allowed address ranges, the random access switch status and current usage status of each allowed address range, it allocates a suitable address resource range to the master device 160 and returns authorization. The master device 160 generates random incentives within the authorized address resource range in conjunction with its own random policy.
[0067] Optionally, the global access control policy may include overlapping or non-overlapping modes. In non-overlapping mode, the address resource manager ensures that address resources allocated to different master devices do not overlap. In overlapping mode, multiple master devices are allowed to access the same address segment. The global access control policy may also include address segment coverage maximization or minimization settings to control the number of address segments accessed in random tests. In maximization mode, more address segments are accessed as much as possible, while in minimization mode, a small number of address segments are accessed in a concentrated manner to create pressure.
[0068] Optionally, the Address Resource Manager 150 enables global control over random access to multiple master devices, avoiding address conflicts and uneven coverage caused by each master device accessing the address independently.
[0069] Optionally, the address resource manager 150 is also used to: receive actual access address information fed back by the master device, and update the usage status of the corresponding address segment in the address segment attribute data structure according to the actual access address information, so as to be used in the next address resource allocation.
[0070] Optionally, after completing the random access, the master device 160 feeds back the actual access address information to the address resource manager 150. Based on the feedback information, the address resource manager 150 updates the usage status of the corresponding address segment in the address segment attribute data structure, including which master device accessed the address segment, the accessed address range, the number of accesses, etc. The updated usage status information will be used for decision-making in the next address resource allocation. For example, in non-overlapping mode, it avoids address ranges that have already been occupied, and in coverage maximization mode, it prioritizes the allocation of address segments that have not yet been accessed.
[0071] The advantage of this setting is that it allows the address resource manager to monitor the usage of all chip address resources in real time and dynamically adjust subsequent resource allocation strategies accordingly.
[0072] Optionally, the address resource manager 150 is a global singleton and is shared among the real-time parser 110, incentive agent builder 120, master device 160, slave device 170, subscriber 130 and scoreboard 140 in the on-chip system's interconnection verification platform through a statically acquired interface.
[0073] Optionally, the Address Resource Manager 150 also provides an application programming interface for performing any of the following control operations during simulation: enabling or disabling random access to a specified address range, enabling or disabling random access to a specified master device, modifying the access permissions of a specified master device to a specified address range, and configuring the access conditions of the master device.
[0074] Optionally, the Address Resource Manager 150 adopts a global singleton pattern in the verification environment, ensuring that there is only one instance of the Address Resource Manager in the entire verification environment. The Address Resource Manager 150 can provide a static retrieval interface, through which any component in the verification environment can obtain the instance of the Address Resource Manager from any location, thereby realizing convenient sharing of global address resource information. The global singleton pattern guarantees the consistency and uniqueness of address resource information, avoiding data synchronization problems caused by multiple instances.
[0075] Optionally, the Address Resource Manager 150 provides an Application Programming Interface (API) for fine-grained control of address resources during simulation. The API may specifically include: an address segment-level control API for enabling or disabling random access to specified address segments; once random access to an address segment is disabled, no random stimulus from any master device will access that address segment; a master device-level control API for enabling or disabling random access to a specified master device; after disabling, the master device will no longer generate random stimuli; a master device-address segment-level control API for modifying the access permissions of a specified master device to a specified address segment, i.e., dynamically modifying the corresponding elements in the master device's address segment access truth table. For example, a master device can bypass certain globally enabled address segments, meaning that the master device will not access these address segments during random testing, thus concentrating the master device's access to a specific area and creating localized data pressure; and a master device access condition configuration API for configuring the master device's operating conditions, such as power status, clock enable, and security status, to simulate access behavior under different operating conditions.
[0076] Optionally, through the application programming interface provided by the address resource manager 150, test cases can dynamically adjust the strategy and scope of random testing during simulation operation to achieve flexible and controllable verification scenarios.
