CN109726413A

CN109726413A - A method and system for accelerating circuit optimization

Info

Publication number: CN109726413A
Application number: CN201711048112.4A
Authority: CN
Inventors: 吴玉平; 陈岚; 张学连; 孙旭
Original assignee: Institute of Microelectronics of CAS
Current assignee: Institute of Microelectronics of CAS
Priority date: 2017-10-31
Filing date: 2017-10-31
Publication date: 2019-05-07
Anticipated expiration: 2037-10-31
Also published as: CN109726413B

Abstract

The invention discloses a method and system for accelerating circuit optimization. The circuit to be optimized is initialized to generate a set of preferred solutions, a new set of candidate solutions is generated according to the set of preferred solutions, and each candidate solution in the set of candidate solutions is used as a design parameter value, and input To the pre-established macro model representing the corresponding relationship between the design parameter value and the circuit performance value, obtain the circuit performance value corresponding to each candidate solution, and update the optimal circuit to be optimized according to the circuit performance value corresponding to each candidate solution. When the optimal solution meets the preset design index, the optimal solution is output. When the optimal solution does not meet the preset design index, continue to generate a new candidate solution set according to the updated preferred solution set until The obtained optimal solution satisfies the preset design criteria. Therefore, the present invention accelerates the acquisition of the circuit performance value of the candidate solution by applying the created macro model to evaluating the circuit performance value corresponding to the candidate solution, thereby shortening the entire circuit optimization cycle.

Description

A kind of method and system of accelerating circuit optimization

Technical field

The present invention relates to electronic design arts, more specifically, being related to a kind of method of accelerating circuit optimization and being System.

Background technique

In Electronic Design process, circuit optimization is to ensure that and is promoted the significant process of circuit performance.Circuit optimization at present Be divided into two classes: one kind is design engineer to circuit artificial optimization, and another kind of is to rely on computer program to circuit Automatic Optimal. Artificial optimization relies on the professional knowledge and design experiences of design engineer, and optimal speed is slow, and design efficiency is low, leads to entire circuit Optimizing cycle is longer.Automatic Optimal process specifically includes that initialization circuit to be optimized generates preferred solution set；According to preferred disaggregation Close the new candidate solution set automatically generated；Evaluate and test circuit performance value corresponding to each candidate solution in candidate solution set；According to each Circuit performance value updates the optimal solution and preferred solution set of circuit to be optimized；It repeats the above process and refers to until obtaining meeting designing Target optimal solution, and export the optimal solution.

In each step of Automatic Optimal circuitry processes, single evaluates and tests circuit corresponding to each candidate solution in candidate solution set Longest the time required to performance, when accurately evaluating and testing circuit performance corresponding to candidate solution especially with circuit simulation, institute's used time Between occupy the major part of the sum of each Step Time, also, the probability that trivial solution generates in entire iteration optimization cyclic process compared with Height, so that loop iteration low efficiency, leads to loop iteration often, it is longer so as to cause the entire circuit optimization period.

Summary of the invention

In view of this, the present invention discloses a kind of method and system of accelerating circuit optimization, to realize the electricity for accelerating candidate solution The acquisition of road performance number, namely the time used in circuit performance value corresponding to single evaluation and test candidate solution is reduced, to reduce entire Circuit performance evaluates and tests the time in iteration optimization cyclic process, improves loop iteration efficiency, shortens the entire circuit optimization period.

A kind of method of accelerating circuit optimization, comprising:

It initializes circuit to be optimized and generates preferred solution set；

New candidate solution set is generated according to the preferred solution set；

Using each candidate solution in the candidate solution set as design parameter value, it is input to the macro model pre-established, Obtain the corresponding circuit performance value of each candidate solution, wherein the macro model characterization design parameter value and circuit performance The corresponding relationship of value；

According to the corresponding circuit performance value of each candidate solution, optimal solution and the institute of the circuit to be optimized are updated State preferred solution set；

When the optimal solution meets preset design index, the optimal solution is exported；

When the optimal solution is unsatisfactory for the preset design index, continue to be generated newly according to the preferred solution set of update Candidate solution set, until obtained optimal solution meets the preset design index.

Preferably, when the macro model is the inquiry table formed by design parameter value and circuit performance value, by the time Each candidate solution in choosing solution set is input to the macro model pre-established, obtains each candidate as design parameter value Corresponding circuit performance value is solved, is specifically included:

Using each candidate solution in the candidate solution set as design parameter value, it is input to the macro model pre-established；

When finding the corresponding circuit performance value of the candidate solution, the then electricity that will be found in the macro model Circuit performance value of the road performance number as the candidate solution；

When not finding the corresponding circuit performance value of the candidate solution in the macro model, then in the macro model It searches the N number of of the candidate solution and closes on the corresponding circuit performance value of candidate solution, according to the candidate solution and N number of described close on time Select solution distance difference and it is N number of it is described close on the corresponding circuit performance value of candidate solution, it is opposite that the candidate solution is calculated The circuit performance value answered, wherein N is positive integer.

Preferably, after step generates new candidate solution set according to the preferred solution set, by the candidate solution Each candidate solution in set is input to the macro model pre-established as design parameter value, obtains each candidate solution phase Before corresponding circuit performance value, further includes:

The corresponding circuit performance value of each candidate solution in the candidate solution set is evaluated and tested using circuit simulation；

Based on each candidate solution and corresponding circuit performance value, the macro model is established.

Preferably, it in step using each candidate solution in the candidate solution set as design parameter value, is input in advance The macro model of foundation, after obtaining the corresponding circuit performance value of each candidate solution, further includes:

Each candidate solution of acquisition and corresponding circuit performance value are added to institute in the form of corresponding relationship Macro model is stated, increment type creates to obtain newest macro model, and the macro model is substituted for the newest macro model.

