A FPGA layout method for optimizing timing by relaxing instances

Through the FPGA layout method that optimizes timing with relaxed instances, the problem of excessive delay in timing paths in FPGA design is solved. By releasing the layout constraints of intermediate instances, adjusting their position, meeting timing requirements, and improving design efficiency and performance.

CN116522837BActive Publication Date: 2024-11-19WUXI ESIONTECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202310437019.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-04-21
Publication Date
2024-11-19
Estimated Expiration
2043-04-21

AI Technical Summary

Technical Problem

In FPGA design, the starting point instance and end point instance of the critical timing path are located in and outside the rectangular area, respectively, resulting in too large distances between instances and too large delays, which cannot meet the timing requirements, affecting chip performance.

Method used

Through the FPGA layout method of optimizing timing of the slack instance, parse the logical netlist, determine the relaxed instance in the pending timing path and release its layout constraints, so that the time margin meets the timing requirements, adjust the position of the intermediate instance to correct the logical netlist layout constraints.

Benefits of technology

The FPGA layout is effectively adjusted to enable the pending timing paths to meet the timing requirements, improve design efficiency and optimize performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116522837B_ABST
    Figure CN116522837B_ABST
Patent Text Reader

Abstract

The present application discloses an FPGA layout method for optimizing timing by relaxing instances, and relates to the field of FPGA technology. The method parses a logic netlist and reads the layout constraints of the logic netlist to determine a timing path to be processed whose layout constraints of two endpoint instances meet the situation targeted by the present application and whose time margin does not meet the timing requirements under the existing layout constraints. Then, the timing path to be processed is preprocessed, and part or all of the intermediate instances in the timing path to be processed are attributed to relaxed instances to release the layout constraints first, so that a subsequent layout algorithm can adjust the positions of these instances according to the timing requirements, so that the timing path to be processed can meet the timing requirements through the subsequent layout algorithm correction, thereby meeting the design requirements, improving the design efficiency and optimizing the performance of the designed FPGA.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of FPGA technology, and in particular to an FPGA layout method for optimizing timing by relaxing instances. Background Art

[0002] Field-Programmable Gate Array (FPGA) is a widely used chip that can be used to implement a variety of complex functions. Before using FPGA, software design needs to be completed under specified user constraints so that the chip can work properly, including the operating frequency of the synchronous timing circuit and the timing requirements of the input and output devices.

[0003] In the design process of FPGA, there are usually specific layout constraints, which are used to constrain the layout of some (or all) instances in a specific module to a specific position or a specific rectangular area, so that when the layout algorithm is correctly implemented, the layout of some (or all) instances in the module can meet the constraints to which they belong. However, in the actual design process, the following situation may occur: the starting instance and the end instance of the critical timing path are respectively specified to be within the rectangular area and outside the rectangular area under their respective constraints. If the distance between the instance outside the rectangular area and the position of the rectangular area is too large, or the area of ​​the rectangular area itself is too large, the distance between the instance outside the rectangular area and the instance corresponding to the upper-level timing element connected to it will be too large, which will result in the failure to meet the timing requirements due to excessive delay, and the subsequent layout algorithm cannot be corrected to meet the timing requirements, which will seriously affect the performance of the chip. Summary of the invention

[0004] In view of the above problems and technical requirements, the applicant proposes an FPGA layout method for optimizing timing by relaxing instances. The technical solution of the application is as follows:

[0005] An FPGA layout method for optimizing timing by relaxing instances, the FPGA layout method comprising:

[0006] Parsing the logic netlist and reading the layout constraints of the logic netlist to determine a timing path to be processed, wherein one endpoint instance in the timing path to be processed has a layout constraint condition located within a layout constraint region, another endpoint instance has a layout constraint condition located at a target layout position outside the layout constraint region, and the timing margin of the timing path to be processed under the layout constraints of each instance contained does not meet the timing requirement;

[0007] For any timing path to be processed, release the layout constraint of the relaxed instance in the timing path to be processed so that the time margin of the timing path to be processed meets the timing requirement, and the relaxed instance is the middle instance in the timing path to be processed except the two endpoint instances;

[0008] The instances in the logic netlist are placed under the revised placement constraints of the logic netlist.

[0009] A further technical solution is that the method for determining a relaxation instance in a timing path to be processed includes:

[0010] Directly use all the intermediate instances in the timing path to be processed as relaxed instances;

[0011] Alternatively, each intermediate instance in the timing path to be processed is classified as a relaxed instance in a predetermined order, until the time margin of the timing path to be processed meets the timing requirement, and the layout constraints of the remaining intermediate instances are retained.

