JP3367869B2 - Wiring reduction processing method, wiring reduction processing device, and storage medium for storing wiring reduction processing program - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wiring shortening method and wiring for shortening the total wiring length in consideration of priority, which is carried out following an area reduction process when one-dimensionally compacting layout data of a semiconductor integrated circuit. Shortening device, and
The present invention relates to a storage medium for recording a wiring shortening program .

[0002]

2. Description of the Related Art In a layout design of a semiconductor integrated circuit (hereinafter referred to as an IC), a process of reducing an area of an IC layout by moving an object to reduce a space between the objects is called a compaction process. Tools that perform compaction processing are called compactors. In addition,
An object is a set of layout parts that are simultaneously moved in the same direction by the same distance in the compaction process (or the relative positional relationship is locally fixed).

When designing a large-scale IC layout, the designer can design only the outline and optimize it with a compactor, thereby reducing the burden on the designer. If advances in manufacturing techniques allow design rules to be modified to reduce the initial object spacing, a compactor can be used to automatically reduce the layout area. In addition, the compactor is used in various situations.

The wiring length shortening problem described later is an important problem that appears in this compaction process.

From the viewpoint of reducing manpower, the compactor is important in optimizing the large-scale layout, but it is difficult to optimize the two-dimensional wiring on the large-scale layout at high speed and strictly. However, in general, even if optimization cannot be performed strictly, it is sufficient if a compact layout that can be practically used can be created at high speed, and thus compaction processing is often performed sequentially in each direction (one-dimensional compaction).

FIG. 28 shows the structure of a compactor for performing one-dimensional compaction. Hereinafter, for convenience of explanation, the left-right direction of the IC layout is referred to as the x direction, and the up-down direction is referred to as the y direction.

First, layout data is input from the input unit 202.

Next, the processing direction determining unit 204 determines the processing target as either the x direction or the y direction. For example, the first and third compactions are in the x direction, and the second and fourth compactions are in the y direction.

Next, the constraint graph creating unit 206 creates a constraint graph expressing a positional constraint relationship with respect to the processing direction of the object. As will be described later in detail, the constraint graph refers to an object as a vertex (node) of the constraint graph
Is stored for each branch connecting the two vertices. After that, the processing is performed with reference to this constraint graph.

Next, the area reduction unit 208 performs processing for packing the objects on one side in the processing direction as much as possible while referring to the constraint graph. For example, in processing in the x direction, objects are packed to the left as much as possible. In the following, the area reduction unit 208 is assumed to pack the objects in a direction in which the coordinates of the vertices are made smaller.

Next, the wiring shortening section 210 moves the element so as to shorten the wiring while referring to the constraint graph. Note that the area reduction unit 208 has already packed each object in a direction in which the coordinates of the vertices are made as small as possible, and the element that makes the coordinates of the vertices smaller cannot be moved. It is only necessary to consider the processing to increase the size.

Next, the end determination unit 212 determines whether or not the end condition is satisfied. If the end condition is satisfied, the operation is ended and the result is output from the output unit 214. If the end condition is not satisfied, the process direction determination unit 204 is returned to, the processing direction is changed, and the same compaction process is repeated. Be seen.

Here, the processing of the area reduction unit 208 and the processing of the wiring reduction unit 210 will be described. As described above, the area reduction unit 208 performs processing for packing the object on one side in the processing direction. As a result, at the stage where the processing of the area reduction unit 208 is completed, there is a portion where the wiring extends. Figure 2
A situation in which the wiring extends due to the processing in the area reduction unit 208 will be described with reference to FIG.

4 of A to D as shown in FIG.
FIG. 29B shows an example in which the object C is moved to the left as a result of the processing of moving the object to the left by the area reduction unit 208 in order to reduce the layout composed of two objects in the x direction. By the way, the object C is slightly on the right side of FIG. 29 (b) (c).
Even in such cases, the area of the entire layout does not change. In other words, it is understood that the object C does not necessarily have to be moved to the left end to minimize the area. Moreover, when the layouts of FIGS. 29B and 29C are compared, it can be seen that the wiring of FIG. 29B is inferior in the following two points. (1) In FIG. 29B, the wiring is too long, so that the resistance becomes larger than that in the given layout. (2) In FIG. 29B, when the compaction is executed next in the y direction, the object C becomes a hindrance and the object A cannot be packed downward. Therefore, the area of the layout finally output becomes large.

Generally, when an object is moved to one side as much as possible to reduce the area, there is a range in which the area does not change even if the object is moved. As can be inferred from the comparison between FIGS. 29 (b) and 29 (c), shortening the wiring length often leads to a reduction in the entire area (for example, the document “YE Cho: A Subjec”).
live Review of Compactio
n, Proc. 22nd Design Autom
ation Conference, 1985, pp.
396-404 ″). Therefore, in the processing of the wiring shortening unit 210, shortening the wiring length within this allowable range is a problem to be solved.

Such a problem is called a wiring length shortening problem and is an important problem in one-dimensional compaction.