[0077] Optional, subscriber 130, specifically used for: The system receives a first address sent by the master device. When the source addressing space of the master device is different from the target addressing space of the slave device, the system converts the first address from the source addressing space address of the master device to the target addressing space address according to the addressing space mapping data structure in the address resource manager, so as to predict the receiving address of the slave device. Based on the receiving address of the slave device and / or the connectivity of the interconnection network under test, a drop prediction is made for the transmission transactions corresponding to the first address; When the prediction result indicates that the transmission transaction will be discarded by the Internet, the first address will not be written into the scoring board; Otherwise, the first address is converted into a scoreboard address structure and then written to the scoreboard.
[0078] Optionally, when the source addressing space of the master device is different from the target addressing space of the slave device, the subscriber 130 can be used to call the address resource manager to convert the first address from the source addressing space address of the master device to the target addressing space address according to the address remapping rules in the addressing space mapping data structure, thereby predicting what address the slave device will receive if the transmission transaction reaches the target slave device through the Internet routing.
[0079] Optionally, in the same address space scenario, forward remapping is not required, as the first address itself is the address reference received by the device.
[0080] Optionally, the subscriber 130 can predict whether the transmission transaction corresponding to the first address will be dropped based on the receiving address of the slave device and / or the connectivity of the Internet under test. Drop prediction means determining whether the transaction will be dropped by a routing node during the routing process of the Internet. If the prediction result is that the transmission transaction will be dropped by the Internet, the first address will not be written into the scoring board. If the prediction result is that the transmission transaction will not be dropped, the first address will be converted into a scoring board address structure and then written into the scoring board.
[0081] Optionally, subscriber 130 can also be specifically used for: Receive the second address sent by the slave device, call the address resource manager, query the base address of the address segment corresponding to the slave device in the slave device address awareness range list, and complete the second address according to the base address; In cross-address space scenarios, based on the address space mapping data structure in the address resource manager, the completed second address is converted into a global address under the source address space, and the global address is converted into a scoreboard address structure and then written into the scoreboard.
[0082] Optionally, the global address is the address that is in the same addressing space base as the first address issued by the master device. In the same addressing space scenario, the address after padding the base address is the global address; in the cross addressing space scenario, a reverse remapping is required to obtain the global address.
[0083] Optionally, base address completion refers to the process of adding the base address of the address segment corresponding to the slave device to the second address reported by the slave device to obtain the complete address in the address space where the slave device is located.
[0084] Optionally, subscriber 130 can call address resource manager 150 to query the base address of the address segment corresponding to the slave device in the slave device address awareness range list, and complete the second address based on the base address. By adding the base address, the complete address under the target address space where the slave device is located can be obtained.
[0085] Optionally, in cross-address space scenarios, subscriber 130 can call address resource manager 150 to convert the completed second address into a global address under the source address space according to the address remapping rules in the address space mapping data structure, thereby restoring the address from the perspective of the slave device to the global address reference from the perspective of the master device, so as to compare it with the first address written to the scoring board by the sender.
[0086] Optionally, in the same address space scenario, there is no need to perform reverse remapping; the address padded with the base address is the global address.
[0087] Optionally, subscriber 130 converts the global address into a scoreboard address structure and writes it to the scoreboard. The address field in the scoreboard address structure is the recovered global address, which is on the same baseline as the first address written by the sender and can be directly compared.
[0088] Optionally, the scoreboard address structure is a normalized data structure used by the scoreboard, which is used to uniformly convert transmission transactions of different protocols into this structure before comparison.
[0089] Optionally, the scoring board 140 can compare address pairs consisting of the sender's address and the receiver's address belonging to the same transmission transaction.
[0090] Specifically, for each transmission transaction expected to reach the slave device, the sending subscriber writes the first address into the scoreboard as the expected result when the transaction is sent; when the transaction arrives at the slave device, the receiving subscriber writes the restored global address into the scoreboard as the actual result; the scoreboard 140 compares the expected result and the actual result of the same transaction: if the addresses match, the address comparison passes; if they do not match, an error is reported and the difference information is recorded.
[0091] Optionally, when the simulation ends, if all address pairs pass the comparison and there are no unmatched expected or actual results, the interconnection under test is deemed to have passed the verification.
