A method for screening critical timing path sets with aging effects

Through the scoring and sorting mechanism combined with static timing analysis and SPICE unit feature library, the key timing paths after circuit aging are screened out, solving the problem of aging performance evaluation in the existing technology, and improving the reliability and timing monitoring efficiency of the circuit.

CN115017847BActive Publication Date: 2024-12-06NANJING HUADA JIUTIAN TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210601606.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-30
Publication Date
2024-12-06
Estimated Expiration
2042-05-30

AI Technical Summary

Technical Problem

The prior art is difficult to quickly and accurately estimate and analyze the aging performance of the entire circuit, especially the impact of negative bias temperature instability and hot carrier injection effect on timing problems, resulting in degradation of circuit reliability.

Method used

Using the method based on static timing analysis and SPICE unit timing feature library, the critical timing path with the worst Slack and larger Degradation after aging is screened through the scoring and sorting mechanism, combined with the timing relaxation and degradation of the timing path, time sequence monitoring and optimization.

Benefits of technology

It realizes the rapid and accurate screening of key timing paths after circuit aging, improves the efficiency and reliability of timing evaluation, and ensures the stability and performance monitoring of the circuit in the aging situation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115017847B_ABST
    Figure CN115017847B_ABST
Patent Text Reader

Abstract

The present invention provides a method for screening a set of critical timing paths with aging effects, comprising: using STA and SPICE unit timing feature library lookup tables to obtain timing feature data with aging effects; using a scoring and sorting mechanism to sort the timing feature data with aging effects to obtain a scoring and sorting order of the timing feature data with aging effects; based on the scoring and sorting order of the timing feature data with aging effects, taking into account the timing relaxation amount and timing degradation amount of the timing path to perform critical path screening to obtain a set of critical timing paths. The method for screening a set of critical timing paths with aging effects of the present invention does not require complex calculations during the data screening process, and can quickly converge to obtain a reliable and unique set of critical timing paths.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of integrated circuit automated design, and in particular to a method for screening a critical timing path set with aging effects. Background Art

[0002] As the design size of transistors continues to decrease, their aging effects become more and more significant, making very large scale integration (VLSI) also face huge challenges of reliability degradation. Studies have shown that negative bias temperature instability (NBTI), hot carrier injection (HCI), time dependent dielectric breakdown (TDDB) and electron migration (EM) can cause transistor parameters to shift, which may eventually lead to circuit equipment failure. Among them, NBTI mainly exists in PMOS transistors, while HCI exists in both PMOS and NMOS transistors. They have negative effects on circuit timing and reliability issues. How to quickly and accurately estimate and analyze the aging performance of the entire circuit is a problem that the industry urgently needs to solve. Summary of the invention

[0003] In order to address the deficiencies in the prior art, the present invention aims to provide a method for screening a set of critical timing paths with aging effects, based on back-labeling the delay of timing units with aging effects back to traditional static timing analysis results (Static Timing Analysis, STA), a timing path screening method that takes into account both the timing slack (Slack) and timing degradation (Degradation) of the timing path, aims to screen out the critical timing paths with large degradation under the worst Slack condition after circuit aging, and then deploy timing monitoring or auxiliary timing optimization.

[0004] To achieve the above object, the present invention provides a method for screening a critical timing path set with aging effects, comprising:

[0005] Use STA and SPICE unit timing feature library lookup table to obtain timing feature data with aging effect;

[0006] Sorting the time series feature data with aging effect by using a scoring and sorting mechanism to obtain a scoring and sorting order of the time series feature data with aging effect;

[0007] Based on the scoring and sorting order of the timing characteristic data with aging effect, the critical paths are screened by taking into account the timing slack and timing degradation of the timing paths to obtain a set of critical timing paths.

[0008] Furthermore, the step of using STA and SPICE unit timing feature library lookup table to obtain timing feature data with aging effect also includes:

[0009] Calculate cell delay by interpolation based on a lookup table of a cell feature library;

[0010] estimating timing slack based on the unit delay;

[0011] estimating data arrival time and data demand time based on the unit delay;

[0012] Calculating the aging timing degradation amount according to the timing relaxation amount;

[0013] Based on the data arrival time and the data demand time, calculating a ratio of a data arrival time degradation amount to a data demand time degradation amount;

[0014] Calculate the absolute difference between the number of data arrival path units and the number of required path units.

[0015] Furthermore, the step of calculating the cell delay by interpolation based on the lookup table of the cell feature library also includes: taking the timing path as the object, and obtaining the initial cell delay and the aged cell delay contained in the path by interpolation based on the lookup table of the cell feature library.

