A peak power consumption credibility determination method and device, and electronic equipment

Abstract

The application discloses a peak power consumption credibility determination method and device and electronic equipment, and relates to the chip technical field, and comprises the following steps: in the case that the back labeling rate of the switch activity file of the chip to the gate level net list does not reach the back labeling rate threshold, the peak power consumption and the peak time corresponding to the peak power consumption are obtained; according to the preset time window half-width, the target time window centered on the peak time is determined; in the target time window, according to the energy contribution information of the power consumption contribution object of the gate level net list, the target power consumption contribution object of the peak power consumption is determined in the contribution object, the activity data source of each target power consumption contribution object is analyzed, and the source information of each target power consumption contribution object is obtained; based on the source information, the credibility data is determined. By determining the target power consumption contribution object in the target time window and analyzing the source information, the credibility data is determined based on the source information, the credibility of the peak power consumption is determined, and the accuracy of the peak power consumption signature conclusion is improved.

Description

Technical Field

[0001] This application relates to the field of chip technology, and in particular to a method, apparatus and electronic device for determining the reliability of peak power consumption. Background Technology

[0002] During the chip design process, the peak power consumption of the chip is determined and verified to ensure that the power supply and heat dissipation design can withstand the chip's maximum instantaneous power consumption.

[0003] Currently, the power analysis tool of Electronic Design Automation (EDA) reads the switching activity file at the gate level and performs power calculations to obtain the peak power consumption and peak time. At the same time, it obtains the back-annotation rate of the switching activity file to the gate-level netlist. If the back-annotation rate meets a preset threshold, the peak power consumption is used as the basis for formal approval.

[0004] However, the back-scanning rate only reflects the overall back-scanning situation of the chip. When un-back-scanned objects become the main contributors to peak power consumption after being supplemented, the judgment method based on the back-scanning rate cannot identify the credibility of peak power consumption, affecting the accuracy of the peak power consumption verification conclusion. Summary of the Invention

[0005] In view of this, this application provides a method, apparatus and electronic device for determining the credibility of peak power consumption. The main purpose is to improve the technical problem in the prior art where the back-scanning rate only reflects the back-scanning situation of the chip as a whole. When the un-scanned objects become the main contributors to the peak power consumption after being supplemented, the judgment method based on the back-scanning rate cannot identify the credibility of the peak power consumption, thus affecting the accuracy of the peak power consumption verification conclusion.

[0006] Firstly, this application provides a method for determining the reliability of peak power consumption, including: If the reverse scaling rate of the chip's switching activity file to the gate-level netlist does not reach the reverse scaling rate threshold, obtain the chip's peak power consumption and the peak time corresponding to the peak power consumption. Based on the preset half-width of the time window, a target time window centered on the peak moment is determined; Within the target time window, based on the energy contribution information of the power contribution objects in the gate-level netlist, the target power contribution object with peak power consumption is determined among the contribution objects. The source of the activity data of each target power contribution object is analyzed to obtain the source information of each target power contribution object. The power contribution object is the network and cell that contributes dynamic power consumption in the gate-level netlist, and the energy contribution information is the value of the dynamic energy contributed by the power contribution object within the target time window. Based on the source information, the credibility data of the peak power consumption is determined, and the credibility data is used to determine whether the peak power consumption has passed the approval.

[0007] Optionally, within the target time window, based on the energy contribution information of the power contribution objects in the gate-level netlist, the target power contribution object with peak power consumption is determined from the target contribution objects, and the source of the activity data for each target power contribution object is analyzed to obtain the source information for each target power contribution object, including: Obtain the energy contribution information of the power consumption contribution objects of the gate-level netlist within the target time window; The power consumption contribution objects are sorted according to the energy contribution information to obtain a power consumption contribution object sequence; The target power consumption contribution object for the peak power consumption is determined based on the power consumption contribution object sequence, wherein the target power consumption contribution object includes network power consumption contribution object and unit power consumption contribution object; Analyze the activity data source of each target power consumption contribution object to obtain the source information of each target power consumption contribution object.

[0008] Optionally, the step of analyzing the activity data source of each target power consumption contributing object to obtain the source information of each target power consumption contributing object includes: Analyze the active data source of each target power consumption contribution object, and determine the source level of the active data source. The source level includes at least one of the following: active file back-annotation level, propagation completion level, preset compensation level, inference level, and unannotated level. The source level is determined as the source information for each of the target power consumption contribution objects.

[0009] Optionally, determining the reliability data of the peak power consumption based on the source information includes: Based on the source information, the target power consumption contribution object whose source level is the activity file back-annotation level is determined as a trusted object; Determine the sum of the first energy contribution information of all the trusted objects, and the sum of the second energy contribution information of all the target power consumption contribution objects; The reliability data of the peak power consumption is determined based on the sum of the first energy contribution information and the sum of the second energy contribution information.

[0010] Optionally, after determining the confidence data of the peak power consumption based on the source information, the method further includes: If the credibility data meets the credibility threshold condition, the peak power consumption is determined to be approved. If the credibility data does not meet the credibility threshold condition, it is determined that the peak power consumption has not passed the approval.

[0011] Optionally, determining the target time window centered on the peak moment based on a preset time window half-width includes: Extend the time window forward by half its width, centered on the peak time, to obtain the starting time after the peak time has been extended; The time window is extended backward from the peak time to obtain the extended end time of the peak time. Based on the start time and the end time, a target time window centered on the peak time is determined.

[0012] Secondly, this application provides a device for determining the reliability of peak power consumption, comprising: The acquisition module is configured to acquire the peak power consumption of the chip and the peak time corresponding to the peak power consumption when the back-scaling rate of the chip's switching activity file to the gate-level netlist does not reach the back-scaling rate threshold. The determination module is configured to determine a target time window centered on the peak moment based on a preset time window half-width; The analysis module is configured to, within the target time window, determine the target power consumption contribution object with peak power consumption from the power consumption contribution objects according to the energy contribution information of the power consumption contribution objects in the gate-level netlist, analyze the activity data source of each target power consumption contribution object, and obtain the source information of each target power consumption contribution object. The power consumption contribution object is the network and cell that contributes dynamic power consumption in the gate-level netlist, and the energy contribution information is the value of the dynamic energy contributed by the power consumption contribution object within the target time window. The determination module is also configured to determine the credibility data of the peak power consumption based on the source information, the credibility data being used to determine whether the peak power consumption has passed the approval process.