[0092] Optionally, the incentive agent pool may also include monitors; monitors can be deployed at the routing nodes of interest in the interconnection under test, without driving any signals, but only collecting the address and data information of the transmission transactions passing through the node, so that the routing information of the interconnection can be incorporated into the verification system without modifying the overall structure of the address information description file, thus assisting in the verification of the correctness of the routing.
[0093] In one specific implementation, the complete workflow of the verification platform of the present invention is as follows: Phase 1: Simulation Initialization (Simulation Time 0) The real-time parser reads and parses the address information description file, maps the four types of information to the four types of data structures in the address resource manager, and generates the address resource manager. The real-time parser reads and parses the incentive agent architecture information description file, obtains the incentive agent configuration information, and calls the incentive agent builder.
[0094] The incentive agent builder dynamically constructs incentive agents such as master devices, slave devices, and monitors based on configuration information, forming an incentive agent pool.
[0095] The Address Explorer is ready as a global singleton, and all components can be obtained through the static interface.
[0096] Phase Two: Random Test Run The master device sends an address resource request to the address resource manager.
[0097] The Address Resource Manager allocates address resources and returns authorization based on the master device address range access truth table, address range attribute data structure, and global access control policy.
[0098] The master device randomly generates access incentives within the authorized range, sends them to the Internet under test, and sends the first address to the sender subscriber.
[0099] The master device sends the actual access address information back to the address resource manager, which then updates the address range usage status.
[0100] The sender subscriber performs positive remapping prediction and discard prediction on the first address. When it is determined that writing is required, the first address is normalized into the scoreboard address structure and written to the scoreboard.
[0101] The interconnected network under test transmits transactions to the target slave device according to routing logic.
[0102] Upon receiving an access request from the device, the second address is sent to the subscriber at the receiving end.
[0103] The receiver subscriber completes the base address of the second address, performs reverse remapping in cross-addressing scenarios, restores it to a global address, standardizes it into a scoreboard address structure, and writes it into the scoreboard.
[0104] The scoring board compares the address pairs.
[0105] Phase 3: Verification Result Determination After the simulation, if all address pairs pass the comparison and there are no unmatched items left in the scoring board, the verification of the Internet under test is confirmed to be successful; otherwise, the comparison failure and difference information are reported.
[0106] The technical solution of this invention adopts a four-layer independent definition structure for the address information description file, decoupling the address segment definition from the mapping relationship between slave or master devices. When the interconnection network design changes, only the description of the corresponding layer needs to be modified, without affecting other parts, thus ensuring the stability of the verification environment. By dynamically parsing and loading the address information description file and the stimulus agent architecture information description file at simulation time 0, simulation can be started without recompiling the verification environment when the design changes, which greatly improves the verification iteration efficiency compared to traditional full-chip incremental compilation. The address resource manager performs global management of random access address resources of multiple master devices, effectively avoiding address conflicts and uneven coverage in random testing. The subscriber uses forward remapping to predict the slave device's received address and performs drop prediction, and uses base address completion and reverse remapping to restore the offset address to the global address, improving the data comparison efficiency in multi-address space and offset address scenarios.
[0107] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.
[0108] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A system-on-a-chip (SoC) interconnection verification platform, characterized in that, This includes a real-time parser, an incentive agent builder, a subscriber, and a scoreboard; The real-time parser is used to parse the pre-built address information description file at simulation time 0, generate the address resource manager, parse the incentive agent architecture information description file, obtain the incentive agent configuration information, and call the incentive agent builder according to the incentive agent configuration information; An incentive agent builder is used to build and configure multiple incentive agents according to the incentive agent configuration information to form an incentive agent pool; wherein, the incentive agent pool includes master devices and slave devices; The master device is used to send address resource requests to the address resource manager, and to randomly access the address resources within the authorized range returned by the address resource manager, and send the first randomly accessed address to the subscriber; The device is used to send a second address to the subscriber upon receiving access from the internet under test; The subscriber is used to standardize the first and second addresses that need to be written to the scoreboard and then write them to the scoreboard. The scoring board is used to compare all written address pairs. If all address pairs pass the comparison, the test network verification is confirmed to be successful.