[0016] Furthermore, the step of estimating the timing relaxation amount based on the unit delay also includes: using the unit delay back-labeling method to back-label the initial unit delay and the aged unit delay contained in the path to the STA data result of the timing path to obtain the initial timing relaxation amount and the aged timing relaxation amount.

[0017] Furthermore, the step of estimating the data arrival time and the data demand time based on the unit delay also includes: using the unit delay back-labeling method to back-label the unit delays of the data arrival path and the data demand path into the STA data results of the timing path to obtain the initial data arrival time, the aged data arrival time, the initial data demand time and the aged data demand time.

[0018] Furthermore, the step of calculating the ratio of the data arrival time degradation amount to the data demand time degradation amount based on the data arrival time and the data demand time further includes:

[0019] Based on the data arrival time and the data demand time, the data arrival time degradation amount and the data demand time degradation amount are calculated respectively;

[0020] The ratio of the data arrival time degradation amount to the data demand time degradation amount is taken as the ratio of the data arrival time degradation amount to the data demand time degradation amount.

[0021] Furthermore, the step of calculating the absolute difference between the number of data arrival path units and the number of required path units further includes:

[0022] The number of sequential units included in the data arrival and demand paths is counted respectively, and the absolute difference between the number of sequential units included in the data arrival path and the number of sequential units included in the data demand path is taken as the absolute difference between the number of data arrival path units and the number of data demand path units.

[0023] Furthermore, the step of sorting the time series feature data with aging effect by using a scoring and sorting mechanism to obtain a scoring and sorting order of the time series feature data with aging effect further includes:

[0024] Scoring and sorting the multiple groups of time series relaxation amount associated data to obtain the scoring and sorting order of the multiple groups of time series relaxation amount associated data;

[0025] Scoring and sorting the multiple groups of time series degradation amount associated data to obtain the scoring and sorting order of the multiple groups of time series degradation amount associated data.

[0026] Furthermore, the step of scoring and sorting the multiple groups of timing slack amount associated data to obtain the scoring and sorting order of the multiple groups of timing slack amount associated data also includes:

[0027] In the original timing path set, the paths are arranged in ascending order according to the aging timing relaxation amount, which serves as the main sorting of the original timing path set;

[0028] On the basis of the primary sorting, the initial timing slack and the STA timing slack are sorted in ascending order to obtain a secondary sorting of the original timing path set;

[0029] The size of the original timing path set is calculated as N, where N is a positive integer, and each timing path is given an initial score according to the order in the main sorting;

[0030] Traverse the primary sorting to the i-th timing path, whose initial score is Ni, where i is a positive integer and N is a positive integer, and adjust the score according to the index position of the timing path in the secondary sorting;

[0031] The score ranking order of the timing relaxation amount associated data is obtained in descending order of the scores.

[0032] Furthermore, the step of scoring and sorting the multiple groups of time series degradation amount associated data to obtain the scoring and sorting order of the multiple groups of time series degradation amount associated data also includes:

[0033] In the original timing path set, the paths are first arranged in descending order according to the absolute values ​​of the aging timing degradation amounts, as the main sorting of the original timing path set;

[0034] On the basis of the primary sorting, the ratio of the data arrival time degradation amount to the data demand time degradation amount is sorted in order from large to small or from small to large with reference to the timing check type, and the absolute difference between the number of data arrival path units and the number of data demand path units is sorted in order from large to small, so as to obtain the secondary sorting of the original timing path set;

[0035] The size of the original timing path set is calculated as N, where N is a positive integer, and each timing path is given an initial score according to the order of the main sorting;

[0036] Traverse the primary sorting to the i-th timing path, whose initial score is Ni, where i is a positive integer and N is a positive integer, and adjust the score according to the index position of the timing path in the secondary sorting;

[0037] The score ranking order of multiple groups of timing degradation amount correlation data is obtained in descending order of scores.

[0038] Furthermore, the step of performing critical path screening based on the scoring and sorting order of the timing feature data with aging effect and taking into account the timing slack and timing degradation of the timing path to obtain a critical timing path set also includes:

[0039] Hybrid screening of critical paths that takes into account both timing slack and timing degradation;

[0040] Critical path compensation screening that considers both timing slack and timing degradation.