[0013] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the confidence method for determining peak power consumption as described in the first aspect.

[0014] Fourthly, this application provides an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the computer program to implement the confidence determination method for peak power consumption described in the first aspect.

[0015] By employing the above technical solution, this application provides a method, apparatus, and electronic device for determining the credibility of peak power consumption, comprising: acquiring the peak power consumption of the chip and the peak time corresponding to the peak power consumption when the back-annotation rate of the switching activity file to the gate-level netlist does not reach the back-annotation rate threshold; determining a target time window centered on the peak time according to a preset time window half-width; within the target time window, determining the target power consumption contributing object of the peak power consumption from the power consumption contributing objects of the gate-level netlist, analyzing the source of the activity data of each target power consumption contributing object, and obtaining the source information of each target power consumption contributing object; and determining the credibility data of the peak power consumption based on the source information, wherein the credibility data is used to determine whether the peak power consumption has passed the approval. Compared with the existing technology, this application achieves targeted analysis of peak power consumption in scenarios with incomplete back-annotation by acquiring the peak power consumption and peak time when the back-annotation rate of the switching activity file to the gate-level netlist does not reach the back-annotation rate threshold. By determining the target time window centered on the peak time, it achieves focusing on the local time range near the peak. By determining the target power consumption contributing object within the target time window and analyzing the source of the activity data to obtain the source information, it achieves the identification of the activity source of the main contributing object of the peak. By determining credibility data based on source information, the credibility of peak power consumption can be quantitatively determined, reducing the risk of false signing caused by relying solely on the global back-annotation rate, and improving the accuracy of peak power consumption signing conclusions. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a method for determining the reliability of peak power consumption according to an embodiment of this application is shown. Figure 2 A flowchart illustrating a method for determining the reliability of peak power consumption according to an embodiment of this application is shown. Figure 3 A schematic diagram of the structure of a peak power consumption reliability determination device provided in an embodiment of this application is shown; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0019] The embodiments of this application will now be described in more detail with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0020] To address the technical problem in existing technologies where the back-scaling rate only reflects the overall back-scaling status of the chip, and where un-scaled components become major contributors to peak power consumption after being supplemented, the back-scaling rate-based determination method cannot identify the reliability of peak power consumption, thus affecting the accuracy of peak power consumption verification conclusions. This embodiment provides a method for determining the reliability of peak power consumption, such as... Figure 1 As shown, the method includes: Step 101: If the back-scaling rate of the chip's switching activity file to the gate-level netlist does not reach the back-scaling rate threshold, obtain the chip's peak power consumption and the peak time corresponding to the peak power consumption.

[0021] In this application embodiment, the chip may refer to an integrated circuit chip, which can be used to perform functions such as data processing, signal transmission, and logical operations. For example, the chip in this application embodiment may include, but is not limited to, a central processing unit chip, a graphics processing unit chip, an application-specific integrated circuit chip, and a system-on-a-chip.

[0022] In the embodiments of this application, the switching activity file can refer to a data file that records the toggling of various signals in the chip over time during simulation. The switching activity file can be used to provide signal activity data for gate-level power consumption analysis. For example, the switching activity file in the embodiments of this application may include, but is not limited to, a FastSignal Database (FSDB) format file and a Value Change Dump (VCD) format file. The FSDB format file can be a waveform file in binary compressed format, which can record the complete toggling sequence of the signal; the VCD format file can be a waveform file in text format, which can record the value changes of the signal.

[0023] In this embodiment, a gate-level netlist can refer to the circuit connection description of a chip during the gate-level design phase. The gate-level netlist can include logic units and the connection relationships between them. For example, the gate-level netlist in this embodiment can specifically be a gate-level circuit netlist generated by a synthesis tool. The logic units in the gate-level netlist can specifically include, but are not limited to, basic logic gates such as AND gates, OR gates, NOT gates, flip-flops, and adders.

[0024] In the embodiments of this application, back-annotation can refer to the process of mapping signal activity data in a switch activity file to networks and cells in a gate-level netlist. Back-annotation can be used to apply activity data from gate-level simulation or higher-level simulation to gate-level power consumption calculation. The back-annotation rate can refer to the proportion of signal activity data in a switch activity file that is successfully mapped to objects in a gate-level netlist. The back-annotation rate can be used to measure the coverage of the gate-level netlist by the switch activity file. For example, the back-annotation rate in the embodiments of this application may specifically include network back-annotation rate and cell back-annotation rate. The network back-annotation rate can refer to the proportion of successfully back-annotated networks to the total number of networks, and the cell back-annotation rate can refer to the proportion of fully back-annotated leaf cells to the total number of leaf cells. A leaf cell can refer to the lowest-level, indivisible logical unit in the gate-level netlist.

[0025] In this embodiment, the back-annotation rate threshold can be a pre-set critical value for determining whether subsequent credibility assessment needs to be performed. The back-annotation rate threshold can be used to offload scenarios with different back-annotation coverage levels. For example, the back-annotation rate threshold in this embodiment can be 100%, or it can be configured to other values ​​such as 99.5% according to project requirements. When either the network back-annotation rate or the unit back-annotation rate is lower than the back-annotation rate threshold, it can be considered that the back-annotation rate of the switch activity file on the gate-level netlist has not reached the back-annotation rate threshold.

[0026] In the embodiments of this application, peak power consumption refers to the maximum instantaneous power consumption reached by the chip at a certain moment during operation. Peak power consumption can be used to guide the power network design, heat dissipation design, and packaging design of the chip. Peak power consumption can be composed of switching power consumption, internal dynamic power consumption, and leakage power consumption. Among them, switching power consumption can be generated by the charging and discharging of the interconnect capacitors of the network; internal dynamic power consumption can be generated by the state transitions within the cell, such as the power consumption consumed by the state dependent path delay (SDPD) arc transition within the cell; leakage power consumption can be generated by the static leakage of the transistor. Leakage power consumption usually changes slowly over time and is generally not considered as the main ranking item for peak hotspots, except in the case of power gating transitions. An SDPD arc can refer to a specific transition path from an input state to an output state within the cell, and the power consumption of different paths can be different. Power gating can refer to the technique of directly cutting off power to unused modules to save power. Power gating can generate large current fluctuations at the moment of power-on or power-off.

[0027] In the embodiments of this application, the peak moment can refer to the moment corresponding to the peak power consumption on the simulation time axis, and the peak moment can be used as the time domain anchor point for subsequently defining the target time window.