More generally, the wiring shortening problem can be defined as "a problem that minimizes cost under constraint conditions". In order to realize more effective one-dimensional compaction, it is important to solve this problem efficiently. Hereinafter, description will be made in relation to this wiring shortening problem.

First, the "cost" will be described. Hereinafter, the process in the x direction will be described, but the same applies to the process in the y direction.

When shortening the total wiring length, it is not possible to sufficiently shorten the entire wiring only by paying attention to each wiring and making each wiring short. Further, when the priorities at the time of shortening are different for each wiring, it is impossible to consider these priorities. Therefore, it is conceivable to define the cost based on all the wirings and move the individual elements so as to reduce the cost.

Here, the straight line and each line segment of the broken line are referred to as a line. The wiring length in the x direction can be obtained by taking the sum of the lines in the x direction.

The wirings have different priorities when shortened. For example, there is a demand that wirings belonging to a mask layer, which generally has a large resistance, be preferentially shortened. Therefore, when calculating the cost, the cost is calculated by multiplying the line length by a certain priority. The cost can be expressed as in Expression (1), where x _{i is} the x-coordinate of each bending point of the wiring and c _ij is the priority. It is assumed that a line having a higher priority value is a line which should be preferentially shortened. Here, J is a set of combinations of subscripts corresponding to both ends of the line.

[0022]

[Equation 1]

If the absolute value of equation (1) is removed, equation (2) is obtained.

[0024]

[Equation 2]

The expressions (1) and (2) are equivalent to each other within the range in which the magnitude relationship of the x-coordinates at both ends of the line is not reversed.

In equation (2), if the coefficient w _i is positive, the coordinate x _i
Can be reduced, and the coefficient w _i is negative, the coordinate x _i can be increased to reduce the cost. Since the equations (1) and (2) must be equivalent, this operation must be performed within a range where the left and right sides of the line are not interchanged. Although it is necessary to replace the left and right of the lines in order to minimize the expression (1), the expression (2) must be corrected when the left and right of one line are replaced.

By rewriting the line having the length of 0 and rewriting the cost of the equation (2) at an appropriate timing, the cost of the equation (1) can be minimized.

Next, the above-mentioned "constraint condition" will be described.

In the following, x _i represents the x coordinate of the side of the polygon indicating the element including the bending point of the wiring unless otherwise specified. The design rule is a condition determined by the manufacturing process and is a constraint condition regarding the positional relationship between elements. This condition can be generally expressed as follows. x _i −x _j ≧ d _ij (3) This x _{i is associated} with the vertex of the constraint graph, and the equation (3) is expressed in the constraint graph as shown in FIG.

In the constraint graph, a vertex v has coordinates X (v),
A value D (e) of the constraint condition is stored in the branch e, and FIG.
Represents the constraint condition X (v _q ) −X (v _p ) ≧ D (e _pq ) (4).

In general, there is a relationship of x _i = X (v _p ) + c _i between the end point xi of the line and the coordinate X (v _p ) of the corresponding vertex, with c _i being a constant. Therefore,
The cost can be expressed as in Expression (5) except for the constant term.

[0032]

[Equation 3]

FIG. 31 is an example of a general constraint graph. Since the sink corresponding to the rightmost point and the source corresponding to the leftmost point of the layout are added, the constraint graph is a connected graph.

The wiring length reduction problem is a problem that minimizes the cost of the equation (1) within a range satisfying the constraint condition represented by the constraint graph.

The local improvement method and the group improvement method, which are conventional wiring shortening methods, will now be described.

To reduce the cost, if the weight W (v _p ) of the vertex v _p is positive, the position of the vertex may be reduced, and if the weight of the vertex is negative, the position of the vertex may be increased. good. By the way, since the area reduction processing is performed prior to the wiring shortening processing to make the coordinates of each vertex as small as possible, the wiring shortening unit cannot further reduce the coordinates of the vertices. Therefore, as described above, only the process of increasing the coordinates of the vertices need be considered in the wiring shortening unit.

If the weight of each vertex is negative, the cost can be considerably reduced by increasing the coordinates of the vertex within the range where the constraint condition is not violated. Such a method is called a local improvement method.

However, it is known that the local improvement method alone causes convergence at a distance from the optimum solution.
FIG. 32 shows an example in which the wiring cannot be sufficiently shortened by the local improvement method. The black circles attached to the vicinity of the wiring in FIG. 32 indicate that the objects on both sides of the line cannot approach any more.

However, in the example of FIG. 32, by moving a plurality of objects at the same time, it is possible to reach an arrangement with a smaller cost. For example, the objects (A to
If you consider a group of H) as a whole, this group is connected to the wiring of priority 4 from the right,
From the left, the wiring of priority 1 and the wiring of priority 2 are connected. Therefore, if all the objects are moved together to the right, the wirings of priority 1 and the wirings of priority 2 are extended, but the wirings of priority 4 are contracted, so that the cost is reduced. Similarly, three shaded objects (B,
The cost can be reduced by moving D and E) together.