[0041] Furthermore, the step of hybrid screening of critical paths taking into account both timing slack and timing degradation also includes: recording the size of the timing path set to be selected as M, and N>M>0; performing hybrid screening based on the scoring and sorting order of the timing slack and timing degradation associated data, traversing the timing paths ranked in the top M positions in the scoring and sorting order of the timing slack associated data, and if the timing path is also ranked in the top M positions in the scoring and sorting order of the timing degradation associated data, taking it out as the critical timing path; if M timing paths are selected, the selection task is completed; otherwise, assuming that only K timing paths are selected, and N>M>K>0, MK timing paths are selected from the remaining NK timing path set for supplementation.

[0042] Furthermore, the step of supplementary screening of critical paths taking into account both timing slack and timing degradation also includes:

[0043] The aged timing degradation is taken out from the remaining NK timing path sets and added to the STA timing slack as the estimated timing slack. The paths are arranged in ascending order, and the timing paths ranked in the first MK positions are taken to complete the supplement of the key timing path set. The screening is ended and a key timing path set with a size of M and unique elements is selected.

[0044] To achieve the above objectives, the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a program running on the processor, and when the processor runs the program, the steps of the method for screening a critical timing path set with aging effects as described above are executed.

[0045] To achieve the above object, the present invention further provides a computer-readable storage medium having computer instructions stored thereon, which, when executed, execute the steps of the method for screening a critical timing path set with aging effects as described above.

[0046] Compared with the prior art, the method for screening the critical timing path set with aging effect of the present invention has the following beneficial effects:

[0047] 1) This patent does not use methods such as solving equations or matrix decomposition with high computational complexity, nor does it use machine learning methods that require specific feature extraction and a large number of sample training. This patent only uses simple data calculation, sorting and combination methods to complete the timing data estimation and screening tasks. In the process of timing data estimation, it fully combines the advantages of STA and SPICE unit timing feature libraries for estimation, which speeds up the timing evaluation of timing paths with aging effects; in the process of timing path screening, it integrates reliable and mature timing analysis experience for optimization, ensuring that it can quickly converge to obtain a set of key timing paths with aging effects as reference data for circuit timing monitoring or optimization;

[0048] 2) In the process of obtaining the Slack and Degradation ranking of the time series path, multiple groups of related time series data are combined and utilized, and the primary and secondary are divided based on the experience of time series analysis. An easy-to-interpret and scalable comprehensive scoring mechanism is designed. This method is more accurate and stable than sorting by a single data. First, the various data required in the comprehensive scoring mechanism are combined and matched according to the purpose requirements; secondly, the score adjustment is completed by using the secondary data ranking, and the setting of the score adjustment step is flexible and controllable. The smaller the step, the smaller the impact of the substituted data;

[0049] 3) Use a fast-converging method to screen the critical timing path set with aging effects. First, use the comprehensive scoring and sorting of Slack and Degradation to complete rigorous mixed screening, ensuring that the Slack of the selected timing path still has a large Degradation in the worst case. The order of Slack and Degradation of the remaining unselected timing paths is no longer strictly matched. At this time, the Slack Degradation in the aging process is superimposed on the STA Slack for estimation and sorting, and the timing paths with the highest ranking are directly selected to quickly complete compensation. The entire data screening solution does not require complex calculations and can quickly converge to obtain a reliable and unique set of critical timing paths.

[0050] Other features and advantages of the present invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0052] Figure 1 A flow chart of a method for screening a critical timing path set with aging effects according to the present invention;

[0053] Figure 2 A schematic diagram of the estimation and calculation process of the time series data with aging effect according to the present invention;

[0054] Figure 3 A schematic diagram of the timing relaxation score sorting according to the present invention;

[0055] Figure 4 A schematic diagram of scoring and sorting of timing degradation according to the present invention;

[0056] Figure 5 FIG. 1 is a schematic diagram of critical path screening with aging effect according to the present invention. DETAILED DESCRIPTION

[0057] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0058] Timing slack (Slack): refers to the difference between the arrival delay of the timing path and the required delay. If the difference is a negative value, it means that the delay of the timing path does not meet the requirements of the timing constraints. When the Slack of the timing path is negative and the value is large, the timing repair of the timing path is more difficult. At this time, the timing path is called a critical path during the timing path analysis process.

[0059] Timing degradation: The degradation of circuit performance caused by transistor aging causes some timing parameters of the timing path to change to varying degrees. It is generally calculated by the difference in timing data before and after aging.

[0060] Data Arrival Time: refers to the actual propagation delay of the timing path. The delay is calculated based on the specific line delay, unit delay and other information and the timing path delay calculation method.

[0061] Required Time: To meet the timing requirements, the delay of the timing path must not exceed the maximum delay limit. This limit value is used to indicate that the timing constraints can be met when the actual delay value is less than or equal to the required delay value.