[0028] For the embodiments of this application, scenarios where the back-annotation rate of the switching activity file to the gate-level netlist does not reach the back-annotation rate threshold may include, but are not limited to: gate-level simulation recording only some layers or some signals resulting in incomplete activity file coverage; uncovered areas still existing after splicing multiple activity files; and incomplete activity mapping resulting from direct analysis of gate-level activity by Register Transfer Level (RTL) activity. In the above scenarios, objects not directly back-annotated by the switching activity file can be processed in subsequent power consumption calculations using propagation padding, default padding, or inference padding. Propagation padding can refer to inferring the activity data of un-back-annotated objects from the actual activity of adjacent objects; default padding can refer to the tool padding un-back-annotated objects using a preset default toggle rate; and inference padding can refer to the tool inferring the activity data of un-back-annotated objects based on logical relationships or clock derivation relationships. The activity data obtained by the above padding methods may not reflect the actual behavior of the chip, therefore, the reliability of peak power consumption needs to be evaluated.

[0029] In this embodiment, the switch activity file can first be read into the entry-level netlist and then back-annotated using a command to read the switch activity file. The command to read the switch activity file can include, but is not limited to, the `read_fsdb` command and the `read_vcd` command. The network back-annotation rate and the cell back-annotation rate can be parsed from the summary information of the back-annotation results. Fields in the summary information can include the total number of networks, the number and proportion of back-annotated networks, the total number of leaf cells, and the number and proportion of fully back-annotated leaf cells.

[0030] In the embodiments of this application, when both the network back-annotation rate and the cell back-annotation rate reach the back-annotation rate threshold, it can be considered that the switch activity file has achieved full coverage of the gate-level netlist, and the risk of the source dimension can be basically eliminated. At this time, it can be directly marked as the source dimension and quickly released without performing subsequent credibility assessment steps. For example, when a fully covered gate-level FSDB file matching the gate-level netlist is generated using tools such as PowerReplay, the back-annotation rate can usually reach 100%, and it can be directly and quickly released. When either the network back-annotation rate or the cell back-annotation rate does not reach the back-annotation rate threshold, it can be considered that the switch activity file has incomplete coverage, and subsequent credibility assessment steps need to be performed.

[0031] In this embodiment, after determining that the back-scaling rate has not reached the back-scaling rate threshold, full-range power consumption calculation can be performed on the gate-level netlist. Power consumption calculation can employ a time-based analysis mode. In this mode, dynamic energy can be accumulated according to timestamped events to obtain a waveform showing power consumption changes over time. Full-range power consumption calculation can be executed using the `update_power` command. After the full-range power consumption calculation is completed, a power consumption report can be generated using the `report_power` command. The peak power consumption and the corresponding peak time can be obtained from the power consumption report. Alternatively, a global maximum point can be extracted from the power consumption waveform file, with the ordinate of the global maximum point determined as the peak power consumption and the abscissa as the peak time. Alternatively, a known peak time can be used for subsequent local time window verification. For example, the power consumption calculation tools in this embodiment may include, but are not limited to, Electronic Design Automation (EDA) power analysis tools such as PrimePower and PTPX.

[0032] Step 102: Determine the target time window centered on the peak moment based on the preset time window half-width.

[0033] In this embodiment, the half-width of the time window can refer to the length of time extending forward and backward from the peak moment. The half-width of the time window can be used to determine the time range of the target time window. For example, the half-width of the time window in this embodiment can specifically be several clock cycles, a preset fixed duration, or a duration adaptively determined based on the half-width of the power consumption waveform. The half-width can refer to the time width between two time points on the left and right when the power consumption value is equal to half of the peak power consumption in the power consumption waveform. The half-width can be used to describe the time width of the power consumption pulse.

[0034] In the embodiments of this application, the target time window can refer to a local time interval centered on the peak time and extending by half the width of each subsequent time window. The target time window can be used to define the time range for subsequent activity source evaluation. The target time window can fall within the time range of the switch activity file to ensure that activity data within the target time window can be read from the switch activity file.

[0035] In this embodiment, the peak moment obtained in step 101 can be used as the time-domain anchor point. Based on a preset half-width time window, a local time interval is defined before and after the peak moment to obtain the target time window. Using a local time window with the peak moment as the time-domain anchor point for attribution determination, rather than focusing solely on the peak, avoids the influence of factors such as event ordering, sampling granularity, instantaneous spike jitter, and the tool's internal integration method, making the determination conclusion more stable and easier to reproduce. The determination object of the target time window can be the source credibility of the local time window near the peak, rather than a rigorous mathematical proof of the peak. The peak moment can be used for time-domain anchoring, and the target time window can be used for attribution and determination.

[0036] In this embodiment of the application, after determining the target time window, the switching activity file can be reread and power consumption calculations performed within the target time window to obtain the dynamic energy contribution of each power consumption contributing object within the target time window. When rereading the switching activity file, only the activity data within the target time window can be read by specifying a time range to improve computational efficiency. Specifying the time range can be achieved using the `-time` option of the `read_fsdb` command.

[0037] As an alternative approach, in addition to defining a local time window centered on the peak moment, other methods can be used to determine the target time window. For example, multiple time windows can be evaluated using multi-scenario analysis, or an approximate full-range method can be used for evaluation. This application does not limit the specific methods used in this embodiment.

[0038] Step 103: Within the target time window, based on the energy contribution information of the power contribution objects in the gate-level netlist, determine the target power contribution object with peak power consumption among the contribution objects, analyze the source of activity data for each target power contribution object, and obtain the source information of each target power contribution object.

[0039] In the embodiments of this application, the power consumption contribution object can refer to an element in the gate-level netlist that can contribute dynamic power consumption. The power consumption contribution object can be used as an object for evaluating the source of activity. For example, the power consumption contribution object in the embodiments of this application may specifically include networks and cells, wherein the network can mainly contribute switching power consumption, and the switching power consumption of the network can be generated by the charging and discharging of interconnecting capacitors; the cell can contribute internal dynamic power consumption and cell-related switching power consumption, and the internal dynamic power consumption of the cell can be generated by the state transition within the cell.

[0040] In this embodiment, the target power consumption contribution object can refer to a power consumption contribution object that contributes significantly to peak power consumption within a target time window. The target power consumption contribution object can be used as a key object for evaluating the credibility of the activity source. For example, in this embodiment, the target power consumption contribution object may specifically include networks and cells that rank highly in dynamic energy contribution within the target time window.