[0012] A further technical solution is that the method of attributing each intermediate instance in the to-be-processed timing path as a relaxed instance in a predetermined order includes:

[0013] The intermediate instances are classified as slack instances in descending order of their influence on the timing margin of the timing path to be processed.

[0014] A further technical solution is that the method of classifying each intermediate instance as a slack instance in descending order of influence on the timing margin of the timing path to be processed includes:

[0015] Determine the influence index of the transmission path between any two instances as d*w, where d is the distance of the transmission path between the two instances and w is the timing criticality of the transmission path between the two instances;

[0016] According to the order of influence index values ​​from large to small, the intermediate instances connecting the two ends of each transmission path are classified as relaxed instances.

[0017] A further technical solution is that the method of attributing each intermediate instance in the to-be-processed timing path as a relaxed instance in a predetermined order includes:

[0018] The intermediate instances in the to-be-processed timing path are sequentially classified as relaxed instances in the order determined by the connection relationship of the intermediate instances.

[0019] A further technical solution is that the method of attributing each intermediate instance to a relaxed instance in sequence according to the order determined by the connection relationship of each intermediate instance includes:

[0020] Along the order from the endpoint instance at the target layout position outside the layout constraint region to another endpoint instance, each intermediate instance connected in sequence is successively attributed to a relaxed instance.

[0021] A further technical solution is that the FPGA layout method further includes:

[0022] When there are multiple timing paths to be processed, each of the timing paths to be processed is processed in order from the smallest to the largest time margin.

[0023] A further technical solution is that before and after the layout constraints of the relaxed instances in the processing timing path are released, the layout position of any other instance obtained after the layout of the instances in the processing timing path is completed remains unchanged or changes.

[0024] A further technical solution is that the method for determining the timing path to be processed includes:

[0025] Parse the logic netlist and determine the subnetlist corresponding to the module to be processed, where the module to be processed is a module containing at least one instance to be processed, and the instance to be processed is an instance with a layout constraint condition that is laid out in a corresponding layout constraint region;

[0026] For each module to be processed, the layout constraint area corresponding to each instance to be processed in the module to be processed is determined, and the timing path to be processed in the module to be processed is determined according to the subnet table of the module to be processed.

[0027] The beneficial technical effects of this application are:

[0028] The present application discloses an FPGA layout method for optimizing timing by relaxing instances. The method first determines a timing path to be processed in which the layout constraints of two endpoint instances meet the situation of the present application and the time margin does not meet the timing requirements. Then, the timing path to be processed is preprocessed, and some or all of the intermediate instances in the timing path to be processed are classified as relaxed instances to first release the layout constraints, so as to correct the layout constraints of the logic netlist, so that the subsequent layout algorithm can adjust the positions of these instances according to the timing requirements, so that the timing path to be processed can meet the timing requirements through the subsequent layout algorithm correction, thereby meeting the design requirements, improving the design efficiency and optimizing the performance of the designed FPGA. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a method flow chart of an FPGA layout method according to an embodiment of the present application.

[0030] Figure 2 It is a schematic diagram of original layout constraints of four instances in a timing path to be processed in an example of the present application.

[0031] Figure 3 In one embodiment of the present application, Figure 2 The diagram shows the layout positions of four instances in the timing path to be processed before and after all the intermediate instances in the timing path to be processed are attributed to the relaxed instances and the layout constraints are released.

[0032] Figure 4 In another embodiment of the present application, Figure 2 The schematic diagram of the layout positions of four instances in the timing path to be processed is shown in FIG. 1 , before and after the instance B in the timing path to be processed is attributed to the relaxed instance and the layout constraints are released.

[0033] Figure 5 In another embodiment of the present application, Figure 2 The schematic diagram of the layout positions of four instances in the timing path to be processed is shown in FIG. 1 , before and after the instance C in the timing path to be processed is attributed to the relaxed instance and the layout constraints are released. DETAILED DESCRIPTION

[0034] The specific implementation of the present application is further described below in conjunction with the accompanying drawings.

[0035] This application discloses an FPGA layout method for optimizing timing by relaxing instances, please refer to Figure 1 The method comprises the following steps:

[0036] Step 1: parse the logic netlist and read the layout constraints of the logic netlist to determine the timing path to be processed.