[0062] Cell Delay: A logic gate composed of transistors can perform logical evaluation on high and low level input signals and generate corresponding high and low level outputs. The time interval from the change of the input signal to the corresponding correct output signal is the logic gate cell delay.

[0063] Data Arrival Path: The timing path segment for calculating data arrival time; Data Required Path: The timing path segment for calculating data requirement time.

[0064] The present invention aims to quickly predict and select key timing paths that have timing violation facts and risks after circuit aging, and the characteristics that should be possessed include the worst Slack and large Degradation. Based on the existing research on back-labeling the aging delay of the timing unit to STA, the advantages of short simulation time of the unit feature library, low cost of characterization library construction, full coverage of STA timing paths, simple and fast calculation speed, etc. can be utilized. Through the look-up table (Look-up Table, LUT) and timing calculation, special timing feature data with aging effect can be obtained, including: timing slack (Slack), data arrival time (Data Arrival Time), data required time (Data Required Time) and cell delay (CellDelay), etc. and their respective degradation amounts, the actual physical meaning of these data and the orderly sequence of the data are integrated to screen out the key timing paths with aging effect to guide the design work such as timing monitoring or timing optimization.

[0065] In the embodiment of the present invention, firstly, the timing estimation with aging effect is completed by combining STA and SPICE unit timing feature library table lookup: STA has the characteristics of full coverage, simple calculation and fast speed, and the timing characterization completed by simulating the timing unit through SPICE simulation is more time-saving than the simulation of the timing path, and the unit delay with aging effect is estimated by looking up the feature library table, and the timing feature data with aging effect is obtained by back-labeling STA, including Slack, Degradation, etc.; secondly, the scoring and sorting mechanism of multiple groups of feature data is integrated to obtain a more reliable and stable special data sorting order: because the timing feature data estimation model is simplified, it is actually more time-saving than the timing path simulation. The SPICE simulation results of the critical path are worse, so it is unreasonable to rely on a single data sorting result as a reference. It is necessary to integrate as many related data as possible to update the special scores of the data, and determine the reliable special data sorting order according to the scores; finally, a fast-converging data screening scheme is designed to obtain a critical path data set with aging effects: a reliable and stable special data sorting order is obtained by integrating the comprehensive scoring and sorting of multiple groups of data, and a part of the critical timing paths can be quickly determined through a simple processing method similar to the set intersection operation; in addition, a set of timing path compensation screening mechanism needs to be designed to ensure that the final critical path set is unique and the number of elements meets the requirements.

[0066] Figure 1 The flowchart of the method for screening the critical timing path set with aging effect of the present invention is as follows. Figure 1 , the method for screening the critical timing path set with aging effect of the present invention is described in detail.

[0067] In the embodiment of the present invention, taking the setup time check (Setup Check) of the circuit as an example, the circuit aging time is recorded as X years.

[0068] First, in step 101, the STA and SPICE unit timing feature library lookup table is used to obtain the timing feature data with aging effect: the STA tool extracts the timing path of the design circuit and calculates the corresponding STA Slack, and the SPICE simulator simulates the timing unit of the design circuit to complete the timing characterization with aging effect, and the two results are combined to quickly estimate the Slack data and its Degradation data after 0 years and aging, as well as the associated timing feature data.

[0069] In the embodiment of the present invention, the timing characteristic data with aging effect is obtained by using the STA and SPICE unit timing characteristic library lookup table, which specifically includes:

[0070] 1) Cell Delay lookup calculation: Taking the timing path as the basic object, bilinear timing interpolation calculation is performed based on the lookup table of the cell feature library to obtain the cell delays contained in the path of 0 years and aged X years, which are recorded as Fresh Cell Delay and Aged Cell Delay. The cell feature library required for the lookup interpolation calculation is obtained by SPICE gate-level simulation netlist through SPICE simulation, and the input conditions such as transition time and output load required for the interpolation calculation need to consider the deviation caused by the STA result and the transmission process. According to the sequence of logic gate units on the timing path, the transition output by the previous level cell lookup calculation is used as the transition input by the next level cell lookup calculation, and the load in STA is taken as the load input by the cell lookup calculation. In addition, the disturbance and deviation of error propagation are considered during the timing transmission process, and the input parameters are corrected in time;

[0071] 2) Timing slack estimation: Use the cell delay back-annotation method to back-annotate the Fresh Cell Delay and Aged Cell Delay obtained in 1) to the STA data result of the timing path (essentially replacing the original STA Cell Delay), and estimate the 0-year and X-year aged Slack, recorded as Fresh Slack and AgedSlack. Note that the clock reconvergence pessimism removal (CRPR) on the common path needs to be solved to ensure that the delay on the common path remains consistent without difference. This process regards the aging effect of cell delay as the most important and main timing influencing factor, ignores the aging effect of interconnect delay, and effectively simplifies the timing estimation model;

[0072] In the embodiment of the present invention, Slack and Degradation are both caused by unit delays on non-common paths, and CRPR is considered to ensure that the delay amount on the common path is the same as the delay amount on the data arrival path and the data demand path.