[0041] In this embodiment, energy contribution information refers to the value of dynamic energy contributed by a power consumption contributing object within a target time window. This energy contribution information can be used to sort power consumption contributing objects to filter target power consumption contributing objects. For example, the energy contribution information in this embodiment may specifically include the switching energy of the network and the internal dynamic energy of the unit.

[0042] In the embodiments of this application, the source of activity data can refer to the method of obtaining the activity data of the target power consumption contribution object. The source of activity data can be used to determine whether the activity data of the target power consumption contribution object is reliable. For example, the source of activity data in the embodiments of this application may specifically include direct back-annotation of activity files, propagation completion, default padding, inference, and unannotated data.

[0043] In this embodiment, source information may refer to information reflecting the source category of the activity data of the target power consumption contributing object. This source information can be used to subsequently calculate the reliability data of peak power consumption. For example, in this embodiment, the source information may specifically be the source level corresponding to each target power consumption contributing object.

[0044] In this embodiment of the application, within the target time window determined in step 102, target power consumption contribution objects that contribute significantly to peak power consumption can be selected from all power consumption contribution objects in the gate-level netlist. The activity data of each target power consumption contribution object is analyzed individually; it is determined whether the data is obtained by direct back-annotation of the switch activity file or by propagation completion, default value supplementation, inference, or other methods. The analysis results are used as the source information for each target power consumption contribution object. Since peak power consumption exhibits temporal locality and energy concentration, the dynamic energy near the peak time is often concentrated on a few power consumption contribution objects. Therefore, selecting target power consumption contribution objects for focused evaluation can improve evaluation efficiency while ensuring evaluation accuracy.

[0045] In this application embodiment, the method of identifying the source of activity data may include, but is not limited to: identifying the back-annotation status of each target power consumption contribution object by combining attribute query with pin query, or identifying the source by cross-hitting the target power consumption contribution object with a known set of unannotated objects.

[0046] As an optional approach, the definition of high-trust objects can be configured according to project requirements. For example, only objects directly back-annotated in the activity file can be designated as high-trust objects, or both objects directly back-annotated in the activity file and objects propagated and completed can be designated as high-trust objects.

[0047] Step 104: Based on the source information, determine the reliability data of the peak power consumption.

[0048] Among them, the credibility data is used to determine whether the peak power consumption has passed the approval.

[0049] In the embodiments of this application, credibility data can refer to quantitative data reflecting the credibility of the activity source of peak power consumption. Credibility data can be used to determine whether peak power consumption has endorsement validity. For example, the credibility data in the embodiments of this application can specifically be a high credibility ratio, which can refer to the proportion of the sum of the dynamic energy of the target power consumption contributing objects with credible activity sources to the sum of the dynamic energy of all target power consumption contributing objects.

[0050] In this embodiment, approval refers to the formal confirmation and signing off of power consumption analysis results in the chip design process. Approval can be used to determine whether peak power consumption can serve as a formal basis for power supply design, heat dissipation design, and packaging design. For example, the approval conclusion in this embodiment may specifically include allowing formal approval, blocking approval, and conditional release.

[0051] In this embodiment of the application, based on the source information of each target power consumption contributing object obtained in step 103, and combined with the dynamic energy contribution of each target power consumption contributing object within the target time window, the credibility data of the peak power consumption can be calculated in a dynamic energy weighted manner. The higher the credibility data, the more credible the source of the activity data of the object contributing the main dynamic energy within the local time window near the peak, and the higher the verification validity of the peak power consumption data; the lower the credibility data, the more dynamic energy within the local time window near the peak comes from objects with unreliable activity sources, and the peak power consumption data may deviate from the actual behavior.

[0052] In some examples, existing technologies typically employ global activity checks to statistically determine the proportion of unlabeled elements in the entire design, or rely on engineers' experience to manually review hotspot reports, or use the global direct back-annotation ratio as a threshold (e.g., allowing peak values ​​to pass if the back-annotation ratio is greater than or equal to 95%), or use a process checklist to check the completeness of activity documents. None of these existing methods limit the peak window, do not weight by energy, or output a validity judgment tied to a single peak value, making them prone to mislabeling due to a high global back-annotation ratio. This application's embodiment, by evaluating the credibility of activity sources for high-dynamic energy objects within a local time window near the peak value using energy weighting, can overcome the blind spot of the global back-annotation ratio threshold and reduce the risk of mislabeling peak values ​​in scenarios with incomplete back-annotation.

[0053] In this embodiment of the application, the credibility data can be compared with a preset credibility threshold to determine whether the peak power consumption has passed the approval. When the credibility data meets the credibility threshold condition, it can be determined that the peak power consumption has passed the approval; when the credibility data does not meet the credibility threshold condition, it can be determined that the peak power consumption has not passed the approval, and the power consumption analysis and credibility assessment need to be re-executed after supplementing the coverage of the switch activity file.

[0054] Compared with existing technologies, this application achieves targeted analysis of peak power consumption in scenarios with incomplete back-annotation by obtaining peak power consumption and peak time when the back-annotation rate of the gate-level netlist in the switch activity file does not reach the back-annotation rate threshold. By determining a target time window centered on the peak time, it focuses on a local time range near the peak. By identifying the target power consumption contributors within the target time window and analyzing the source of activity data to obtain source information, it identifies the main contributors to the peak power consumption. By determining credibility data based on the source information, it quantitatively judges the credibility of peak power consumption, reducing the risk of missigning due to relying solely on the global back-annotation rate for verification, and improving the accuracy of peak power consumption verification conclusions.

[0055] As an optional approach, when performing the task of "within the target time window, based on the energy contribution information of the power contribution objects in the gate-level netlist, determining the target power contribution object with peak power consumption among the contribution objects, analyzing the activity data source of each target power contribution object, and obtaining the source information of each target power contribution object," the following methods can be used, but are not limited to them: Figure 2 As shown, it includes: Step 201: Obtain the energy contribution information of the power consumption contribution objects in the gate-level netlist within the target time window.