Preferably, it using each candidate solution in the candidate solution set as design parameter value, is input to and pre-establishes Macro model obtains the corresponding circuit performance value of each candidate solution, specifically includes:

Using each candidate solution in the candidate solution set as design parameter value, it is input to the macro model pre-established, Obtain the corresponding circuit performance value of each candidate solution；

Whether within a preset range to judge each circuit performance value obtained；

When the circuit performance value is not in the preset range, then using the circuit performance value as corresponding time Select the objective circuit performance number of solution；

When the circuit performance value is in the preset range, the circuit performance value of being tested and assessed again using circuit simulation The circuit performance value of corresponding candidate solution, and using the circuit performance value tested and assessed using the circuit simulation as corresponding The objective circuit performance number of candidate solution；

According to the corresponding circuit performance value of each candidate solution, optimal solution and the institute of the circuit to be optimized are updated Preferred solution set is stated, is specifically included:

According to the corresponding objective circuit performance number of each candidate solution, update the optimal solution of the circuit to be optimized with And the preferred solution set.

Preferably, the preset range are as follows:

|(Preal–Pspec)/Pspec|<Rerr；

Wherein, Preal is that gained circuit performance value is evaluated and tested based on the macro model, and Pspec is target capabilities value, and Rerr is The relative error magnitudes of circuit performance index item.

Preferably, the preset range are as follows:

PboundL≤Preal≤Pspec；

Or, Pspec≤Preal≤PboundR

Wherein, Preal is that gained circuit performance value is evaluated and tested based on the macro model, and Pspec is target capabilities value, PboundL is the lower limit of circuit performance index item, and PboundR is the upper limit of circuit performance index item.

Preferably, new candidate solution set is generated according to the preferred solution set, specifically included:

Based on the macro model, the corresponding circuit performance value of each candidate solution in the candidate solution set is obtained；

When the circuit performance value is not in the preset range, give up the corresponding candidate of the circuit performance value Solution updates the preferred solution set, and generates new candidate solution set according to the preferred solution set of update；

When the circuit performance value is in the preset range, using the corresponding candidate solution of the circuit performance value as Target candidate solution.

A kind of system of accelerating circuit optimization, comprising:

Initialization unit generates preferred solution set for initializing circuit to be optimized；

Candidate solution generation unit, for generating new candidate solution set according to the preferred solution set；

Circuit performance value generation unit, for using each candidate solution in the candidate solution set as design parameter value, It is input to the macro model pre-established, obtains the corresponding circuit performance value of each candidate solution, wherein the macro model table Levy the corresponding relationship of design parameter value and circuit performance value；

Updating unit, for updating the circuit to be optimized according to the corresponding circuit performance value of each candidate solution Optimal solution and preferred solution set；

Output unit, for exporting the optimal solution when the optimal solution meets preset design index；

Return unit is generated for when the optimal solution is unsatisfactory for the preset design index, returning to the candidate solution Unit is continued to generate new candidate solution set according to the preferred solution set of update by the candidate solution generation unit, until obtaining Optimal solution meet the preset design index.

Preferably, the circuit performance value generation unit is specifically used for:

When the macro model is the inquiry table formed by design parameter value and circuit performance value, by the candidate solution set In each candidate solution as design parameter value, be input to the macro model pre-established；

Preferably, further includes:

Unit is evaluated and tested, after generating new candidate solution set in the candidate solution generation unit, in the circuit Generation unit can be worth using each candidate solution in the candidate solution set as design parameter value, be input to the macro mould pre-established Type is evaluated and tested in the candidate solution set before obtaining the corresponding circuit performance value of each candidate solution using circuit simulation The corresponding circuit performance value of each candidate solution；

Macro model establishes unit, for being based on each candidate solution and corresponding circuit performance value, described in foundation Macro model.

Preferably, further includes:

Macro model replacement unit, in the circuit performance value generation unit by each time in the candidate solution set Choosing solution is used as design parameter value, is input to the macro model pre-established, obtains the corresponding circuit performance of each candidate solution After value, each candidate solution of acquisition and corresponding circuit performance value are added in the form of corresponding relationship described Macro model, increment type creates to obtain newest macro model, and the macro model is substituted for the newest macro model.

Preferably, the circuit performance value generation unit specifically includes:

First obtains subelement, for using each candidate solution in the candidate solution set as design parameter value, input To the macro model pre-established, the corresponding circuit performance value of each candidate solution is obtained；

First judgment sub-unit, within a preset range whether each circuit performance value for judging to obtain；

First determines subelement, in the case where first judgment sub-unit is judged as NO, by the circuit It can objective circuit performance number of the value as corresponding candidate solution；

Second determines subelement, in the case where first judgment sub-unit is judged as YES, using circuit simulation It tests and assesses again the circuit performance value of the corresponding candidate solution of the circuit performance value, and will test and assess to obtain using the circuit simulation Objective circuit performance number of the circuit performance value as corresponding candidate solution；

The updating unit includes:

Subelement is updated, for updating described to excellent according to the corresponding objective circuit performance number of each candidate solution Change the optimal solution and the preferred solution set of circuit.

Preferably, the preset range are as follows:

|(Preal–Pspec)/Pspec|<Rerr；

Preferably, the preset range are as follows:

PboundL≤Preal≤Pspec；

Or, Pspec≤Preal≤PboundR

Preferably, the candidate solution generation unit specifically includes:

Subelement is generated, for generating new candidate solution set according to the preferred solution set；

Second obtains subelement, for being based on the macro model, obtains each candidate solution phase in the candidate solution set Corresponding circuit performance value；

Second judgment sub-unit, within a preset range whether each circuit performance value for judging to obtain；

Give up subelement, for giving up the circuit performance value in the case where the second judgment unit is judged as NO Corresponding candidate solution updates the preferred solution set, and generates new candidate solution set according to the preferred solution set of update；

Subelement is chosen, in the case where the second judgment unit is judged as YES, by the circuit performance value phase Corresponding candidate solution is as target candidate solution.