2. The system-on-chip interconnect network verification platform of claim 1, wherein, The description of each address space in the address information description file includes: address segment information, mapping relationship information between slave devices and address segments, allowed access range information of master devices, and address space index information; The address segment information is defined independently of the mapping relationship information between the slave device and the address segment and the allowed access range information of the master device. The mapping relationship information between the slave device and the address segment is used to establish the association between the slave device and the defined address segment.
3. The system-on-chip interconnect network verification platform of claim 2, wherein, The address segment information includes the address segment name, starting address, address size, address segment type, access attributes, and random access switch status; The mapping relationship information between slave devices and address segments includes the slave device instance name and the address segment list corresponding to each slave device; the address segment list uses wildcards and / or set operators to describe the set of address segments; The allowed access range information of the master device includes the master device instance name, the list of allowed address ranges for each master device, and the random access switch status; the address range list uses wildcards and / or set operators to describe the set of address ranges, or uses preset keywords to represent all address ranges under this address space; The address space index information includes unique identifiers for multiple address spaces, as well as address remapping rules used to describe the address change patterns of the same address segment across different address spaces; the address change patterns include addition offsets, subtraction offsets, and / or high-order address string replacements.
4. The on-chip system interconnection verification platform according to claim 2, characterized in that, The real-time parser is specifically used for: At simulation time 0, the pre-built address information description file is parsed; Load the address segment information in the address information description file into the address segment attribute data structure in the address resource manager; Based on the mapping relationship information between slave devices and address segments in the address information description file, analyze and generate a slave device address awareness range list in the address resource manager; Based on the allowed access range information of the master device in the address information description file, generate the master device address segment access truth table in the address resource manager; The address space index information in the address information description file is loaded into the address space mapping data structure in the address resource manager.
5. The on-chip system interconnection verification platform according to claim 4, characterized in that, The address resource manager is used for: The system receives an address resource request from the master device, and allocates an authorized range of address resources to the master device based on the master device address segment access truth table, address segment attribute data structure, and global access control policy in the address resource manager.
6. The on-chip system interconnection verification platform according to claim 4, characterized in that, The address resource manager is also used for: The system receives actual access address information from the master device and updates the usage status of the corresponding address segment in the address segment attribute data structure based on the actual access address information, so that it can be used in the next address resource allocation.
7. The on-chip system interconnection verification platform according to claim 4, characterized in that, The address resource manager is a global singleton, which is shared among the real-time parser, incentive agent builder, master device, slave device, subscriber and scoreboard in the on-chip system's interconnection verification platform through a static acquisition interface; The address resource manager also provides an application programming interface for performing any of the following control operations during simulation: enabling or disabling random access to a specified address range, enabling or disabling random access to a specified master device, modifying the access permissions of a specified master device to a specified address range, and configuring the access conditions of the master device.
8. The on-chip system interconnection verification platform according to claim 4, characterized in that, The subscriber is specifically used for: The system receives a first address sent by the master device. When the source addressing space of the master device is different from the target addressing space of the slave device, the system converts the first address from the source addressing space address of the master device to the target addressing space address according to the addressing space mapping data structure in the address resource manager, so as to predict the receiving address of the slave device. Based on the receiving address of the slave device and / or the connectivity of the interconnection network under test, a drop prediction is made for the transmission transactions corresponding to the first address; When the prediction result indicates that the transmission transaction will be discarded by the Internet, the first address will not be written into the scoring board; Otherwise, the first address is converted into a scoreboard address structure and then written to the scoreboard.
9. The on-chip system interconnection verification platform according to claim 4, characterized in that, The subscriber is specifically used for: Receive the second address sent by the slave device, call the address resource manager, query the base address of the address segment corresponding to the slave device in the slave device address awareness range list, and complete the second address according to the base address; In cross-address space scenarios, based on the address space mapping data structure in the address resource manager, the completed second address is converted into a global address under the source address space, and the global address is converted into a scoreboard address structure and then written into the scoreboard.
10. The on-chip system interconnection verification platform according to claim 1, characterized in that, The incentive agent pool also includes a monitor; the monitor is deployed at the route node of interest in the interconnected network under test.