[0056] In this embodiment, after re-performing the power consumption calculation within the target time window, the dynamic energy contribution of each power consumption contributor can be obtained from the power consumption report within the target time window, thus obtaining the energy contribution information of each power consumption contributor. The network's energy contribution information can be the switching energy consumed by the charging and discharging of the network's interconnect capacitors within the target time window, and the cell's energy contribution information can be the internal dynamic energy consumed by the state switching of the cell's internal transistors within the target time window. The methods for obtaining energy contribution information may include, but are not limited to, obtaining it from the power value in the power consumption report, estimating it by multiplying the number of switching times by the capacitance, or statistically analyzing it through event counting.

[0057] Step 202: Sort the power consumption contribution objects according to the energy contribution information to obtain the power consumption contribution object sequence.

[0058] In the embodiments of this application, the power consumption contribution object sequence can refer to an ordered sequence obtained by sorting all power consumption contribution objects according to the magnitude of their energy contribution information. The power consumption contribution object sequence can be used to filter target power consumption contribution objects that contribute significantly to peak power consumption. For example, the power consumption contribution object sequence in the embodiments of this application may specifically include a network sequence sorted from largest to smallest dynamic energy and a unit sequence sorted from largest to smallest dynamic energy.

[0059] In this embodiment, based on the energy contribution information obtained in step 201, all power consumption contribution objects in the gate-level netlist can be arranged in descending order of dynamic energy contribution to obtain a power consumption contribution object sequence. The network and cells can be sorted separately to obtain a network power consumption contribution object sequence and a cell power consumption contribution object sequence, so that target power consumption contribution objects in the network and cells can be selected subsequently.

[0060] Step 203: Determine the target power contribution object for peak power consumption based on the power contribution object sequence.

[0061] In this embodiment of the application, a preset number of power consumption contribution objects are selected from the power consumption contribution object sequence obtained in step 202, and these are determined as the target power consumption contribution objects for peak power consumption. For example, a preset number of networks can be selected from the network power consumption contribution object sequence as network target power consumption contribution objects, and a preset number of units can be selected from the unit power consumption contribution object sequence as unit target power consumption contribution objects. The preset number can be configured according to chip size and evaluation accuracy requirements; for example, the top 100 networks and the top 100 units can be selected.

[0062] As an optional approach, in addition to filtering the target power consumption contribution objects separately using both network and unit channels, the method of filtering only units can also be adopted, or the method of first aggregating by hierarchy and then filtering can be adopted. This application embodiment does not limit this approach.

[0063] In the embodiments of this application, since peak power consumption has energy concentration, a small number of power consumption contributing objects ranked at the top can often contribute most of the dynamic energy within the target time window. Therefore, selecting a preset number of power consumption contributing objects ranked at the top as target power consumption contributing objects can reduce the number of evaluation objects and improve evaluation efficiency while ensuring that the evaluation covers the main energy contribution.

[0064] Step 204: Analyze the source of activity data for each target power consumption contribution object to obtain the source information for each target power consumption contribution object.

[0065] In this embodiment, the activity data of each target power consumption contribution object determined in step 203 can be queried one by one to determine whether it was obtained by direct back-annotation of the switch activity file or by processing methods such as propagation padding, default padding, or inference. The query results are used as the source information for each target power consumption contribution object. For the network, it can be queried whether the signal switching activity on the network is directly recorded by the switch activity file; for the unit, it can be queried whether the activity of each pin of the unit is directly back-annotated by the switch activity file. When all pins of the unit are successfully back-annotated, the unit can be considered to be fully back-annotated.

[0066] As an optional approach, when performing the "analysis of the activity data source of each target power consumption contribution object to obtain the source information of each target power consumption contribution object", the following method can be used, but is not limited to: analyzing the activity data source of each target power consumption contribution object, determining the source level of the activity data source, the source level including at least one of the activity file back-annotation level, propagation completion level, preset compensation level, inferred level and unannotated level; and determining the source level as the source information of each target power consumption contribution object.

[0067] In the embodiments of this application, the source level can refer to the classification of activity data sources based on their credibility. The source level can be used to quantify and distinguish the credibility of different activity data sources.

[0068] In this embodiment, the activity file back-annotation level can refer to the source level of the activity data of the target power consumption contribution object obtained by directly back-annotating the switch activity file. The activity file back-annotation level can indicate that the activity data comes from the real simulation record and has the highest degree of credibility, and can be counted as high credibility by default.

[0069] In the embodiments of this application, the propagation completion level can refer to the source level obtained by propagating the activity data of the target power consumption contributing object from the real activity of neighboring objects. The propagation completion level can indicate that the activity data is calculated based on the real neighbor signals, with a high degree of reliability. It can be included in the high reliability category by default and is configurable.

[0070] In this embodiment, the inference level can refer to the source level of the activity data of the target power consumption contributor, which is inferred by the tool based on logical relationships or clock derivation relationships. The inference level can indicate that the activity data is based on logical inference or clock signal derivation, and has a medium or low confidence level. By default, it is not counted as high confidence or is counted as medium confidence and is configurable. Clock derivation can refer to the activity of deriving related signals from the activity of clock signals.

[0071] In the embodiments of this application, the preset compensation level can refer to the source level obtained by the tool using a preset default toggle rate to compensate the activity data of the target power consumption contribution object. The preset compensation level can indicate that the activity data lacks real basis and has a low degree of credibility, and is not counted as high credibility by default.

[0072] In this application embodiment, "unlabeled level" can refer to the source level where the activity data of the target power consumption contribution object is not labeled or is assumed to be constant. An unlabeled level indicates missing activity data, has the lowest reliability, and is not considered high reliability by default. "Constant assumption" can mean that the tool assumes the signal does not flip and treats it as constant.

[0073] Optionally, the above method of dividing source levels is only an example. In actual applications, the number of source levels can be increased or decreased as needed, and the high-trust attribution configuration of each source level can also be adjusted. For example, only the active file back-annotation level can be determined as high-trust, or both the active file back-annotation level and the propagation completion level can be determined as high-trust, or the inference level can be configured to be included in high-trust. This application embodiment does not limit this.

[0074] As an optional approach, when performing the "determining the credibility data of peak power consumption based on source information", the following method can be used, but is not limited to: based on source information, identifying target power consumption contribution objects with a source level of active file back-annotation level as credible objects; determining the sum of the first energy contribution information of all credible objects and the sum of the second energy contribution information of all target power consumption contribution objects; and determining the credibility data of peak power consumption based on the sum of the first energy contribution information and the sum of the second energy contribution information.