[0037] The timing path in the logic netlist refers to the logic path that the data signal passes through during the propagation process in the design. Each timing path has a starting instance and an end instance, and usually also includes several intermediate instances. The data signal starts from the starting instance and passes through each intermediate instance in turn to reach the end instance. The starting instance of the timing path can be an IO input port or a clock input port of a timing unit. Here, the IO input port is sometimes specifically implemented as a data input port of a combinational logic unit. The end instance of the timing path can be an IO output port or a data input port of a timing unit. Here, the IO output port is sometimes specifically implemented as a data output port of a combinational logic unit. According to the different types of starting instances and end instances, the timing paths can be roughly divided into four categories: (1) The timing path between the IO input port and the IO output port. (2) The timing path between the clock input port of the timing unit and the IO output port. (3) The timing path between the IO input port and the data input port of the timing unit. (4) The timing path between the clock input port of the timing unit and the data input port of the timing unit. Each timing path has a corresponding time margin (slack), which refers to the difference between the actual time used and the time required for the design. It is a term that indicates whether the design meets the timing. A positive slack indicates that the timing requirements are met, and a negative slack indicates that the timing requirements are not met.

[0038] Each of the timing paths to be processed in the present application is any one of the above four categories, and the starting point instance and the end point instance of the timing path to be processed determined to have their own layout constraints: one of the endpoint instances has a layout constraint condition located within the layout constraint region, and the other endpoint instance has a layout constraint condition located at the target layout position outside the layout constraint region, and one of the two endpoint instances is the starting point instance of the timing path to be processed, and the other is the end point instance of the timing path to be processed. The intermediate instances other than the two endpoint instances in the timing path to be processed also have their own corresponding layout constraints.

[0039] In addition to the aforementioned layout constraints for each instance included in the timing path to be processed, there is another requirement that the time margin of the timing path to be processed under the layout constraints for each instance included does not meet the timing requirement.

[0040] For example, Figure 2 In the example shown, a timing path includes four instances, namely instance A, instance B, instance C, and instance D. The two endpoint instances of the timing path are instance A as the starting instance and instance D as the end instance. The timing path includes two intermediate instances, namely instance B and instance C. The layout constraint condition of instance A as the starting instance is that it needs to be placed within the layout constraint area shown in the dotted box, while the layout constraint condition of instance D as the end instance is that it needs to be placed within Figure 2 The target layout position shown in the middle shaded area is outside the layout constraint area. Instance B and Instance C, which are intermediate instances, also have their own corresponding layout constraints. In this instance, the layout constraints of the two intermediate instances also require layout within the layout constraint area shown in the dotted box. Under the layout constraints corresponding to each of the four instances, if the time margin of the timing path does not meet the timing requirements, it can be determined that the timing path is the timing path to be processed in this application, otherwise it is not a timing path to be processed.

[0041] According to the characteristics of the timing paths to be processed provided in the present application, one or more timing paths to be processed can be determined by parsing the logic netlist and layout constraints.

[0042] Considering that in practical applications, the layout method of FPGA is often implemented based on modules, where the modules mainly include Module and IP core: Module is the basic description unit and basic design unit of Verilog, and is used to describe a certain design function or structure and the external port for communication with other modules. IP core (Intellectual Property Core) is a general term for integrated circuit cores with intellectual property rights. It is a macro module with specific functions that has been repeatedly verified. It is independent of the chip manufacturing process and can be transplanted to different semiconductor processes. The layout constraints of the logic netlist often constrain some or all instances in a specific module to a specific layout position or a specific layout constraint area. Therefore, when determining the timing path to be processed, the present application first parses the logic netlist and determines the subnet table corresponding to the module to be processed. The module to be processed is a module containing at least one instance to be processed, and the instance to be processed is an instance with layout constraints that are laid out in the corresponding layout constraint area. That is, first locate the module to be processed, and then for each module to be processed, determine the layout constraint area corresponding to each instance to be processed in the module to be processed according to the above principles, and determine the timing path to be processed in the module to be processed according to the subnet table of the module to be processed. Each module to be processed is located and processed in sequence, and the timing path to be processed in each module to be processed is determined. The processing order of multiple modules to be processed can be customized.

[0043] Step 2: For any timing path to be processed, release the layout constraint conditions of the relaxed instances in the timing path to be processed so that the time margin of the timing path to be processed meets the timing requirement.

[0044] The relaxed instances in the timing path to be processed are the intermediate instances in the timing path to be processed except the two endpoint instances. In one embodiment, all the intermediate instances in the timing path to be processed are relaxed instances, or some of the intermediate instances in the timing path to be processed are relaxed instances, and the remaining intermediate instances still retain the original layout constraints.