From above-mentioned technical solution it is found that the invention discloses a kind of method and system of accelerating circuit optimization, initialization Circuit to be optimized generates preferred solution set, generates new candidate solution set according to the preferred solution set, will be in candidate solution set Each candidate solution is input to the corresponding relationship of the characterization design parameter value and circuit performance value pre-established as design parameter value Macro model, obtain the corresponding circuit performance value of each candidate solution, according to the corresponding circuit performance value of each candidate solution, more The optimal solution of new circuit to be optimized and preferred solution set export optimal solution, instead when optimal solution meets preset design index It continues to generate new candidate disaggregation according to the preferred solution set of update when optimal solution is unsatisfactory for the preset design index It closes, until obtained optimal solution meets preset design index.It can thus be seen that the present invention passes through the macro model application that will be created In the corresponding circuit performance value of evaluation and test candidate solution, the acquisition of the circuit performance value of candidate solution is accelerated, compared to traditional scheme Speech, The present invention reduces the times used in circuit performance value corresponding to single evaluation and test candidate solution, reduce entire iteration optimization circulation The circuit performance evaluation and test time reduces loop iteration number to improve loop iteration efficiency in the process, shortens entire electricity Road optimizing cycle.

Detailed description of the invention

In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this The embodiment of invention for those of ordinary skill in the art without creative efforts, can also basis Disclosed attached drawing obtains other attached drawings.

Fig. 1 is a kind of method flow diagram of accelerating circuit optimization disclosed by the embodiments of the present invention；

Fig. 2 is the method flow diagram of another accelerating circuit optimization disclosed by the embodiments of the present invention；

Fig. 3 is the method stream of circuit performance value corresponding to a kind of adaptive evaluation and test candidate solution disclosed by the embodiments of the present invention Cheng Tu；

Fig. 4 is that a kind of basis disclosed by the embodiments of the present invention preferably solves the method flow that set generates new candidate solution set Figure；

Fig. 5 is a kind of system structure diagram of accelerating circuit optimization disclosed by the embodiments of the present invention；

Fig. 6 is a kind of composed structure schematic diagram of circuit performance value generation unit disclosed by the embodiments of the present invention；

Fig. 7 is a kind of composed structure schematic diagram of candidate solution generation unit disclosed by the embodiments of the present invention.

Specific embodiment

Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other Embodiment shall fall within the protection scope of the present invention.

The embodiment of the invention discloses a kind of method and system of accelerating circuit optimization, to realize the circuit for accelerating candidate solution The acquisition of performance number, namely the time used in circuit performance value corresponding to single evaluation and test candidate solution is reduced, to reduce entire change Circuit performance evaluates and tests the time in generation optimization cyclic process, improves loop iteration efficiency, reduces loop iteration number, shortens entire electricity Road optimizing cycle.

Referring to Fig. 1, a kind of method flow diagram of accelerating circuit optimization disclosed in one embodiment of the invention, this method includes step It is rapid:

Step S101, it initializes circuit to be optimized and generates preferred solution set；

Wherein, the process that the preferred solution set of circuit generation to be optimized is initialized in the present embodiment can be found in existing scheme use Technological means, details are not described herein again.

Step S102, new candidate solution set is generated according to preferred solution set；

Existing scheme use can be found according to the process that preferred solution set generates new candidate solution set in the present embodiment Technological means, details are not described herein again.

Step S103, it using each candidate solution in the candidate solution set as design parameter value, is input to and pre-establishes Macro model, obtain the corresponding circuit performance value of each candidate solution；

Specifically, macro model can be inquiry table, it is also possible to analytic expression.Macro model characterizes design parameter value and circuit Can value corresponding relationship, input design parameter value in macro model, the corresponding circuit performance value of available design parameter value. Wherein, each candidate solution is a specific value of the design parameter value.

Unlike the prior art, the present invention is generated according to optimization process intermediate data (specially candidate solution and The corresponding circuit performance value of the candidate solution) macro model is automatically created, macro model can be according to the design parameter value of input, quickly The corresponding circuit performance value of the design parameter value is obtained, speed is significantly faster than existing scheme and is waited to using circuit simulation Choosing solves the process of corresponding circuit performance value.Therefore the macro model of creation can be applied to accelerate corresponding to evaluation and test candidate solution The process of circuit performance value reach and mention to reduce the evaluation and test time of circuit performance value in entire iteration optimization cyclic process High circulation iteration efficiency shortens the purpose in entire circuit optimization period.

Step S104, according to the corresponding circuit performance value of each candidate solution, the circuit to be optimized is updated most Excellent solution and the preferred solution set；

Specifically, in practical applications, can be by judging it is pre- whether the corresponding circuit performance value of each candidate solution meets If condition, for example, whether circuit performance value within the scope of some or outside the range, to determine that the circuit performance value is corresponding Candidate solution whether be that the optimal solution of circuit to be optimized determines the circuit when some circuit performance value meets preset condition It is optimal solution that corresponding candidate solution, which can be worth,.

Wherein, the preferred disaggregation updated in this step is combined into the preferred solution set in step S101, at this point, step S102 root New candidate solution set is generated according to the new preferred solution set.

Step S105, judge whether the candidate solution meets preset design index, if so, thening follow the steps S106, instead It, return step S102 generates new candidate solution set according to the preferred solution set of update；

Wherein, depending on preset design index concrete foundation actual needs, the present invention is it is not limited here.

Step S106, the optimal solution is exported.