[0075] In this embodiment, a trusted object can refer to a target power consumption contribution object whose source level is high trust level. The trusted object can be used to calculate the trusted energy component in the trust data. For example, in this embodiment, the trusted object can specifically be a target power consumption contribution object whose source level is the activity file back-annotation level.

[0076] In the embodiments of this application, the sum of the first energy contribution information may refer to the sum of the dynamic energy contributions of all trusted objects within the target time window, and the sum of the first energy contribution information may be used to represent the total amount of dynamic energy contributed by the activity data of trusted sources.

[0077] In the embodiments of this application, the sum of the second energy contribution information may refer to the sum of the dynamic energy contributions of all target power consumption contribution objects within the target time window, and the sum of the second energy contribution information may be used to represent the total dynamic energy of the target power consumption contribution objects within the target time window.

[0078] For example, the formula for calculating credibility data can be as shown in Formula 1, where C can represent credibility data (high credibility percentage), E_trust can represent the sum of the first energy contribution information, and E_total can represent the sum of the second energy contribution information.

[0079] C = E_trust / E_total (Formula 1) For example, when performing trustworthiness assessments on the network and cells separately, the formula for calculating the high trustworthiness percentage of the network can be shown in Formula 2, and the formula for calculating the high trustworthiness percentage of the cells can be shown in Formula 3. Here, C_net can represent the high trustworthiness percentage of the network, E_(net_trust) can represent the sum of the switching energies of trusted network objects, and E_(net_total) can represent the sum of the switching energies of all network target power consumption contributing objects; C_cell can represent the high trustworthiness percentage of the cell, E_(cell_trust) can represent the sum of the dynamic energies of trusted cell objects, and E_(cell_total) can represent the sum of the dynamic energies of all cell target power consumption contributing objects.

[0080] C_net=E_(net_trust) / E_(net_total) (Formula 2) C_cell = E_(cell_trust) / E_(cell_total) (Formula 3) For example, the determination of comprehensive credibility data can be achieved in two ways, including but not limited to the following: The first method is to take the smaller of the network's high credibility percentage and the unit's high credibility percentage as the comprehensive credibility data. This employs a "weakest link" strategy; if the credibility of any channel in the network or unit is low, the overall credibility is low. The second method is to calculate the comprehensive credibility data by weighting the network's high credibility percentage and the unit's high credibility percentage according to the total energy of the network channels and the total energy of the unit channels. The formula for calculating the comprehensive credibility data is shown in Formula 4, where C_total represents the comprehensive credibility data, E_(net_total) represents the total energy of the network channels, and E_(cell_total) represents the total energy of the unit channels.

[0081] C_total=(E_(net_total)×C_net+E_(cell_total)×C_cell) / (E_(net_total)+E_(cell_total)) (Formula 4) In the embodiments of this application, the credibility data can be compared with the credibility threshold. The credibility threshold can be configured according to project requirements, for example, it can be configured to 0.90. When the credibility data is greater than or equal to 0.90, it can be said that the source credibility of the local time window near the peak meets the approval requirements.

[0082] Optionally, other methods of calculating confidence data based on dynamic energy weighting can also be used. For example, the energy contribution can be estimated by multiplying the number of flips by the capacitance and then the proportion can be calculated. Alternatively, the weighted calculation can be performed by counting events. This application does not limit the specific methods used.

[0083] As an optional approach, after performing "determining the credibility data of peak power consumption based on source information", the following method can be used, but is not limited to: if the credibility data meets the credibility threshold condition, determine that the peak power consumption has passed the approval; if the credibility data does not meet the credibility threshold condition, determine that the peak power consumption has not passed the approval.

[0084] In the embodiments of this application, the credibility threshold condition can refer to a pre-set condition used to determine whether the credibility data meets the approval requirements. The credibility threshold condition can be used to determine whether the peak power consumption passes the approval. For example, in the embodiments of this application, the credibility threshold condition can specifically be that the credibility data is greater than or equal to the credibility threshold. The credibility threshold can be configured according to the risk tolerance of the project, for example, it can be configured as 0.90.

[0085] In this embodiment, when the credibility data meets the credibility threshold condition, it indicates that the credibility of the activity source within the local time window near the peak meets the approval requirements, and the peak power consumption can be determined to have passed the approval, allowing it to be used as the basis for formal approval. When the credibility data does not meet the credibility threshold condition, it indicates that a significant amount of dynamic energy within the local time window near the peak originates from untrusted sources, and the peak power consumption data may deviate from the actual behavior. In this case, the peak power consumption can be determined to have failed the approval, requiring the approval process to be blocked and the coverage of the activity file to be supplemented before re-executing the analysis.

[0086] In this embodiment of the application, after determining the approval conclusion, the judgment record can be bound and archived with the identification information of this operation. The judgment record may include, but is not limited to, information such as the back-bill ratio, peak time, target time window, time window mode, dual-channel confidence data, low-confidence object examples, confidence threshold, switch activity file path, process corner, analysis mode, and diversion results. Binding and archiving the judgment record with the operation identifier facilitates subsequent reproduction and auditing, ensuring the traceability of the approval conclusion.

[0087] For example, when the switch activity file records only some layers or some signals, resulting in a network back-annotation rate of approximately 95.99% and a unit back-annotation rate of approximately 96.2%, the prior art can directly allow the process to proceed because the global back-annotation rate reaches more than 95%. However, when evaluating using the method of this application embodiment, if approximately 12% of the hotspot energy within the target time window comes from the hole-filling object, and the overall confidence data is 0.84 and lower than the confidence threshold of 0.90, it can be determined that the peak power consumption has not passed the signing, thereby avoiding false signing.

[0088] For example, when using the Power-Replay tool to generate a full-coverage gate-level FSDB file that matches the gate-level netlist, both the network back-annotation rate and the cell back-annotation rate can be 100%. In this case, it can be marked as fully back-annotated and the detailed source evaluation can be skipped.

[0089] For example, when the switch activity file only records a portion of the layers, resulting in a global back-annotation rate of approximately 92%, but almost all hot objects within the peak window are within the recording range and are directly back-annotated by the switch activity file, the overall confidence score can be 0.95, which is higher than the confidence threshold of 0.90. This indicates that the peak power consumption has passed the approval process, demonstrating that a back-annotation rate of less than 100% does not necessarily mean rejection.