[0045] There are two main categories of methods for determining slack instances in the timing path to be processed:

[0046] In the first method, all intermediate instances in the timing path to be processed are directly used as relaxed instances. After all the intermediate instances are used as relaxed instances to release the layout constraints, the time margin of the timing path to be processed can generally always meet the timing requirements.

[0047] For example, Figure 2 Taking the timing path to be processed as an example, based on the first method, both instance B and instance C are directly classified as relaxed instances and the layout constraints are released. Instance B and instance C can be placed outside the layout constraint area later. Please refer to Figure 3 , which shows a schematic diagram of the layout after the instances in the timing path to be processed are laid out according to the original layout constraints of the four instances, and a schematic diagram of the layout after the four instances in the timing path to be processed are laid out after the layout constraints of instance B and instance C are released. Figure 3 It can be seen that instance D as an endpoint instance is still placed at the target placement position according to its own placement constraints, and instance A as an endpoint instance is still placed in the placement constraint area according to its own placement constraints. According to the original placement constraints, instance B and instance C are also placed in the placement constraint area. After releasing the placement constraints of instance B and instance C, subsequent instances B and instance C can be placed outside the placement constraint area, so that the time margin of the timing path to be processed meets the timing requirements.

[0048] It should be noted that in this method, all intermediate instances are directly classified as relaxed instances to release the layout constraints, but in reality, it may only be necessary to release the layout constraints of some intermediate instances to make the time margin of the timing path to be processed meet the timing requirements. Therefore, the layout position of any relaxed instance after releasing the layout constraints may not meet its original layout constraints or still meet its original layout constraints. For example, Figure 3 In the example shown, it is also possible that the final layout position of instance C is still within the layout constraint area.

[0049] In the second method, each intermediate instance in the timing path to be processed is classified as a relaxed instance in a predetermined order until the time margin of the timing path to be processed meets the timing requirement, and the layout constraints of the remaining intermediate instances are retained. In this method, some intermediate instances may be classified as relaxed instances, or all intermediate instances may be classified as relaxed instances.

[0050] In the second method described above, the predetermined order for attributing each intermediate instance to a relaxed instance mainly includes the following methods:

[0051] (1) The intermediate instances are assigned to relaxed instances in descending order of their influence on the timing margin of the timing path to be processed. The greater the influence of an intermediate instance on the timing margin of the timing path to be processed, the more likely it is that the timing margin of the timing path to be processed can meet the timing requirements after the layout constraints are released by assigning the intermediate instance to the relaxed instance.

[0052] Based on this processing principle, in one embodiment, the influence index of the transmission path between any two instances is first determined to be d*w, where the instances include intermediate instances and two endpoint instances. Where d is the distance of the transmission path between the two instances, and w is the timing criticality of the transmission path between the two instances. After determining the influence index of each transmission path, the intermediate instances connecting the two ends of each transmission path are classified as relaxed instances in descending order of the influence index values, until the time margin of the timing path to be processed meets the timing requirements, and the subsequent intermediate instances are no longer processed.

[0053] For example, Figure 2 As an example of the timing path to be processed shown in FIG. 1 , based on the method provided in this embodiment, the transmission path L between instance A and instance B is calculated. AB The influence indicator Q AB =d AB *w AB , the transmission path L between instance B and instance C BC The influence indicator Q BC =d BC *w BC , the transmission path L between instance C and instance D CD The influence indicator Q CD =d CD *w CD . Transmission path L AB The distance is d AB , the timing criticality is w AB , transmission path L BC The distance is d BC , the timing criticality is w BC , transmission path L CD The distance is d CD , the timing criticality is w CD .

[0054] Assuming the calculated Q AB >Q BC >Q CD , then firstly set the transmission path L ABThe intermediate instance connected at both ends is classified as a relaxed instance. AB Only one side is connected to an intermediate instance, which is instance B. Therefore, instance B is first classified as a relaxed instance. Assuming that the time margin of the timing path to be processed can meet the timing requirements at this time, the original layout constraints of instance C are retained.