In summary, the method for accelerating circuit optimization disclosed by the invention, initializes circuit to be optimized and generates preferred disaggregation It closes, new candidate solution set is generated according to the preferred solution set, using each candidate solution in candidate solution set as design parameter Value, is input to the macro model of the corresponding relationship of the characterization design parameter value and circuit performance value pre-established, obtains each candidate Corresponding circuit performance value is solved, according to the corresponding circuit performance value of each candidate solution, updates the optimal solution of circuit to be optimized And preferably solution set exports optimal solution when optimal solution meets preset design index, conversely, described in being unsatisfactory for when optimal solution When preset design index, continue to generate new candidate solution set according to the preferred solution set of update, until obtained optimal solution is full Sufficient preset design index.It can thus be seen that the present invention is corresponding applied to evaluation and test candidate solution by the macro model that will be created Circuit performance value accelerates the acquisition of the circuit performance value of candidate solution, and compared to for traditional scheme, The present invention reduces singles to comment The time used in circuit performance value corresponding to candidate solution is surveyed, when reducing circuit performance evaluation and test in entire iteration optimization cyclic process Between, to improve loop iteration efficiency, reduce loop iteration number, shortens the entire circuit optimization period.

It should be noted that macro model is after creation, it is subsequent to be multiplexed.Therefore, acceleration can be executed in first time When circuit optimization process, macro model is created.Therefore, when first time executing accelerating circuit Optimizing Flow, in the above-described embodiments Step S102 after, before step S103, further includes:

Using the corresponding circuit performance value of candidate solution each in circuit simulation evaluation and test candidate solution set；Based on each described Candidate solution and corresponding circuit performance value, establish the macro model.

Wherein, using the process of the corresponding circuit performance value of each candidate solution in circuit simulation evaluation and test candidate solution set, It can be found in existing scheme, details are not described herein again.

Specifically, in using circuit simulation evaluation and test candidate solution set the corresponding circuit performance value of each candidate solution it Afterwards, according to the one-to-one relationship of candidate solution and circuit performance value each in candidate solution set, creation describes this one-to-one correspondence The macro model of relationship, circuit performance value corresponding to the intermediate variable candidate solution for subsequent quick evaluation circuit design.

In practical applications, macro model can be inquiry table, be also possible to analytic expression.For inquiring the macro mould of sheet form Type when the circuit performance value corresponding to the design parameter value for finding input, directly exports the circuit performance value found； When the circuit performance value corresponding to the design parameter value for not finding input, it is N number of that the design parameter value is searched in inquiry table The corresponding circuit performance value of design parameter value is closed on, and according to N number of range difference for closing on design parameter value and design parameter value Value, is calculated the corresponding circuit performance of design parameter value in conjunction with N number of corresponding circuit performance value of design parameter value of closing on Value.

It should be noted that candidate solution is the specific value of design parameter value in macro model in the present invention.

Therefore, in order to further optimize the above embodiments, step S103 is specifically included:

In summary, when macro model is look-up table, the circuit performance value corresponding to the candidate solution for not finding input When, the candidate solution is N number of to close on the corresponding circuit performance value of candidate solution to the present invention by searching in inquiry table, and according to N number of Time is calculated in conjunction with N number of corresponding circuit performance value of candidate solution of closing in the distance difference for closing on candidate solution Yu the candidate solution Choosing solves corresponding circuit performance value and therefore substantially increases the reliable of the corresponding circuit performance value of evaluated and tested candidate solution Property.

In view of in circuit design iterative optimization procedure, constantly having new candidate solution and the corresponding essence of part candidate solution True circuit performance value needs to join macro model to design to keep the assessment of the circuit performance value based on macro model more accurate Numerical value and the description of the corresponding relationship of circuit performance value are more accurate, therefore macro mould can be dynamically updated in circuit design iterative process Type makes macro model accurately reflect the candidate solution being newly included in and its circuit performance value.

Referring to fig. 2, a kind of method flow diagram of accelerating circuit optimization, this method disclosed in another embodiment of the present invention include Step:

Step S201, it initializes circuit to be optimized and generates preferred solution set；

Step S202, new candidate solution set is generated according to preferred solution set；

Step S203, it using each candidate solution in the candidate solution set as design parameter value, is input to and pre-establishes Macro model, obtain the corresponding circuit performance value of each candidate solution；

Step S204, by each candidate solution of acquisition and corresponding circuit performance value in the form of corresponding relationship It is added to the macro model, increment type creates to obtain newest macro model, and the macro model is substituted for the newest macro model, And return to step S203；

Step S205, according to the corresponding circuit performance value of each candidate solution, the circuit to be optimized is updated most Excellent solution and the preferred solution set；

Step S206, judge whether the candidate solution meets preset design index, if so, thening follow the steps S207, instead It, return step S202 generates new candidate solution set according to the preferred solution set of update；

Step S207, the optimal solution is exported.

In summary, the present invention generates the candidate solution of several optimization circuits before step S206 Optimized Iterative, Circuit performance assessment is carried out to each candidate solution based on macro model, obtains the corresponding circuit performance value of each candidate solution.It changes in optimization During generation, new candidate solution and corresponding circuit performance value are added in macro model, the creation of increment type obtains most New macro model, and the newest macro model is replaced into original macro model, used in the circuit performance evaluation and test for subsequent design candidate solution, To provide more accurate evaluation and test.The present invention accelerates electricity corresponding to evaluation and test candidate solution by the way that newly created macro model to be applied to Road performance number reduces the time used in circuit performance value corresponding to single evaluation and test candidate solution, to reduce entire iteration optimization Circuit performance evaluates and tests the time in cyclic process, improves loop iteration efficiency, reduces loop iteration number, shortens entire electricity Road optimizing cycle.

It, may further unlike technological means used by circuit performance value corresponding to front evaluation and test candidate solution Using circuit performance value namely above-described embodiment step S103 and step S203 corresponding to adaptive approach evaluation and test candidate solution In, using each candidate solution in the candidate solution set as design parameter value, it is input to the macro model pre-established, is obtained every The corresponding circuit performance value of a candidate solution, can be using circuit performance corresponding to adaptive approach evaluation and test candidate solution Value, to reduce the time used in single evaluation circuit performance.