[0090] As an optional approach, when performing the "determine the target time window centered on the peak time based on the preset half-width of the time window", the following method can also be used, but is not limited to: extending the half-width of the time window forward with the peak time as the center to obtain the start time after the peak time is extended; extending the half-width of the time window backward with the peak time as the center to obtain the end time after the peak time is extended; and determining the target time window centered on the peak time based on the start time and the end time.

[0091] In the embodiments of this application, the starting time can refer to the starting point of the target time window, and the starting time can be obtained by extending the peak time forward by half the width of the time window.

[0092] In the embodiments of this application, the end time can refer to the end of the target time window, and the end time can be obtained by extending the time window halfway backward from the peak time.

[0093] In the embodiments of this application, the time range of the target time window can be limited by the start time and the end time.

[0094] In this embodiment, the half-width of the time window can be a number of clock cycles, a preset fixed number of nanoseconds, or a duration adaptively determined based on the half-width of the power consumption waveform. The target time window needs to fall within the time range of the switching activity file to ensure that the activity data within the target time window can be read from the switching activity file. After determining the target time window, the activity data within the target time window in the switching activity file can be reread by specifying a time range, and the power consumption calculation can be re-executed within the target time window to obtain the dynamic energy contribution of each power consumption contributing object within the target time window.

[0095] Compared with existing technologies, this application's embodiments obtain the energy contribution information of each power consumption contributing object within a target time window and sort them by energy contribution. The target power consumption contributing object is then determined from the sorting results, achieving precise screening of the main energy contributing objects for peak power consumption. This allows subsequent activity source assessment to focus on objects with a significant impact on peak power consumption, balancing assessment accuracy and computational efficiency. By classifying the activity data sources of each target power consumption contributing object, the reliability of the activity source is quantitatively differentiated. Target power consumption contributing objects with a source level equal to the activity file back-annotation level are identified as reliable objects. The sum of the first energy contribution information of reliable objects and the sum of the second energy contribution information of all target power consumption contributing objects are calculated separately. Based on these two, reliability data is determined, achieving a reliability calculation weighted by dynamic energy contribution. This ensures that the reliability data reflects the actual contribution ratio of reliable sources to peak energy, rather than simply being based on the number of objects. By comparing credibility data with credibility threshold conditions, peak power consumption is approved when the conditions are met and rejected when the conditions are not met. This automates the judgment of peak power consumption approval conclusions, ensuring consistent and reproducible judgment standards. It avoids inconsistencies caused by relying solely on engineer experience, and automatically blocks approval when credibility is insufficient to reduce the risk of false approvals. The start and end times are obtained by extending the time window half-width forward and backward from the peak time, and the target time window is determined based on the start and end times. This achieves symmetrical delineation of the local time range with the peak time as the time domain anchor point, ensuring that the activity source assessment covers the local energy range before and after the peak, guaranteeing the stability and reproducibility of attribution judgments.

[0096] Furthermore, as Figure 1 and Figure 2 The specific implementation of the method shown in this embodiment provides a device for determining the reliability of peak power consumption, such as... Figure 3 As shown, the device includes: an acquisition module 31, a determination module 32, and an analysis module 33.

[0097] The acquisition module 31 is configured to acquire the peak power consumption of the chip and the peak time corresponding to the peak power consumption when the back-scaling rate of the chip's switching activity file to the gate-level netlist does not reach the back-scaling rate threshold. The determination module 32 is configured to determine a target time window centered on the peak moment based on a preset time window half-width; Analysis module 33 is configured to, within the target time window, determine the target power consumption contribution object with peak power consumption among the power consumption contribution objects based on the energy contribution information of the power consumption contribution objects in the gate-level netlist, analyze the source of the activity data of each target power consumption contribution object, and obtain the source information of each target power consumption contribution object. The power consumption contribution object is the network and cell that contributes dynamic power consumption in the gate-level netlist, and the energy contribution information is the value of the dynamic energy contributed by the power consumption contribution object within the target time window. The determination module 32 is also configured to determine the credibility data of the peak power consumption based on the source information. The credibility data is used to determine whether the peak power consumption has passed the approval.

[0098] In some examples of this embodiment, the analysis module 33 is specifically configured to: obtain the energy contribution information of the power contribution objects of the gate-level netlist within the target time window; sort the power contribution objects according to the energy contribution information to obtain a power contribution object sequence; determine the target power contribution object for peak power consumption based on the power contribution object sequence; and analyze the activity data source of each target power contribution object to obtain the source information of each target power contribution object.

[0099] In some examples of this embodiment, the analysis module 33 is further configured to analyze the source of activity data for each target power consumption contribution object, determine the source level of the activity data source, and the source level includes at least one of activity file back-annotation level, propagation completion level, preset compensation level, inference level, and unannotated level; and determine the source level as the source information for each target power consumption contribution object.

[0100] In some examples of this embodiment, the determining module 32 is specifically configured to determine the target power consumption contribution object with a source level of active file back-annotation level as a trusted object based on the source information; determine the sum of the first energy contribution information of all trusted objects and the sum of the second energy contribution information of all target power consumption contribution objects; and determine the credibility data of peak power consumption based on the sum of the first energy contribution information and the sum of the second energy contribution information.

[0101] In some examples of this embodiment, the determining module 32 is further configured to determine that the peak power consumption passes the approval if the credibility data meets the credibility threshold condition, and to determine that the peak power consumption fails the approval if the credibility data does not meet the credibility threshold condition.

[0102] In some examples of this embodiment, the determining module 32 is further configured to extend the time window forward by half the width of the peak time to obtain the start time after the peak time extension; extend the time window backward by half the width of the peak time to obtain the end time after the peak time extension; and determine the target time window centered on the peak time based on the start time and the end time.

[0103] It should be noted that other corresponding descriptions of the functional units involved in the peak power consumption reliability determination device provided in this embodiment can be found in [reference needed]. Figure 1 and Figure 2 The corresponding description in [the document] will not be repeated here.

[0104] Based on the above, Figure 1 and Figure 2 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figure 1 and Figure 2 The method shown.

[0105] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.

[0106] like Figure 4 The diagram shown is a hardware structure schematic of an electronic device according to the present invention, comprising: At least one processor 401; and, A memory 402 connected to the processing of at least one processor 401 terminal device; wherein, The memory 402 stores instructions that can be executed by at least one processor to enable the at least one processor to perform the aforementioned peak power consumption confidence determination method.

[0107] Figure 4 Take a processor 401 as an example.

[0108] The electronic device may also include an input device 403 and an output device 404.