[0055] Please refer to Figure 4 , which shows a schematic diagram of the layout after the instances in the timing path to be processed are laid out according to the original layout constraints of the four instances, and a schematic diagram of the layout after the four instances in the timing path to be processed are laid out after the layout constraints of instance B are released. Figure 4 It can be seen that instance D, as an endpoint instance, is still placed at the target placement position according to its own placement constraints, and instance A, as an endpoint instance, is still placed in the placement constraint region according to its own placement constraints. According to the original placement constraints, instances B and C are also placed in the placement constraint region, and after the placement constraints of instance B are released, instance B is placed outside the placement constraint region, and instance C, which is not classified as a relaxed instance, is still placed in the placement constraint region according to the original placement constraints, so that the time margin of the timing path to be processed meets the timing requirements.

[0056] (2) The intermediate instances in the timing path to be processed are sequentially classified as relaxed instances in the order determined by the connection relationship of the intermediate instances.

[0057] Based on this processing principle, reverse tracing is generally started from the endpoint instance outside the layout constraint area, that is, along the order from the endpoint instance at the target layout position outside the layout constraint area to another endpoint instance, each intermediate instance connected in sequence is successively attributed to a relaxed instance until the time margin of the timing path to be processed meets the timing requirements, and the subsequent intermediate instances are no longer processed.

[0058] For example, Figure 2 Taking the timing path to be processed as an example, based on the method provided in this embodiment, reverse tracing starts from instance D, and instance C is first classified as a relaxed instance. If the time margin of the timing path to be processed meets the timing requirements at this time, the subsequent instance B will not be processed. If the time margin of the timing path to be processed still does not meet the timing requirements at this time, the next intermediate instance, that is, instance B, will be classified as a relaxed instance in the connection order.

[0059] Figure 5 For example, only instance C is classified as a relaxed instance. Figure 5The following is a schematic diagram showing the layout of the instances in the timing path to be processed after the original layout constraints of the four instances are laid out, and a schematic diagram showing the layout of the four instances in the timing path to be processed after the layout constraints of instance C are released. Figure 5 It can be seen that instance D, as an endpoint instance, is still placed at the target placement position according to its own placement constraints, and instance A, as an endpoint instance, is still placed in the placement constraint region according to its own placement constraints. According to the original placement constraints, instances B and C are also placed in the placement constraint region, and after the placement constraints of instance C are released, instance C is placed outside the placement constraint region, and instance B, which is not classified as a relaxed instance, is still placed in the placement constraint region according to the original placement constraints, so that the time margin of the timing path to be processed meets the timing requirements.

[0060] Regardless of which of the above methods is used to determine the relaxed instances in the timing path to be processed, before and after the placement constraints of the relaxed instances in the timing path to be processed are released, the placement positions of any other instances obtained after the placement of the instances in the timing path to be processed remain unchanged or change, including: (1) before and after the placement constraints of the relaxed instances are released, one of the endpoint instances of the timing path to be processed is always placed at the target placement position outside the placement constraint area according to the placement constraints, such as Figure 3-Figure 5 For instance D in the example, since the layout constraint of the endpoint instance remains unchanged, the layout position remains unchanged. (2) Before and after the layout constraint of the relaxed instance is released, the other endpoint instance of the timing path to be processed is always placed in the layout constraint area according to the layout constraint, but the layout position of the endpoint instance in the layout constraint area is the same or different before and after the layout constraint of the relaxed instance is released. Figure 3-Figure 5 For instance A in the example, since the layout constraint of the endpoint instance remains unchanged, instance A is always placed in the layout constraint area. However, the layout position of instance A in the layout constraint area can change, such as Figure 3 and Figure 4 As shown. Alternatively, the layout position of instance A in the layout constraint area can remain unchanged, as shown Figure 5 As shown. (3) Before and after the layout constraints of the relaxed instances are released, the layout positions of any intermediate instances in the to-be-processed timing path that are not attributed to the relaxed instances can be the same or different, which is similar to case (2) and will not be repeated here. (4) More specifically, when all intermediate instances are attributed to the relaxed instances, there may be some relaxed instances whose layout positions are the same before and after the layout constraints are released, as described above and will not be repeated here.

[0061] The above embodiment discloses a method for preprocessing each timing path to be processed to release the layout constraint of the relaxation instance. When the logic netlist is parsed to determine that there are multiple timing paths to be processed, in one embodiment, each timing path to be processed can be preprocessed in sequence in a random order. Or in another embodiment, each timing path to be processed can be processed in sequence in order from small to large time margin.