Therefore, in order to further optimize the above embodiments, referring to Fig. 3, one kind disclosed in one embodiment of the invention is adaptively commented The method flow diagram of circuit performance value corresponding to candidate solution is surveyed, this method namely step S103 and step S203 specifically include:

Step S301, using each candidate solution in candidate solution set as design parameter value, be input to pre-establish it is macro Model obtains the corresponding circuit performance value of each candidate solution；

Step S302, whether within a preset range each circuit performance value obtained is judged, if it is not, then executing step Otherwise rapid S303 executes step S304；

Wherein, the judgment criteria of the preset range in this step namely circuit performance value may is that

|(Preal–Pspec)/Pspec|<Rerr；

In formula, Preal is that gained circuit performance value is evaluated and tested based on the macro model, and Pspec is target capabilities value, and Rerr is The relative error magnitudes of circuit performance index item, the value can be fixed value, it also may be preferable for be dynamic value, with iterative optimization procedure by Step reduces.

Preset range is also possible to: PboundL≤Preal≤Pspec；Or, Pspec≤Preal≤PboundR；

Wherein, Preal is that gained circuit performance value is evaluated and tested based on the macro model, and Pspec is target capabilities value, PboundL is the lower limit of circuit performance index item, and PboundR is the upper limit of circuit performance index item, PboundL and PboundR Value can be fixed value, it also may be preferable for for dynamic value, as iterative optimization procedure gradually levels off to Pspec.

Step S303, using the circuit performance value as the objective circuit performance number of corresponding candidate solution；

Step S304, it is tested and assessed again using circuit simulation the circuit performance of the corresponding candidate solution of the circuit performance value Value, and using the circuit performance value tested and assessed using the circuit simulation as the objective circuit performance number of corresponding candidate solution.

It should be noted that after obtaining the objective circuit performance number of each candidate solution using the present embodiment, correspondingly, Step S104 and step S205 in above-described embodiment are specifically included:

In summary, the method for circuit performance value corresponding to adaptive evaluation and test candidate solution disclosed by the invention, effectively subtracts The probability for having lacked the generation of invalid candidate solution, improves the validity of evaluation circuit performance, to improve loop iteration efficiency, subtracts Lack loop iteration number, shortens the entire circuit optimization period.

It in a preferred embodiment, can be by the preferred solution set of basis shown in macro model and traditional namely step S102 It generates new both candidate solution set to combine, obtains a kind of scheme of new selection candidate solution set.

Referring to fig. 4, a kind of basis disclosed in one embodiment of the invention preferably solves the side that set generates new candidate solution set Method flow chart, the method comprising the steps of:

Step S401, new candidate solution set is generated according to preferred solution set；

Step S402, it is based on macro model, obtains the corresponding circuit performance of each candidate solution in the candidate solution set Value；

Step S403, whether within a preset range each circuit performance value obtained is judged, if it is, executing step Otherwise rapid S404 executes step S405；

Wherein, the preset range in this step may is that | (Preal-Pspec)/Pspec | < Rerr；

Preset range is also possible to: PboundL≤Preal≤Pspec；Also or, Pspec≤Preal≤PboundR；

Step S404, using the corresponding candidate solution of the circuit performance value as target candidate solution；

Step S405, give up the corresponding candidate solution of the circuit performance value, update and initialize what circuit to be optimized generated It is preferred that solution set, and return step S401, new candidate solution set is generated according to the preferred solution set of update.

In summary, the present invention is by generating new candidate solution set for macro model and the preferred solution set of traditional basis Method combines, and effectively reduces the probability of invalid candidate solution generation, the validity of evaluation circuit performance number is improved, to mention High loop iteration efficiency, reduces loop iteration number, shortens the entire circuit optimization period, and reduce design cost.

In order to further optimize the above embodiments, in the scheme of another preferred accelerating circuit optimization, it can be further by Fig. 3 The method of circuit performance value corresponding to the adaptive evaluation and test candidate solution of the one kind shown and the preferred solution set of basis shown in Fig. 4 produce The method of raw new candidate solution set combines, and the program specifically includes:

S1, initialization circuit to be optimized generate preferred solution set；

S2, candidate solution set is obtained using scheme disclosed in embodiment illustrated in fig. 4；

S3, the corresponding circuit of each candidate solution in candidate solution set is obtained using scheme disclosed in embodiment illustrated in fig. 3 Performance number；

S4, according to the corresponding circuit performance value of each candidate solution, update the optimal solution of the circuit to be optimized with And the preferred solution set；

S5, judge whether the candidate solution meets preset design index, if it is, S6 is executed, conversely, returning to S2, root New candidate solution set is generated according to the preferred solution set of update；

S6, the output optimal solution.

It should be noted that in the present embodiment, the process of increment type building macro model can also be increased.Wherein, this reality It applies example and obtains the process of macro model for the first time and can be found in above-mentioned corresponding embodiment, details are not described herein again.

In summary, the method for accelerating circuit optimization disclosed by the invention, initializes circuit to be optimized and generates preferred disaggregation It closes, is combined, generated new by the method that macro model and the preferred solution set of traditional basis are generated new candidate solution set It is corresponding to obtain each candidate solution using the method for circuit performance value corresponding to adaptive evaluation and test candidate solution for candidate solution set Circuit performance value the optimal solution and preferably of circuit to be optimized is updated according to the corresponding circuit performance value of each candidate solution Solution set exports optimal solution when optimal solution meets preset design index, conversely, when optimal solution is unsatisfactory for the preset design When index, continue to generate new candidate solution set according to the preferred solution set of update, until obtained optimal solution satisfaction is set Count index.It can thus be seen that the present invention is applied to the corresponding circuit performance value of evaluation and test candidate solution with the macro model created, add The speed acquisition of the circuit performance value of candidate solution, compared to for traditional scheme, The present invention reduces single evaluation and test candidate solution, institute is right Time used in the circuit performance value answered reduces circuit performance in entire iteration optimization cyclic process and evaluates and tests the time, to improve Loop iteration efficiency reduces loop iteration number, shortens the entire circuit optimization period.