As described above, a method in which a plurality of vertices are regarded as one group and the vertices in the group are simultaneously moved is called a group improvement method. It can be seen by moving the whole group to the right side and seeing if the cost is reduced by summing the weights of all the vertices included in the group. As in the case of the local improvement method, it can be seen that if this value is negative, the cost is reduced by moving it to the right.

In the group improvement method, the process can be executed efficiently by storing the vertices to be moved simultaneously in the group in a tree. In the tree created here, there is a tight constraint graph branch from the parent to the child. During processing, the group of vertices that should be moved simultaneously changes every moment, so the tree members must be modified each time the group is moved.

In FIG. 32, the cost can be reduced by moving the three shaded objects (B, D, E) to the right rather than moving all the objects to the right. , E) is the optimal operation in this state. The problem is how to efficiently obtain a tree that has these three objects (B, D, E) as members.

A conventional method (for example, the document “W. L. Sch.
iele, Improved Compaction
by Minimized Length of W
ires, Proc. 20th Design A
automation Conf. , 1983, pp. 1
21-127 ", reference" G. Lakhani, R .; Va
radarajan: A Wire-Length M
inimation Algorithm for
Circuit Layout Compaction
n. IEEE International Sympo
sium onCircuits and System
ms, Vol. 1, pp. 276-279 ", reference" C. Kingsley: A Hiererachic
al, Error-Tolerant Compact
or, IEEE 21st Design Automa
section Conf. , 1984, pp. 126-13
2 "), the sum of the weights of the vertices on the path from the root of the tree to each vertex is calculated as an accumulated value. If the accumulated value of a leaf is positive, the vertex on this path is fixed, The other vertices were separated and moved.
Will be described with reference to.

In FIG. 33A, the numerical value in the circle representing the apex is the weight of the apex, and the numerical value in parentheses beside the apex is the accumulated value. According to the conventional method, it was found that the vertices 301 to 303 on the bottom path should be fixed, and
Separate as in (b). Next, among the separated subtrees, a tree whose sum of weights is negative is searched for. As a result,
It is understood that the two vertices surrounded by the dotted line in (b) should be moved further to the right.

However, even if this method is adopted, in the case of the layout of FIG. 32, the three shaded objects (B, D, E) cannot be moved properly.

FIG. 34 shows the accumulated value calculated in the group corresponding to FIG. As can be seen from FIG. 34, since the accumulated value of the leaves of all trees is negative, it is not possible to decompose this tree to separate objects B, D and E.

In addition to the method described above, a method for obtaining a good solution by executing a linear programming method on a graph (reference “J. Lee, CK Wong: AP”).
erformance-Aimed Cell Com
actor with Automatic Jog
s, IEEE Trans. Computer-Ai
ded Design, Vol. 11, No. 12, 1
992, pp. 1495-1507, 1992 ", document" T. Yoshimura: A Graph Th
eoretical compaction Algo
ritm, IEEE Proc. ISCA 85,
1985, pp. 1455-1458 ") has been proposed, but these methods have a problem that the processing speed becomes slow.

[0048]

In the conventional one-dimensional compaction process, when the group improvement method is used in the wiring shortening process performed after the area reduction process, the value of the wiring cost is reduced. There is a problem that sufficient compaction cannot be performed because the object cannot be moved even though the object can be moved. In addition, there is a problem that the processing takes time in the case of using the linear programming in the wiring shortening processing.

The present invention has been made in view of the above circumstances, and is a wiring shortening method for shortening the wiring that is performed following the area reduction processing in the one-dimensional compaction processing for the layout of the semiconductor integrated circuit. Wiring shortening method and wiring shortening processing device capable of moving an object to reduce wiring cost effectively and effectively
And a storage medium for recording the wiring shortening program .

[0050]

According to the present invention, a specific object in a wiring pattern formed by wiring between objects is moved according to a load showing a performance of wiring shortening in consideration of shortening priority set for each wiring. As a result, in the wiring shortening method for shortening the length of the wiring, a first step of forming a tree by using a plurality of objects having a predetermined positional relationship as nodes, and A second step of sequentially calculating the load of the parent node with respect to the child node based on the load of the child node from the leaf side, and moving the object corresponding to the tree part whose root node is the node having the load satisfying a predetermined condition Third to identify as a target
And the predetermined positional relationship is a relationship in which the two objects are located at a limit distance that allows the two objects to approach each other, and the object located upstream of the two objects in the object moving direction in the tree. Is a parent node, and the object located downstream is a child node, and the load of a node that does not have a child node, including leaf nodes, is expanded by moving the node in the wiring part that has one end Although the wiring is shortened by moving the node among the wiring line segments having one end at the node from the sum of the shortening priorities, the weight is obtained by subtracting the sum of the shortening priorities, and the load of the node having the child node is set. , The weight value of the node plus all positive loads of the child nodes of the node, Matter, wherein said load is a negative value. In addition, the present invention moves a specific object in a wiring pattern formed by wiring between objects according to a load indicating a wiring shortening performance in consideration of a shortening priority set for each wiring, thereby lengthening the wiring length. A first step of forming a tree with a plurality of objects having a predetermined positional relationship as nodes, respectively,
For each tree, from the leaf side of the tree,
A second step of calculating the load of the parent node with respect to the child node based on the load of the child node, and a second step of specifying an object corresponding to a tree part whose root node is a node having a load satisfying a predetermined condition as a movement target. 3), and the shortening priority of each wiring line segment is a function of the length of the wiring line segment, the width of the wiring line segment, and the mask layer to which the wiring line segment belongs. .