[0073] 3) Data arrival time and data required time (Data Arrival Time and Data Required Time) estimation. According to the concepts of data arrival time and data required time, the size of Slack data is closely related to the relative sizes of these two sets of data. Using the same Cell Delay back-labeling method in 2), the data arrival time and data required time of 0 years and aged X years are estimated, recorded as the fresh data arrival and required time (Fresh Data Arrival and Required Time) and the aged data arrival and required time (Aged Data Arrival and Required Time), where the fresh or aged data arrival and data required time are the results of the estimation after adding CRPR compensation;

[0074] 4) Slack degradation calculation. Calculate the degradation of Slack data from 0 years to X years after aging, that is, Aged Slack-Fresh Slack, recorded as Slack Degradation or Degradation. Its value is generally negative, indicating that the circuit aging process makes Slack worse to varying degrees. The order of its absolute value can measure the amount of degradation;

[0075] 5) Calculate the ratio of the data arrival time degradation to the required time degradation. Based on the results in 3), calculate the degradation of Data Arrival and Required Time, namely Aged Data Arrival Time-Fresh Data Arrival Time, recorded as Data Arrival Time Degradation (DATD), and Aged Data Required Time-Fresh Data Required Time, recorded as Data Required Time Degradation (DRTD). DATD and DRTD data are always positive values. The ratio of the two sets of data, DATD / DRTD, is the ratio of data arrival and data required time degradation. This ratio value can characterize the relative difference in timing degradation on the data arrival path and the data required path.

[0076] 6) Count the absolute difference between the number of cells in the data arrival and data requirement paths. Studies have shown that a longer data arrival or data requirement path will accumulate more aging effects, making the timing degradation more significant. Count the number of timing cells contained in the data arrival and data requirement paths in the same timing path, respectively, and record them as Data Arrival Path Cell Count and Data Required Path Cell Count. The absolute difference between the two numbers |Data ArrivalPath Cell Count-Data Required Path Cell Count| is the absolute difference between the number of cells in the data arrival and data requirement paths. This absolute difference can also reflect the difference in the degree of timing degradation on the data arrival and data requirement paths.

[0077] In step 102, the acquired time series feature data is scored and sorted to obtain a scoring and sorting order of the time series feature data.

[0078] In the embodiment of the present invention, the multiple groups of Slack-related data and the multiple groups of Degradation-related data are scored and sorted respectively.

[0079] In the embodiment of the present invention, Slack-related data includes: Aged Slack, Fresh Slack and STA Slack; Degradation-related data includes: Slack Degradation absolute value, data arrival and data demand time degradation ratio and data arrival and data demand path unit number absolute difference.

[0080] 1) Comprehensive scoring and sorting of multiple groups of Slack-related data. The degradation of interconnect delay is ignored in Slack data estimation, and the propagation and deviation of the signal on the timing path cannot be fully reproduced. Therefore, it is necessary to comprehensively analyze multiple groups of Slack-related data to predict the order of Slack after aging. In the original timing path set P, the Aged Slack is first arranged in order from small to large, as the main sorting SS1 of the set P. On the basis of the main sorting SS1, the Fresh Slack and STASlack are sorted in order from small to large, respectively, to obtain the secondary sorting SS2 and SS3 of the set P. The size of the original timing path set is N, and each timing path is given an initial score {N-1, N-2, ..., 0} according to the order in the main sorting SS1. The secondary sorting SS2 and SS3 are mainly used to adjust the score. Traverse the main sorting SS1 to the i-th timing path p i , whose initial score is Ni, according to the timing path p i Index position index in sub-sorts SS2 and SS3 SS2 and indexSS3 To adjust the score, as shown in formula 1:

[0081]

[0082] Finally, the score ranking order S of multiple groups of Slack related data is obtained in descending order. slack .