Corresponding with above method embodiment, the invention also discloses a kind of systems of accelerating circuit optimization.

Referring to Fig. 5, a kind of system structure diagram of accelerating circuit optimization, the system packet disclosed in one embodiment of the invention It includes:

Initialization unit 501 generates preferred solution set for initializing circuit to be optimized；

Candidate solution generation unit 502, for generating new candidate solution set according to the preferred solution set；

Circuit performance value generation unit 503, for using each candidate solution in the candidate solution set as design parameter Value is input to the macro model pre-established, obtains the corresponding circuit performance value of each candidate solution, wherein the macro mould The corresponding relationship of type characterization design parameter value and circuit performance value；

Updating unit 504, for updating the electricity to be optimized according to the corresponding circuit performance value of each candidate solution The optimal solution on road and the preferred solution set；

Output unit 505, for exporting the optimal solution when the optimal solution meets preset design index；

Return unit 506 generates list for when the optimal solution is unsatisfactory for the preset design index, returning to candidate solution Member 502 is continued to generate new candidate solution set according to the preferred solution set of update by the candidate solution generation unit, until obtaining Optimal solution meet the preset design index.

In summary, the system of accelerating circuit optimization disclosed by the invention, initializes circuit to be optimized and generates preferred disaggregation It closes, new candidate solution set is generated according to the preferred solution set, using each candidate solution in candidate solution set as design parameter Value, is input to the macro model of the corresponding relationship of the characterization design parameter value and circuit performance value pre-established, obtains each candidate Corresponding circuit performance value is solved, according to the corresponding circuit performance value of each candidate solution, updates the optimal solution of circuit to be optimized And preferably solution set exports optimal solution when optimal solution meets preset design index, conversely, described in being unsatisfactory for when optimal solution When preset design index, continue to generate new candidate solution set according to the preferred solution set of update, until obtained optimal solution is full Sufficient preset design index.It can thus be seen that the present invention is corresponding applied to evaluation and test candidate solution by the macro model that will be created Circuit performance value accelerates the acquisition of the circuit performance value of candidate solution, and compared to for traditional scheme, The present invention reduces singles to comment The time used in circuit performance value corresponding to candidate solution is surveyed, when reducing circuit performance evaluation and test in entire iteration optimization cyclic process Between, to improve loop iteration efficiency, reduce loop iteration number, shortens the entire circuit optimization period.

It should be noted that macro model is after creation, it is subsequent to be multiplexed.Therefore, the system that accelerating circuit optimizes is also Include:

Unit is evaluated and tested, after generating new candidate solution set in the candidate solution generation unit 502, in the circuit Performance number generation unit 503 is input to and pre-establishes using each candidate solution in the candidate solution set as design parameter value Macro model, before obtaining the corresponding circuit performance value of each candidate solution, the candidate solution is evaluated and tested using circuit simulation The corresponding circuit performance value of each candidate solution in set；

Therefore, in order to further optimize the above embodiments, circuit performance value generation unit 503 is specifically used for:

In order to further optimize the above embodiments, accelerating circuit optimization further include:

Macro model replacement unit, for will be each in the candidate solution set in the circuit performance value generation unit 503 A candidate solution is input to the macro model pre-established as design parameter value, obtains the corresponding circuit of each candidate solution After performance number, each candidate solution of acquisition and corresponding circuit performance value are added in the form of corresponding relationship The macro model, increment type creates to obtain newest macro model, and the macro model is substituted for the newest macro model.

In summary, the present invention generates the candidate solution of several optimization circuits before Optimized Iterative, is based on macro mould Type carries out circuit performance assessment to each candidate solution, obtains the corresponding circuit performance value of each candidate solution.During Optimized Iterative, New candidate solution and corresponding circuit performance value are added in macro model, the creation of increment type obtains newest macro model, And the newest macro model is replaced into original macro model, it is used in the circuit performance evaluation and test of subsequent design candidate solution, to provide more Accurately to evaluate and test.The present invention accelerates circuit performance corresponding to evaluation and test candidate solution by the way that newly created macro model to be applied to Value reduces the time used in circuit performance value corresponding to single evaluation and test candidate solution, is circulated throughout to reduce entire iteration optimization Circuit performance evaluates and tests the time in journey, improves loop iteration efficiency, reduces loop iteration number, shortens entire circuit optimization Period.

It, may further unlike technological means used by circuit performance value corresponding to front evaluation and test candidate solution Using circuit performance value generation unit in circuit performance value corresponding to adaptive approach evaluation and test candidate solution namely above-described embodiment 503 using each candidate solution in the candidate solution set as design parameter value, is input to the macro model pre-established, obtains every The process of the corresponding circuit performance value of a candidate solution, can be using circuit corresponding to adaptive approach evaluation and test candidate solution Performance number, to reduce the time used in single evaluation circuit performance.

Therefore, in order to further optimize the above embodiments, referring to Fig. 6, a kind of circuit performance disclosed in one embodiment of the invention It is worth the composed structure schematic diagram of generation unit,

Circuit performance value generation unit specifically includes:

First obtains subelement 601, defeated for using each candidate solution in the candidate solution set as design parameter value Enter to the macro model pre-established, obtains the corresponding circuit performance value of each candidate solution；

First judgment sub-unit 602, within a preset range whether each circuit performance value for judging to obtain；

Wherein, the judgment criteria of the preset range in the present embodiment namely circuit performance value may is that

|(Preal–Pspec)/Pspec|<Rerr；

First determines subelement 603, will be described in the case where first judgment sub-unit 603 is judged as NO Objective circuit performance number of the circuit performance value as corresponding candidate solution；

Second determines subelement 604, in the case where first judgment sub-unit 603 is judged as YES, using electricity Road emulates the circuit performance value for the corresponding candidate solution of the circuit performance value of testing and assessing again, and will be surveyed using the circuit simulation Objective circuit performance number of the circuit performance value commented as corresponding candidate solution.