[0051]

Preferably, the procedure is executed after the movement of the object for reducing the layout area is executed, and the movement of the object for reducing the layout area is performed in the one-dimensional compaction direction. The object is moved so as to be arranged on one side as close as possible, and in the procedure, objects to be moved may be sequentially arranged from the object located on the other side in the one-dimensional compaction direction.

Preferably, the condition that all the objects cannot be moved even after executing the above procedure is repeated twice, the condition that all the objects cannot move even after performing the above procedure, and the number of repetitions is One of the conditions in which the specified number of times is exceeded, and the change in cost corresponding to the weighted sum of the lengths of the wire segments is smaller than the specified value.
The above procedure may be repeatedly executed until an end condition consisting of one condition or a condition defined by a predetermined combination is satisfied.

Preferably, the relationship between the respective objects may be represented by a constraint graph, and the objects may be moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed.

Preferably, the object is moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed, and the length of the wiring line segment becomes 0 when the movement of the object is completed. If so, the positional relationship of both ends of the wiring line segment is reversed, the structure of the tree is corrected, the load is obtained again, and then the tree part having a load satisfying a predetermined condition as the root node is added to the tree part. All corresponding objects may be moved as a unit.

Preferably, the object may be moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed, and the length of the wiring line segment becomes 0 when the end condition is satisfied. For example, after reversing the positional relationship of both ends of the wiring line segment, until the end condition is satisfied again,
You may make it repeat the said procedure.

Preferably, the object is moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed, and the length of the wiring line segment becomes 0 at the time when the end condition is satisfied or If there is a tight branch between the vertices corresponding to both ends of the line segment, both ends of the wiring line segment are merged, and the above procedure is repeated until the end condition is satisfied again. You may do it.

It is preferable that the portion where the layout is changed is specified and the difference between before and after the change is displayed or the change state of at least one of the total wiring length and the cost is displayed. The state may be presented and the number of repetitions input by the user may be used as the ending condition.

[0059]

Preferably, a specific design rule and a shortening priority for wiring line classification may be set only in a specific portion of the layout.

The inventions relating to the above method also hold as inventions relating to the apparatus. Further, the invention according to the above method is also realized as a machine-readable medium in which a program for causing a computer to execute a corresponding procedure or means is recorded. That is, wire between objects
A specific object in the wiring pattern
A wiring shortening performance that considers the priority set for each line.
The load is indicated by
Predetermined position on the wire shortening device that shortens the wire length
Multiple related objects as nodes
Means for forming a tree and the tree for each tree
Based on the load of the child node, in order from the leaf side of Lee
A means for calculating the load of the parent node on the node, and a predetermined
The node with the load that satisfies the condition of is set as the root node
The object corresponding to the tree part is specified as the movement target.
And means for determining. Also, of
Of the wiring pattern formed by wiring between
Considering the priority set for each object,
Move according to the load showing the performance of shortened wiring
Wiring shortening program that shortens the length of the wiring by
A computer-readable storage medium that stores RAM
In addition, the program is a plurality of programs having a predetermined positional relationship.
Form a tree with each object as a node
For each program tree and each tree,
Based on the load of the child node, from the leaf side of
Program code that calculates the load of the parent node on the node
And a node that has a load that satisfies a predetermined condition.
Move the object corresponding to the tree part
With the program code specified as an elephant
Characterize.

According to the present invention, a tree is formed with each of a plurality of objects having a predetermined positional relationship as nodes, and the load of each node is reflected from the downstream side of the tree to the upstream side of the tree. Since the groups are sequentially obtained while being performed, the group of objects to be moved can be easily obtained based on the load, and the movement of the objects can be executed efficiently and effectively so as to reduce the wiring cost.

[0063]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

The present invention shortens the wiring length by the above-mentioned group improvement method, and improves the tree construction method for specifying the group.

FIG. 1 shows a structural example of a compaction processing apparatus according to an embodiment of the present invention. FIG. 2 shows an example of the processing procedure of the compaction processing apparatus.

FIG. 3 shows an example of the internal structure of the wiring shortening unit of this embodiment. FIG. 4 shows an example of the processing procedure of the main wiring shortening unit. Further, FIG. 5 shows an example of a procedure of weight determination processing and element movement processing of the wiring shortening unit of the present embodiment.

As shown in FIG. 1, the compaction processing apparatus of this embodiment has an input unit 2, a processing direction determination unit 4, a constraint graph creation unit 6, an area reduction unit 8, a wiring reduction unit 10, an end determination unit 12, The output unit 14 is provided. The compaction processing device can be composed of a computer and a program (or a computer and an operating system and a program).