[0083] 2) Comprehensive scoring and sorting of multiple groups of Degradation-related data. Similarly, it is also necessary to comprehensively analyze multiple groups of Degradation-related data to predict the order of Degradation after aging. In the original timing path set P, first arrange them in order from large to small according to the absolute value of SlackDegradation, as the main sorting DS1 of the set P. On the basis of the main sorting DS1, the ratio of the data arrival and data demand time degradation is referred to the timing check type, and sorted in order from large to small or from small to large, and the absolute difference of the number of data arrival and data demand path units is sorted in order from large to small, and the secondary sorting DS2 and DS3 of the set P are obtained. Similarly, each timing path is given an initial score {N-1, N-2, ..., 0} in the order of the main sorting DS1. The secondary sorting DS2 and DS3 are mainly used to adjust the score. Traverse from the main sorting DS1 to the i-th timing path p i , whose initial score is Ni, according to the timing path p i Index position index in the sub-sorts DS2 and DS3 DS2 and index DS3 To adjust the score, as shown in formula 2:

[0084]

[0085] Finally, the score ranking order S of multiple groups of Degradation related data is obtained according to the order of scores from large to small. degradation .

[0086] Different score adjustment strategies are used in Formula 1 and Formula 2. Formula 1 intentionally weakens the influence of the STA Slack secondary ranking data, while Formula 2 regards the influence of all secondary ranking data as the same. At the same time, the score adjustment step size can be set according to experience, and its value should not be too large, otherwise it will seriously disrupt the score order of the main ranking data.

[0087] In step 103, based on the scoring and sorting order of the timing feature data, the key paths of Slack and Degradation are screened to obtain a set of key timing paths.

[0088] In the embodiment of the present invention, the hybrid screening of critical paths taking into account both Slack and Degradation includes: the size of the timing path set to be selected is M, and N>M>0. slack and S degradation Perform mixed screening and traverse the scoring and sorting S slack The timing path ranked in the top M positions in the order of S degradation If the ranking in is also in the top M, it will be taken out as the key timing path. If M timing paths are selected according to the mixed screening strategy, the selection task is completed; otherwise, assuming that only K timing paths are selected, and N>M>K>0, then it is necessary to select the remaining NK timing paths in the set P N-K Select MK timing paths from the above to supplement.

[0089] In the embodiment of the present invention, the critical path compensation screening taking into account both Slack and Degradation includes: N-K , take STA Slack + Slack Degradation as Evaluated Slack, and arrange them in order from small to large, take the timing path ranked in the first MK position to complete the supplement of the key timing path set, and end the screening.

[0090] By using the mixed screening of key paths that take into account both Slack and Degradation and the compensated screening of key paths that take into account both Slack and Degradation (i.e., the collaborative screening strategy of key paths that take into account both Slack and Degradation-specific associated data), we can quickly select a key timing path set P of size M with unique elements. critical .

[0091] Figure 2 Schematic diagram of the estimation and calculation process of the timing data with aging effect according to the present invention. Figure 2 As shown in the figure, SPICE Cell Libs is the SPICE cell timing aging feature library, Δ(Cell Count) represents the absolute difference between the number of cells in the data arrival and data demand paths, Degradation represents the Slack degradation, and Rate (ΔA / ΔR) represents the ratio of the data arrival and data demand time degradation. As can be seen from the process in the figure, combining STA timing analysis and SPICE cell feature library table lookup calculation can quickly estimate the Slack of each timing path in 0 years and X years of aging, and it is also easy to statistically or calculate the associated data of each group of Degradation.

[0092] Figure 3This is a schematic diagram of the timing relaxation score sorting according to the present invention. Figure 4 FIG. 1 is a schematic diagram of the scoring and sorting of the timing degradation amount according to the present invention, as shown in FIG. Figure 3 and Figure 4 The comprehensive scoring sorting mechanism of Slack and Degradation is shown respectively. Their ideas and methods are similar, and the differences are mainly reflected in different input data and their sorting behaviors, as well as different score adjustment step settings. The comprehensive scoring sorting mechanism is easy to explain and scalable. The master sorting data (Master Sort), as the most influential data, determines the initial score of the special data. Under a reasonable empirical background, more secondary sorting data can be integrated and combined, and the special score of the data can be updated through different judgment conditions and adjustment steps. This method is more reliable and stable than the results of relying on single data sorting.