Based on the present embodiment, updating unit 504 includes:

In a preferred embodiment, macro model and the preferred solution set of traditional basis can be generated into new candidate solution set The two combines, and obtains a kind of scheme of new selection candidate solution set.

Referring to Fig. 7, a kind of composed structure schematic diagram of candidate solution generation unit, candidate solution disclosed in one embodiment of the invention Generation unit includes:

Subelement 701 is generated, for generating new candidate solution set according to the preferred solution set；

Second obtains subelement 702, for being based on the macro model, obtains each candidate solution in the candidate solution set Corresponding circuit performance value；

Second judgment sub-unit 703, within a preset range whether each circuit performance value for judging to obtain；

Wherein, the preset range in the present embodiment may is that | (Preal-Pspec)/Pspec | < Rerr；

Give up subelement 704, for giving up the circuit in the case where the second judgment unit 703 is judged as NO The corresponding candidate solution of performance number updates the preferred solution set, and generates new candidate solution according to the preferred solution set of update Set；

Subelement 705 is chosen, in the case where the second judgment unit 703 is judged as YES, by the circuit Corresponding candidate solution can be worth as target candidate solution.

In order to further optimize the above embodiments, in the scheme of another preferred accelerating circuit optimization, it can be further by Fig. 6 Disclosed circuit performance value generation unit and the candidate solution generation unit shown in Fig. 7 combine, and the program specifically includes:

Candidate solution generation unit specifically obtains candidate solution set using candidate solution generation unit shown in Fig. 7；

Circuit performance value generation unit specifically obtains candidate solution set using circuit performance value generation unit described in Fig. 6 In the corresponding circuit performance value of each candidate solution；

Return unit, for returning to candidate solution generation unit when the optimal solution is unsatisfactory for the preset design index, Continued to generate new candidate solution set according to the preferred solution set of update by the candidate solution generation unit, until obtaining optimal Solution meets the preset design index.

In summary, the system of accelerating circuit optimization disclosed by the invention, initializes circuit to be optimized and generates preferred disaggregation It closes, is combined, generated new by the method that macro model and the preferred solution set of traditional basis are generated new candidate solution set It is corresponding to obtain each candidate solution using the method for circuit performance value corresponding to adaptive evaluation and test candidate solution for candidate solution set Circuit performance value the optimal solution and preferably of circuit to be optimized is updated according to the corresponding circuit performance value of each candidate solution Solution set exports optimal solution when optimal solution meets preset design index, conversely, when optimal solution is unsatisfactory for the preset design When index, continue to generate new candidate solution set according to the preferred solution set of update, until obtained optimal solution satisfaction is set Count index.It can thus be seen that the present invention is applied to the corresponding circuit performance value of evaluation and test candidate solution with the macro model created, add The speed acquisition of the circuit performance value of candidate solution, compared to for traditional scheme, The present invention reduces single evaluation and test candidate solution, institute is right Time used in the circuit performance value answered reduces circuit performance in entire iteration optimization cyclic process and evaluates and tests the time, to improve Loop iteration efficiency reduces loop iteration number, shortens the entire circuit optimization period.

It should be noted that in system embodiment each component part concrete operating principle, refer to embodiment of the method pair Part is answered, details are not described herein again.

Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that A little elements, but also including other elements that are not explicitly listed, or further include for this process, method, article or The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged Except there is also other identical elements in the process, method, article or apparatus that includes the element.

Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other The difference of embodiment, the same or similar parts in each embodiment may refer to each other.

The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the invention It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one The widest scope of cause.

Claims

1. a kind of method of accelerating circuit optimization characterized by comprising

It initializes circuit to be optimized and generates preferred solution set；

Using each candidate solution in the candidate solution set as design parameter value, it is input to the macro model pre-established, is obtained The corresponding circuit performance value of each candidate solution, wherein the macro model characterization design parameter value and circuit performance value Corresponding relationship；

According to the corresponding circuit performance value of each candidate solution, the optimal solution of the circuit to be optimized and described excellent is updated Select Xie Jihe；

When the optimal solution is unsatisfactory for the preset design index, continue to generate new candidate according to the preferred solution set of update Solution set, until obtained optimal solution meets the preset design index.

2. the method according to claim 1, wherein when the macro model is by design parameter value and circuit performance When being worth the inquiry table formed, using each candidate solution in the candidate solution set as design parameter value, it is input to and pre-establishes Macro model, obtain the corresponding circuit performance value of each candidate solution, specifically include:

When finding the corresponding circuit performance value of the candidate solution, the then circuit that will be found in the macro model It can circuit performance value of the value as the candidate solution；

When not finding the corresponding circuit performance value of the candidate solution in the macro model, then searched in the macro model The N number of of the candidate solution closes on the corresponding circuit performance value of candidate solution, according to the candidate solution and N number of described closes on candidate solution Distance difference and it is N number of it is described close on the corresponding circuit performance value of candidate solution, it is corresponding that the candidate solution is calculated Circuit performance value, wherein N is positive integer.

3. the method according to claim 1, wherein generating new candidate according to the preferred solution set in step Solution set after, using each candidate solution in the candidate solution set as design parameter value, be input to pre-establish it is macro Model, before obtaining the corresponding circuit performance value of each candidate solution, further includes:

4. the method according to claim 1, wherein in step by each candidate solution in the candidate solution set As design parameter value, be input to the macro model pre-established, obtain the corresponding circuit performance value of each candidate solution it Afterwards, further includes:

Each candidate solution of acquisition and corresponding circuit performance value are added in the form of corresponding relationship described macro Model, increment type creates to obtain newest macro model, and the macro model is substituted for the newest macro model.