For convenience of explanation, the left and right directions of the IC layout will be referred to as the x direction and the up and down directions will be referred to as the y direction.
Further, in the case of using a specific example, it is assumed that the object is first moved to the left side and then moved to the right side to reduce the cost.

The input unit 2 inputs the layout data of the IC to be processed (step S11).

The layout data is composed of, for example, the coordinates of each pattern for each mask layer (for example, the coordinate values of two points on the diagonal of a polygon indicating an element).

The input section 2 also inputs design rule data. A design rule database is provided, and design rule identification information is input from the outside through the input unit 2, and the design rule database is searched for using the identification information as a key to obtain the data. May be.

The priority of each wiring is determined according to the attribute (eg mask layer) of the wiring. The wiring priority data (for example, the data indicating the correspondence between the wiring attributes and the priorities) is input from the input unit 2 or provided with a database, and the identification information is input from the outside through the input unit 2. It is assumed that the data is retrieved from the database by using the identification information as a key.

The priority may be set for each line. For example, the line priority may be a function of line length, line width, and the mask layer in which the line is created. This function is, for example, a function that increases as the value determined according to the mask layer increases, increases as the line length increases, and decreases as the line width increases. Is. Also, this function may be selectable.

Further, it is possible to set a specific design rule and a priority for each line only in a specific portion of the layout.

The processing direction determination unit 4 determines the processing target to be either the x direction or the y direction (step S12). For example, the first compaction is determined as the x direction and the second compaction is determined as the y direction. For example, when two sets of compaction are performed with this set as one set, the processing direction is determined in the order of x direction, y direction, x direction, y direction.

The constraint graph creating section 6 refers to the layout data and the design rule data to create a constraint graph for the processing direction (step S13).

For example, a constraint graph as shown in FIG. 14 is created for a layout including eight objects A to H as shown in FIG. In FIG. 14, the identification information A to H of the corresponding object and the coordinates of the vertices are shown in the circles representing the vertices. In addition, the minimum required distance between the vertices that connect the vertices (called the approach limit distance)
Is written.

In FIG. 10, the value of the priority is shown for each wiring. Further, the approach margin distance (called slack), which is obtained by subtracting the approach limit distance from the difference between the coordinates of both vertices in FIG. 14, and which indicates how much margin both objects corresponding to both vertices can approach, is 0 or more. 10 are indicated by parenthesized numbers with respect to the wiring of FIG. 10, and those with zero slack are indicated by black circles with respect to the wiring of FIG. In the constraint graph, a branch with 0 slack is called a tight branch.

In the constraint graph, the branches give a positional constraint relationship between objects and do not necessarily correspond to wiring (objects corresponding to two vertices at both ends of a branch are not connected by wiring). Sometimes).

Here, the relationship between each element pattern, the object, and the vertex (node) of the constraint graph will be described.

FIG. 6 shows an IC for explaining the above relation.
An example of the layout pattern (partially extracted) is shown. As described above, the processing direction is the x direction. In FIG. 6, 30 is a wiring pattern (which is a processing target), 31 and 33 are diffusion layer patterns, and 36 and 3
8 is another wiring pattern, 32, 37, 34 and 39
Is a pattern of contact holes, and 35 is a pattern of another wiring having a bond with the diffusion layer 31. In the case of FIG. 6, the pattern of 31, 32, 36, 37, the left end of the wiring 30 and the right end of the wiring 35 are a group of patterns or elements that move together, and this is one object (J. ) Is formed. Similarly, 33, 34, 3
8, 39 and the right end of the wiring 30 are a group of patterns or elements that move together, which forms one object (denoted as I).

The constraint graph has a connection from the vertex corresponding to the object J (denoted as vertex j) to the vertex corresponding to the object I (denoted as vertex i). Further, for example, the x-coordinate of the lower left corner of the pattern 31 a point of the object J and X _j, when the x-coordinate of the lower left corner of the reference point of the object I patterns 33 and X _i, as the coordinates of the vertices j X _j is stored, and X _i is stored as the coordinate of the vertex i. Also, the value of the constraint condition (approach limit distance) given between both objects, that is, between both vertices, is
Of the design rules corresponding to both objects, the one that minimizes the approach limit distance will be defined. The approach limit distance is stored in correspondence with the branch connecting the corresponding two vertices.

The movement in the area reduction processing and the movement in the wiring shortening processing are performed by moving the coordinates of the vertices of the constraint graph, that is, the coordinates of the reference point of the object I in the processing direction. Here, in one object, the relative positional relationship between the reference point and each pattern is always fixed. Therefore, for example, as shown in FIG. 7, for each pattern, the identification information of the vertex corresponding to the object to which it belongs and the coordinates of the vertex (reference point of the object)
When the coordinates of the vertices are moved, the coordinates of each pattern can be obtained from the table and constraint graph shown in FIG.