[0093] Figure 5 The schematic diagram of the critical path screening with aging effect according to the present invention is as follows: Figure 5 As shown in the figure, a fast-converging critical path screening solution is provided, which is mainly divided into a hybrid screening process combining Sorted Slack and Sorted Degradation data and a compensatory screening process combining STA Slack and Degradation data. The hybrid screening process quickly obtains a part of the critical paths that meet the target requirements through a simple set intersection operation, and then decides to end the screening or compensate the screening based on whether the number of paths meets the standard. The compensatory screening process superimposes the estimated Slack degradation amount on the STA Slack to obtain the Evaluated Slack, and arranges it in order from the smallest to the smallest, and takes it out in turn in the feedback mechanism and supplements it into the critical path set until the number of paths meets the standard.

[0094] In the embodiment of the present invention, the STA tool extracts the timing path of the design circuit and calculates the corresponding STA Slack, and the SPICE simulator completes the timing characterization with aging effects on the timing unit of the design circuit, and combines the two results to quickly estimate the Slack data and its Degradation data after 0 years and aging, as well as the associated timing feature data. Slack and Degradation are important timing feature data with aging effects. In their corresponding comprehensive scoring mechanisms, multiple groups of related data are integrated to collaboratively complete the special scoring of the data. In the scoring process, it is necessary to combine the experience of timing analysis to determine the influence of each group of data, accurately complete the data sorting, and reasonably set the score adjustment step to obtain a more accurate special data score sorting order. The ultimate task goal is to design a simple and fast-converging data screening solution to select the timing path with the worst Slack and the largest corresponding Slack Degradation after circuit aging. Therefore, the timing path screening solution takes both Slack and Degradation special data into consideration, with Slack data as the main data and Degradation data as the supplementary data. The design solution through hybrid screening and compensation screening can converge quickly to obtain a unique and reliable set of key timing paths.

[0095] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a program running on the processor, and when the processor runs the program, the steps of the method for screening a critical timing path set with aging effects are executed. The method for screening a critical timing path set with aging effects is described in the aforementioned part and will not be repeated here.

[0096] The present invention also provides a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed, the steps of the above-mentioned method for screening a critical timing path set with aging effects are executed. The method for screening a critical timing path set with aging effects is described in the aforementioned part and will not be repeated here.

[0097] Those skilled in the art can understand that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions recorded in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for screening a critical timing path set with aging effects, comprising the following steps: Use STA and SPICE unit timing feature library lookup table to obtain timing feature data with aging effect; Sorting the time series feature data with aging effect by using a scoring and sorting mechanism to obtain a scoring and sorting order of the time series feature data with aging effect; Based on the scoring and sorting order of the timing characteristic data with aging effect, the critical paths are screened by taking into account the timing slack and timing degradation of the timing paths to obtain a critical timing path set; The step of performing critical path screening based on the scoring and sorting order of the timing feature data with aging effect and taking into account the timing slack and timing degradation of the timing path to obtain a critical timing path set also includes: Hybrid screening of critical paths that takes into account both timing slack and timing degradation; Critical path compensation screening that takes into account both timing slack and timing degradation; The step of hybrid screening of key paths taking into account both timing slack and timing degradation also includes: recording the size of the timing path set to be selected as M, and N>M>0, where N represents the total sample size of all timing path sets; performing hybrid screening based on the scoring and sorting order of the timing slack and timing degradation associated data, traversing the timing paths ranked in the top M positions in the scoring and sorting order of the timing slack associated data, and if the ranking of this timing path in the scoring and sorting order of the timing degradation associated data is also ranked in the top M positions, taking it out as the key timing path; if M timing paths are selected, the selection task is completed; otherwise, assuming that only K timing paths are selected, and N>M>K>0, MK timing paths are selected from the remaining NK timing path sets for supplementation; The step of supplementary screening of the critical paths taking into account both the timing slack and the timing degradation also includes: The aged timing degradation is taken out from the remaining NK timing path sets and added to the STA timing slack as the estimated timing slack. The paths are arranged in ascending order, and the timing paths ranked in the first MK positions are taken to complete the supplement of the key timing path set. The screening is ended and a key timing path set with a size of M and unique elements is selected.

2. The method for screening a critical timing path set with aging effect according to claim 1, characterized in that: The step of using STA and SPICE unit timing feature library lookup table to obtain timing feature data with aging effect also includes: Calculate cell delay by interpolation based on a lookup table of a cell feature library; estimating timing slack based on the unit delay; estimating data arrival time and data demand time based on the unit delay; Calculating the aging timing degradation amount according to the timing relaxation amount; Based on the data arrival time and the data demand time, calculating a ratio of a data arrival time degradation amount to a data demand time degradation amount; Calculate the absolute difference between the number of data arrival path units and the number of required path units.