[0109] The processor 401, memory 402, input device 403, and output device 404 can be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0110] Memory 402, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the peak power consumption reliability determination method in the embodiments of this application, for example, Figure 1 and Figure 2 The method flow is shown. The processor 401 executes various functional applications and terminal device processing by running non-volatile software programs, instructions, and modules stored in the memory 402, thereby realizing the peak power consumption reliability determination method in the above embodiments.

[0111] Memory 402 may include a program storage area and a data storage area. The program storage area may store an operating system and an application program required for at least one function. The data storage area may store data created using the peak power consumption confidence determination method, etc. Furthermore, memory 402 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 402 may optionally include memory remotely located relative to processor 401, and these remote memories may be connected via a network to means of performing the peak power consumption confidence determination method. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, processing networks of mobile terminal devices, and combinations thereof.

[0112] Input device 403 can receive user clicks and generate signal inputs related to user settings and function control, such as those relating to the peak power consumption reliability determination method. Output device 404 may include display devices such as a display screen.

[0113] One or more modules are stored in memory 402, and when run by one or more processors 401, the peak power consumption reliability determination method in any of the above method embodiments is executed.

[0114] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.

[0115] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0116] The storage medium may also include an operating system and a processing module for network terminal devices. The operating system is a program that manages the hardware and software resources of the aforementioned physical devices, supporting the operation of information processing programs and other software and programs. The processing module for network terminal devices is used to enable terminal device processing between the various components within the storage medium, as well as terminal device processing with other hardware and software within the information processing physical device.

[0117] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented using software plus necessary general-purpose hardware platforms, or it can be implemented in hardware. By applying the scheme of this embodiment, compared with the prior art, this application embodiment obtains peak power consumption and peak time when the back-annotation rate of the gate-level netlist in the switching activity file does not reach the back-annotation rate threshold, thus achieving targeted analysis of peak power consumption in scenarios with incomplete back-annotation. By determining a target time window centered on the peak time, focusing on a local time range near the peak is achieved. By identifying the target power consumption contributing object within the target time window and analyzing the source of activity data to obtain source information, the source of the activity of the main contributing object of the peak is identified. By determining credibility data based on the source information, the credibility of the peak power consumption is quantitatively judged, reducing the risk of missigning caused by relying solely on the global back-annotation rate for verification, and improving the accuracy of the peak power consumption verification conclusion.

[0118] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0119] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for determining the reliability of peak power consumption, characterized in that, include: If the reverse scaling rate of the chip's switching activity file to the gate-level netlist does not reach the reverse scaling rate threshold, obtain the chip's peak power consumption and the peak time corresponding to the peak power consumption. Based on the preset half-width of the time window, a target time window centered on the peak moment is determined; Within the target time window, based on the energy contribution information of the power contribution objects in the gate-level netlist, the target power contribution object with peak power consumption is determined among the contribution objects. The source of the activity data of each target power contribution object is analyzed to obtain the source information of each target power contribution object. The power contribution object is the network and cell that contributes dynamic power consumption in the gate-level netlist, and the energy contribution information is the value of the dynamic energy contributed by the power contribution object within the target time window. Based on the source information, the credibility data of the peak power consumption is determined, and the credibility data is used to determine whether the peak power consumption has passed the approval.

2. The method according to claim 1, characterized in that, Within the target time window, based on the energy contribution information of the power contribution objects in the gate-level netlist, the target power contribution object with peak power consumption is determined from the contribution objects. The activity data source of each target power contribution object is analyzed to obtain the source information of each target power contribution object, including: Obtain the energy contribution information of the power consumption contribution objects of the gate-level netlist within the target time window; The power consumption contribution objects are sorted according to the energy contribution information to obtain a power consumption contribution object sequence; The target power contribution object for the peak power consumption is determined based on the power consumption contribution object sequence; Analyze the activity data source of each target power consumption contribution object to obtain the source information of each target power consumption contribution object.

3. The method according to claim 2, characterized in that, The analysis of the activity data source of each target power consumption contribution object to obtain the source information of each target power consumption contribution object includes: Analyze the active data source of each target power consumption contribution object, and determine the source level of the active data source. The source level includes at least one of the following: active file back-annotation level, propagation completion level, preset compensation level, inference level, and unannotated level. The source level is determined as the source information for each of the target power consumption contribution objects.

4. The method according to claim 3, characterized in that, The determination of the reliability data of the peak power consumption based on the source information includes: Based on the source information, the target power consumption contribution object whose source level is the activity file back-annotation level is determined as a trusted object; Determine the sum of the first energy contribution information of all the trusted objects, and the sum of the second energy contribution information of all the target power consumption contribution objects; The reliability data of the peak power consumption is determined based on the sum of the first energy contribution information and the sum of the second energy contribution information.

5. The method according to claim 1, characterized in that, After determining the reliability data of the peak power consumption based on the source information, the method further includes: If the credibility data meets the credibility threshold condition, the peak power consumption is determined to be approved. If the credibility data does not meet the credibility threshold condition, it is determined that the peak power consumption has not passed the approval.

6. The method according to claim 1, characterized in that, The step of determining the target time window centered on the peak time based on the preset time window half-width includes: Extend the time window forward by half its width, centered on the peak time, to obtain the starting time after the peak time has been extended; The time window is extended backward from the peak time to obtain the extended end time of the peak time. Based on the start time and the end time, a target time window centered on the peak time is determined.

7. A device for determining the reliability of peak power consumption, characterized in that, include: The acquisition module is configured to acquire the peak power consumption of the chip and the peak time corresponding to the peak power consumption when the back-scaling rate of the chip's switching activity file to the gate-level netlist does not reach the back-scaling rate threshold. The determination module is configured to determine a target time window centered on the peak moment based on a preset time window half-width; The analysis module is configured to, within the target time window, determine the target power consumption contribution object with peak power consumption from the power consumption contribution objects according to the energy contribution information of the power consumption contribution objects in the gate-level netlist, analyze the activity data source of each target power consumption contribution object, and obtain the source information of each target power consumption contribution object. The power consumption contribution object is the network and cell that contributes dynamic power consumption in the gate-level netlist, and the energy contribution information is the value of the dynamic energy contributed by the power consumption contribution object within the target time window. The determination module is also configured to determine the credibility data of the peak power consumption based on the source information, the credibility data being used to determine whether the peak power consumption has passed the approval process.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

9. An electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

10. A computer program product, the computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images