For example, x at the left end of the wiring pattern 30.
Coordinates and x _t, the coordinates relative to the coordinates X _j of the corresponding vertices j and d _2, x _t is represented by _{x t = X j + d 2} . Similarly, the x coordinate of the right end of the wiring pattern 30 is set to x
and _s, the relative coordinates from the coordinates X _i of the corresponding vertex i and d _1, x _s is expressed by _{x s = X i + d 1} .

When the length of the wiring pattern 30 in the x direction is L, L is L = x _s −x _t = (X _i + d ₁ ) − (X _j + d ₂ ) = (X _i −X _j ) + (D ₁ −d ₂ ) = L _ij + constant such that the offset difference ( _ij) between the reference points of the object I to which the right end of the wiring pattern 30 belongs and the object J to which the left end of the wiring pattern 30 belongs. d ₁ −d ₂ ) is added. That is, the length L of the wiring pattern 30 in the x direction is uniquely determined by the distance L _ij between the reference points of the object I and the object J. Therefore, as described above, the cost can be expressed as in Expression (5) except for the constant term. In order to use the formula (1) when evaluating the cost, it is necessary to obtain the length of each wiring, but the formula (5) has an advantage that it can be calculated only by the coordinates of the vertices.

The movement in the area reduction processing, the movement in the wiring shortening processing, the cost calculation / evaluation, etc. can be processed by handling the reference points of the object, that is, the vertex coordinates of the constraint graph and the constraint conditions between the vertex coordinates. The coordinates after the movement of each pattern may be obtained last.

It should be noted that, like the wiring having the bent portion shown in FIG. 8, the wiring line segment 40 in the y direction in the x direction processing and the wiring line segment 43 in the x direction in the y direction processing are the wiring line segments themselves. May form an object. In FIG. 8, considering the y-direction processing, the wiring line segments 42, 43, and 44 in the x direction form objects, respectively. In such a case, when the wiring line segment 43 is moved upward. , The y-coordinate of the wiring line segment 43 does not exceed the y-coordinates of the wiring line segments 42 and 44.
To move upward (that is, approach limit distance = 0
Will be). The y coordinate of the wiring line segment 43 is the wiring line segments 42 and 44.
In order to enable movement beyond the y coordinate of, the line inversion processing described later is performed. In order to move the wiring line segments 42, 43, 44 together upward, a merge process described below is performed.

Next, the area reduction unit 8 performs processing for packing the objects as much as possible to one side in the processing direction while referring to the constraint graph (step S14). For example, in processing in the x direction, objects are packed to the left as much as possible. In the following, the area reduction unit 8 packs the objects in the direction in which the coordinates of the vertices are made smaller (left in the processing in the x direction and lower in the processing in the y direction).

Actually, the coordinate of the vertex is made as small as possible by referring to the constraint graph, and the coordinate value of the moved vertex in the constraint graph is updated.

Next, the wiring shortening unit 10 creates a tree (to be described later) and shortens the wiring while referring to and updating the constraint graph (more specifically, to reduce the cost as much as possible). Is moved (step S15). Note that the area reduction unit 8 has already packed each object in the direction in which the coordinates of the vertices are made as small as possible, and the objects that make the coordinates of the vertices smaller cannot be moved.
Only the process of increasing the coordinates of the vertices need be considered.

By the way, in the present embodiment, instead of the "accumulated value" in the conventional group improvement method, the "load" of the vertices is obtained, and the optimum vertex group is obtained based on this load. Therefore, first, the “load” used in this embodiment will be described with reference to FIG. 9.

FIG. 9 shows an example of a tree created to obtain a group of vertices. FIG. 9 corresponds to the layout of FIG. In Figure 9, in the circle that represents the vertex,
The identification information of the corresponding object and the weight of the vertex are entered. Here, the weight is the weight W (v _p ) of the equation (5). For example, since the object A includes the right end portion of the wiring having the priority 1 and the left end portions of the two wirings having the priority 1 and the priority 5, the weight of the object A is +1.
-1-5 = -5.

To obtain the load R (v) of the vertex v, the vertex (leaf node) which is the leaf of the tree is calculated according to the equation (6).
From then on, you can proceed to the branch in order. Where S (v) is the set of children of vertex v in the tree. Further, θ (x) is a function that becomes 0 when x <0 and becomes 1 when x ≧ 0 (or 0 when x ≦ 0 and 1 when x> 0).
Function).

[0094]

[Equation 4]

That is, the load of the leaf node becomes equal to the weight of the leaf node. Then, the load of the parent vertex (parent node) is a value obtained by adding the weight of the parent node and the positive load of the child vertex (child node).

This value corresponds to the load when moving each vertex to the right. For example, in the case of the vertex vA, the weight of the vertex vA is -5, but the descendants on the right of the vertex vA prevent the movement to the right. To see how disturbed it is, look at the load value at the top of the child. Specifically, since the load of the vertex vC is +7, this is the vertex vA.
Prevent you from going to the right. On the other hand, the load on the vertex vB is −
Since it is 3, it does not prevent the vertex vA from going to the right.
That is, θ (R (vB)) = θ (−3) = 0, θ (R
Since (vC)) = θ (+7) = 1, the load R (vA) at the vertex vA is R (vA) = W (vA) + R (vC)
= (-5) + (+ 7) = + 2.