3. The method for screening a critical timing path set with aging effect according to claim 2, characterized in that: The step of calculating the cell delay by interpolation based on the lookup table of the cell feature library also includes: taking the timing path as the object, and obtaining the initial cell delay and the aged cell delay contained in the path by interpolation based on the lookup table of the cell feature library.

4. The method for screening a critical timing path set with aging effect according to claim 2, characterized in that: The step of estimating the timing relaxation amount according to the unit delay also includes: using the unit delay back-labeling method to back-label the initial unit delay and the aged unit delay contained in the path to the STA data result of the timing path to obtain the initial timing relaxation amount and the aged timing relaxation amount.

5. The method for screening a critical timing path set with aging effect according to claim 2, characterized in that: The step of estimating the data arrival time and the data demand time based on the unit delay also includes: using the unit delay back-labeling method to back-label the unit delays of the data arrival path and the data demand path back to the STA data results of the timing path, including the initial data arrival time, the aged data arrival time, the initial data demand time and the aged data demand time.

6. The method for screening a critical timing path set with aging effect according to claim 2, characterized in that: The step of calculating the ratio of the data arrival time degradation amount to the data demand time degradation amount based on the data arrival time and the data demand time further includes: Based on the data arrival time and the data demand time, the data arrival time degradation amount and the data demand time degradation amount are calculated respectively; The ratio of the data arrival time degradation amount to the data demand time degradation amount is taken as the ratio of the data arrival time degradation amount to the data demand time degradation amount.

7. The method for screening a critical timing path set with aging effect according to claim 2, characterized in that: The step of calculating the absolute difference between the number of data arrival path units and the number of required path units also includes: The number of sequential units included in the data arrival and demand paths is counted respectively, and the absolute difference between the number of sequential units included in the data arrival path and the number of sequential units included in the data demand path is taken as the absolute difference between the number of data arrival path units and the number of data demand path units.

8. The method for screening a critical timing path set with aging effect according to claim 1, characterized in that: The step of using a scoring and sorting mechanism to sort the time series feature data with aging effects to obtain a scoring and sorting order for the time series feature data with aging effects also includes: Scoring and sorting the multiple groups of time series relaxation amount associated data to obtain the scoring and sorting order of the multiple groups of time series relaxation amount associated data; Scoring and sorting the multiple groups of time series degradation amount associated data to obtain the scoring and sorting order of the multiple groups of time series degradation amount associated data.

9. The method for screening a critical timing path set with aging effect according to claim 8, characterized in that: The step of scoring and sorting the multiple groups of timing slack associated data to obtain the scoring and sorting order of the multiple groups of timing slack associated data also includes: In the original timing path set, the paths are arranged in ascending order according to the aging timing relaxation amount, which serves as the main sorting of the original timing path set; Based on the primary sorting, the initial timing slack and the STA timing slack are sorted in ascending order to obtain a secondary sorting of the original timing path set; The size of the original timing path set is recorded as N, where N is a positive integer, and each timing path is given an initial score according to the order in the main sorting; Traverse the primary sorting to the i-th timing path, whose initial score is Ni, where i is a positive integer and N is a positive integer, and adjust the score according to the index position of the timing path in the secondary sorting; The score ranking order of the timing relaxation amount associated data is obtained in descending order of the scores.

10. The method for screening a critical timing path set with aging effect according to claim 8, characterized in that: The step of scoring and sorting the multiple groups of time series degradation amount associated data to obtain the scoring and sorting order of the multiple groups of time series degradation amount associated data also includes: In the original timing path set, the paths are first arranged in descending order according to the absolute values ​​of the aging timing degradation amounts, as the main sorting of the original timing path set; On the basis of the primary sorting, the ratio of the data arrival time degradation amount to the data demand time degradation amount is sorted in order from large to small or from small to large with reference to the timing check type, and the absolute difference between the number of data arrival path units and the number of data demand path units is sorted in order from large to small, so as to obtain the secondary sorting of the original timing path set; The size of the original timing path set is recorded as N, where N is a positive integer, and each timing path is given an initial score according to the order of the main sorting; Traverse the primary sorting to the i-th timing path, whose initial score is Ni, where i is a positive integer and N is a positive integer, and adjust the score according to the index position of the timing path in the secondary sorting; The score ranking order of multiple groups of timing degradation amount correlation data is obtained in descending order of scores.

11. An electronic device comprising a memory and a processor, characterized in that: The memory stores a program running on the processor, and the processor executes the steps of the method for screening a critical timing path set with aging effects as described in any one of claims 1 to 10 when running the program.

12. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed, the steps of a method for screening a critical timing path set with aging effects as described in any one of claims 1 to 10 are executed.