5. the method according to claim 1, wherein using each candidate solution in the candidate solution set as setting Parameter value is counted, the macro model pre-established is input to, obtains the corresponding circuit performance value of each candidate solution, it is specific to wrap It includes:

Using each candidate solution in the candidate solution set as design parameter value, it is input to the macro model pre-established, is obtained The corresponding circuit performance value of each candidate solution；

When the circuit performance value is not in the preset range, then using the circuit performance value as corresponding candidate solution Objective circuit performance number；

When the circuit performance value is in the preset range, the circuit performance value of being tested and assessed again using circuit simulation is opposite The circuit performance value for the candidate solution answered, and using the circuit performance value tested and assessed using the circuit simulation as corresponding candidate The objective circuit performance number of solution；

According to the corresponding circuit performance value of each candidate solution, the optimal solution of the circuit to be optimized and described excellent is updated Xie Jihe is selected, is specifically included:

According to the corresponding objective circuit performance number of each candidate solution, optimal solution and the institute of the circuit to be optimized are updated State preferred solution set.

6. according to the method described in claim 5, it is characterized in that, the preset range are as follows:

|(Preal–Pspec)/Pspec|<Rerr；

Wherein, Preal is that gained circuit performance value is evaluated and tested based on the macro model, and Pspec is target capabilities value, and Rerr is circuit The relative error magnitudes of performance indicator item.

7. according to the method described in claim 5, it is characterized in that, the preset range are as follows:

PboundL≤Preal≤Pspec；

Or, Pspec≤Preal≤PboundR

Wherein, Preal is that gained circuit performance value is evaluated and tested based on the macro model, and Pspec is target capabilities value, and PboundL is The lower limit of circuit performance index item, PboundR are the upper limit of circuit performance index item.

8. the method according to claim 1, wherein generating new candidate disaggregation according to the preferred solution set It closes, specifically includes:

When the circuit performance value is not in the preset range, give up the corresponding candidate solution of the circuit performance value, The preferred solution set is updated, and new candidate solution set is generated according to the preferred solution set of update；

9. a kind of system of accelerating circuit optimization characterized by comprising

Circuit performance value generation unit, for using each candidate solution in the candidate solution set as design parameter value, input To the macro model pre-established, the corresponding circuit performance value of each candidate solution is obtained, wherein the macro model characterization is set Count the corresponding relationship of parameter value and circuit performance value；

Updating unit, for updating the circuit to be optimized most according to the corresponding circuit performance value of each candidate solution Excellent solution and the preferred solution set；

Return unit, for returning to the candidate solution generation unit when the optimal solution is unsatisfactory for the preset design index, Continued to generate new candidate solution set according to the preferred solution set of update by the candidate solution generation unit, until obtaining optimal Solution meets the preset design index.

10. system according to claim 9, which is characterized in that the circuit performance value generation unit is specifically used for:

It, will be in the candidate solution set when the macro model is the inquiry table formed by design parameter value and circuit performance value Each candidate solution is input to the macro model pre-established as design parameter value；

11. system according to claim 9, which is characterized in that further include:

Unit is evaluated and tested, after generating new candidate solution set in the candidate solution generation unit, in the circuit performance value Generation unit is input to the macro model pre-established using each candidate solution in the candidate solution set as design parameter value, Before obtaining the corresponding circuit performance value of each candidate solution, evaluated and tested using circuit simulation each in the candidate solution set The corresponding circuit performance value of the candidate solution；

Macro model establishes unit, for being based on each candidate solution and corresponding circuit performance value, establishes the macro mould Type.

12. system according to claim 9, which is characterized in that further include:

Macro model replacement unit, in the circuit performance value generation unit by each candidate solution in the candidate solution set As design parameter value, be input to the macro model pre-established, obtain the corresponding circuit performance value of each candidate solution it Afterwards, each candidate solution of acquisition and corresponding circuit performance value are added to the macro mould in the form of corresponding relationship Type, increment type creates to obtain newest macro model, and the macro model is substituted for the newest macro model.

13. system according to claim 9, which is characterized in that the circuit performance value generation unit specifically includes:

First obtains subelement, for being input to pre- using each candidate solution in the candidate solution set as design parameter value The macro model first established obtains the corresponding circuit performance value of each candidate solution；

First determines subelement, in the case where first judgment sub-unit is judged as NO, by the circuit performance value Objective circuit performance number as corresponding candidate solution；

Second determines subelement, in the case where first judgment sub-unit is judged as YES, again using circuit simulation It tests and assesses the circuit performance value of the corresponding candidate solution of the circuit performance value, and the electricity that will be tested and assessed using the circuit simulation Objective circuit performance number of the road performance number as corresponding candidate solution；

The updating unit includes:

Subelement is updated, for updating the electricity to be optimized according to the corresponding objective circuit performance number of each candidate solution The optimal solution on road and the preferred solution set.

14. system according to claim 13, which is characterized in that the preset range are as follows:

|(Preal–Pspec)/Pspec|<Rerr；

15. system according to claim 13, which is characterized in that the preset range are as follows:

PboundL≤Preal≤Pspec；

Or, Pspec≤Preal≤PboundR

16. system according to claim 9, which is characterized in that the candidate solution generation unit specifically includes:

Second obtains subelement, and for being based on the macro model, each candidate solution obtained in the candidate solution set is corresponding Circuit performance value；

Give up subelement, for it is opposite to give up the circuit performance value in the case where the second judgment unit is judged as NO The candidate solution answered updates the preferred solution set, and generates new candidate solution set according to the preferred solution set of update；

Subelement is chosen, it is in the case where the second judgment unit is judged as YES, the circuit performance value is corresponding Candidate solution as target candidate solution.