In the case of FIG. 9, all the vertices (that is, the vertices v whose roots are the vertices vB whose load is negative)
Increase the coordinates of (B, vD, vE) (move to the right)
be able to.

As described with reference to FIG. 33, the layout of FIG. 32 cannot be moved any more by the conventional group improvement method, but according to this embodiment,
The cost can be further reduced by moving the objects B, D and E in the layout of FIG.

Next, the structure and operation of the wiring shortening section 10 will be described.

As shown in FIG. 3, the main wiring shortening unit 10
Includes a weight determining unit 100, an element moving unit 102, an end determining unit 104, and a line inverting unit 106.

The weight determining section 100 obtains the weight of the vertex corresponding to each object (step S21). The element moving unit 102 moves an object, that is, selects a vertex to be moved, determines a movement amount, updates the coordinates of the vertices of a constraint graph, and a tree within a range in which the magnitude relation of the coordinates of both ends of the line in the processing direction is not reversed. The structure is updated and the load is updated (step S22). The termination determination unit 104 determines whether to terminate the wiring shortening process, perform the line inversion process and perform the process again, or perform the process again without performing the line inversion process (steps S23 and S24). The line inversion unit 106 corrects the constraint graph, the tree, and the cost function so that the connection relationship between the two corresponding vertices is reversed with respect to the line whose length becomes 0 in the case of FIG. Inversion processing is performed (step S25).

The weight determining process and the element moving process in steps S21 and S22 of FIG. 4 will be described in detail below with reference to FIG. Note that step S31 in FIG. 5 corresponds to step S21, and the other steps in FIG. 5 correspond to step S22.

10 to 13 show how objects are moved in the wiring shortening process. 10 to 13, the priority is shown for each wiring, and in the case where the spacing between objects cannot be reduced further, a black circle is added to the wiring between the objects, and if there is a margin for the reduction, The value (slack) is shown in parentheses. For example, in the objects A and B in FIG. 10, the interval can be reduced by 3. Further, the space between B and D cannot be further reduced.

FIG. 14 shows a constraint graph created from layout data and design rules in the constraint graph creation processing section 6. In FIG. 14, the identification information of the corresponding object and the coordinates of the vertices are shown in the circle indicating the vertices. The values of the coordinates of the vertices are as shown in FIGS. 15 to 17 as the object moves. It will be updated sequentially. That is, the coordinates of the vertices in FIGS. 14 to 17 correspond to the arrangements in FIGS. 10 to 13, respectively. In addition, the value given to the branch represents the necessary interval (approach limit distance) between the two vertices corresponding to the branch.

The slack between corresponding objects can be obtained from the coordinates of the two vertices of the constraint graph and the approach limit distance of its branches. For example, in FIG. 14, vertex B
The position of is 26 and the position of the vertex A is 20, and the difference is 6, while it can be seen that the required distance between A and B is 3, so the margin between AB is 3. I understand.
This corresponds to the fact that there is a margin of 3 between the objects AB in FIG. In actual processing, this constraint graph is used to calculate how much slack (slack) exists between individual objects.

18 to 21 show how a tree changes during processing.

The condition for moving in groups is that the load on the root of the tree is negative, and the coordinates of all the vertices included in the tree having a root with a negative load are increased. Therefore, if YES in step S35, the process of increasing the coordinates of the vertices of the tree is performed, and if NO, the next tree is selected.

As described above, the main wiring shortening process is, for example, a process of moving the elements, which are arranged close to the left side, to the right side. In this case, when the element on the right side is moved, a gap is created between the element on the left side and the element on the left side, and the moving range of the element on the left side is expanded. Therefore, by sequentially moving from the element on the right side, the number of repetitions until the processing is completed can be reduced. Therefore, in step S33, it is advisable to select trees in descending order of coordinates.

When actually moving a tree, it is necessary to determine the amount of movement. If you move the tree to the right,
When the equal sign of the constraint between a certain vertex included in the tree and another vertex not included in the tree is satisfied and further rightward movement is reached, the limit at which this constraint is violated is reached. The vertices included in the tree at this time are called critical nodes, and the vertices not included in the tree are called limit nodes. The amount of movement of the tree at this time is slack. In step S37, the tree is moved by the size of the slack. Further, since the limit node is added to the tree in step S38, the limit node also moves together with other vertices as a part of the tree thereafter.

Here, it is assumed that the objects are arranged in the state of FIG. 10 (the tree is in the state of FIG. 18). Immediately after the start of the process, the process is started with load = weight at each vertex in step S31, but the tree is updated every time the subsequent step S38 is executed.

If the root A is selected in step S33, referring to FIG. 18 in step S35, the load on the root is negative, so the flow advances to step S36.

When considering moving A to the right as shown in FIG. 10, the smallest margin is B, so in step S36, the limit node is vertex B, the critical node is vertex A, It turns out that the slack is three. These can be calculated by referring to the constraint graph of FIG. Since the slack can be obtained by comparing the difference between the value of the branch coming out of the vertex A (approach limit distance) and the coordinates of the end point and the start point of the branch, all branches coming out of the vertex A Slack is calculated, and the minimum slack, and the limit node and critical node corresponding thereto are calculated.