[0062] Step 3, after completing the processing of all the timing paths to be processed through the above step 2, the original layout constraints of the logic netlist can be corrected. In the revised layout constraints of the logic netlist, the two endpoint instances of each timing path to be processed still have their original layout constraints, the intermediate instances of the timing path to be processed that are not attributed to the relaxed instances still have their original layout constraints, and the intermediate instances attributed to the relaxed instances no longer have layout constraints. Then, the instances in the logic netlist are laid out under the revised layout constraints of the logic netlist. Subsequently, operations such as wiring processing, timing analysis, and code stream generation are continued to be completed, and the entire design process is finally completed.

[0063] The above is only a preferred embodiment of the present application, and the present application is not limited to the above embodiments. It is understood that other improvements and changes directly derived or associated by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the protection scope of the present application.

Claims

1. An FPGA layout method for optimizing timing by relaxing instances, characterized in that: The FPGA layout method comprises: Parsing a logic netlist and reading layout constraints of the logic netlist to determine a timing path to be processed, wherein one endpoint instance in the timing path to be processed has a layout constraint condition located within a layout constraint region, another endpoint instance has a layout constraint condition located at a target layout position outside the layout constraint region, and the timing margin of the timing path to be processed under the layout constraints of each instance contained does not meet the timing requirement; For any one of the timing paths to be processed, release the layout constraints of the relaxed instances in the timing path to be processed so that the time margin of the timing path to be processed meets the timing requirements, and the relaxed instances are the middle instances in the timing path to be processed except the two endpoint instances; Complete placement of instances in the logic netlist under the modified placement constraints of the logic netlist; The method for determining a timing path to be processed includes: parsing a logic netlist and determining a subnetlist corresponding to a module to be processed, wherein the module to be processed is a module containing at least one instance to be processed, and the instance to be processed is an instance with layout constraints that are laid out into a corresponding layout constraint region; for each module to be processed, determining the layout constraint region corresponding to each instance to be processed in the module to be processed, and determining the timing path to be processed in the module to be processed according to the subnetlist of the module to be processed.

2. The FPGA layout method according to claim 1, characterized in that: The method for determining the relaxation instance in the timing path to be processed includes: Directly taking all the intermediate instances in the to-be-processed timing path as relaxed instances; Alternatively, each intermediate instance in the to-be-processed timing path is classified as a relaxed instance in a predetermined order, until the time margin of the to-be-processed timing path meets the timing requirement, and the layout constraints of the remaining intermediate instances are retained.

3. The FPGA layout method according to claim 2, characterized in that: The method of attributing each intermediate instance in the to-be-processed timing path as a relaxed instance in a predetermined order comprises: The intermediate instances are classified as relaxed instances in descending order of influence on the timing margin of the timing path to be processed.

4. The FPGA layout method according to claim 3, characterized in that: The method of classifying each intermediate instance as a relaxed instance in descending order of influence on the timing margin of the timing path to be processed includes: Determine the influence index of the transmission path between any two instances as d*w, where d is the distance of the transmission path between the two instances and w is the timing criticality of the transmission path between the two instances; According to the order of influence index values ​​from large to small, the intermediate instances connecting the two ends of each transmission path are classified as relaxed instances.

5. The FPGA layout method according to claim 2, characterized in that: The method of attributing each intermediate instance in the to-be-processed timing path as a relaxed instance in a predetermined order comprises: The intermediate instances in the to-be-processed timing path are sequentially classified as relaxed instances in the order determined by the connection relationship of the intermediate instances.

6. The FPGA layout method according to claim 5, characterized in that: The method of sequentially assigning each intermediate instance to a relaxed instance according to the order determined by the connection relationship of each intermediate instance includes: Along the sequence from the endpoint instance at the target layout position outside the layout constraint region to another endpoint instance, each intermediate instance connected in sequence is successively attributed to a relaxed instance.

7. The FPGA layout method according to claim 1, characterized in that: The FPGA layout method further includes: When there are multiple timing paths to be processed, each timing path to be processed is processed in sequence from small to large time margin.

8. The FPGA layout method according to claim 1, characterized in that: Before and after the layout constraint condition of the relaxed instance in the to-be-processed timing path is released, the layout position of any other instance obtained after the layout of the instance in the to-be-processed timing path is completed remains unchanged or changes.

Citation Information

Patent Citations

  • FPGA chip global placement optimization method based on temporal constraint

    CN105808795A

  • Looseness weight time sequence driving layout method

    CN113191107A

  • Time sequence drive layout method and device, equipment and storage medium

    CN114386352A

  • Time sequence optimization method and system for FPGA (Field Programmable Gate Array)

    CN115099177A

  • Solving constraint satisfaction problems using a field programmable gate array

    US20150365092A1