Then, the process proceeds to step S37. Here, since the slack is 3, the coordinates of the vertex A are changed from 20 to 23 as shown in FIG. 15, and the branch between AB becomes a tight branch.

As a result, the arrangement changes as shown in FIG.
A black circle is written between B. Further, the tree is also changed in step S38, as shown in FIG.

After that, the process proceeds to step S35, but since the load on the root A of the tree is negative here, the process proceeds to step S36 again.

Hereinafter, steps S36 to S38
Up to the above, the state changes to those shown in FIGS. 12, 16 and 20, but since the root load is positive this time, the process proceeds to step S33 in step S35. Here, the processing for the tree having the vertex A as a root is completed, and the processing for the next tree is performed.

Next, it is assumed that the vertex C is selected in step S33. Referring to FIG. 20, since the load on the vertex C is positive, the process returns from step S35 to step S33.
In this way, vertices with a positive load will be skipped.

Next, it is assumed that the vertex B is selected in step S33. Referring to FIG. 20, since the load on the vertex C is -5, which is negative, the process proceeds to step S36. Since the tree to be processed at this time is a tree whose root is B, it is assumed that the tree at this point is as shown in FIG. Further, by repeating the same processing as the above, the objects B, D, E
Moves to the right, as shown in FIGS. 13 and 17.

Next, FIG. 22 shows another example of the procedure of the weight determining process and the element moving process. The procedure of FIG. 22 is a slight modification of the procedure of FIG. Here, the difference will be described.

Since the condition for moving the objects in groups was that the load on the root of the tree was negative,
It is not necessary to construct a tree rooted at a vertex whose load is always positive. As can be seen from the equation (6), if the weight of the root of the tree is positive, the load on the root of the tree will not be negative.
Therefore, in step S42, only vertices having a negative weight are selected.

Further, prior to the processing of FIG. 22, a tight branch is traced backward from the sink of the constraint graph, and the apex of the traceable range is marked with BLOCKED. It is desirable to execute this processing immediately after the processing of the area reduction unit 8 is completed. Here, the tight branch is a branch in which the size of the constraint condition indicated by the branch is equal to the distance between the vertices at both ends of the branch. At this time, when the vertex with the BLOCKED mark is moved, the coordinates of the sink of the graph increase, and the layout area increases.

Therefore, in step S47, it is checked whether or not the BLOCKED mark is added to the limit node. If the mark is added, in step S50 BLO is added to all vertices on the path from the critical node to the root of the tree.
The mark of CKED is added, and if not added, the moving process is performed in steps S48 and S49. This prevents the vertices marked BLOCKED from moving if the layout area should not be increased.

When the element moving process is completed, the termination determining unit 104 of the wiring shortening unit 10 performs the line inversion process and re-executes the process cycle, or re-executes the process cycle without performing the line inversion process. Or whether to end the process.

The following conditions can be considered as the termination determination conditions in the termination determination unit 104. (1) The positions of all objects were not corrected. (2) The positions of all objects were not corrected twice. (3) The fixed number of times t1 specified by the user is exceeded. (4) The change in cost (for example, (previous cost-current cost) or (previous cost-current cost) / current cost) becomes smaller than the value s1 designated by the user. (5) A combination of some of the above condition 1 or condition 2, condition 3 and condition 4 by logical sum or the like.

The convergence state may be presented so that the user can determine the number of repetitions and input or update it. Alternatively, the converged state may be presented, and when the user gives an instruction to end, the processing may be ended regardless of whether the above condition is satisfied.

As the presentation of the convergence state, for example, a method of identifying the portion where the layout is changed and displaying the difference before and after the change, the change state of the cost or the change of the total wiring length and the cost You may make it display a state.

Note that the presentation of the convergence state may be switchable between a mode performed each time the vertex is moved and a mode performed when the end is determined, according to a user's instruction.

Next, if the above end condition is not satisfied, the processing cycle will be executed again. At that time, the following may be considered to determine whether or not to perform the line inversion processing. To be (A) The positions of all objects were not corrected (except when the above (1) is adopted). (B) The fixed number of times t2 specified by the user (t2 <t1 when the above (3) is adopted) is exceeded. (C) A combination of the above condition a and condition b by logical sum.

In addition to the above, when the above end condition is satisfied, the line process is performed, and then the weight determination process and the element movement process are executed only once again until the end condition is satisfied. When the processing cycle is re-executed as described above, various methods such as a method of always executing the line inversion processing and a method of executing the line inversion processing each time the object corresponding to one tree in FIG. 24 is moved are considered. To be

Now, the line inversion unit 106 performs the line inversion process of switching the right and left of the line when the line length becomes 0 as described above. Specifically, the connection relation of the constraint graph and the approach limit distance, the tree structure, the weight and load of the vertex of the tree are modified so that the relation between the start point and the end point of the two vertices corresponding to both ends of the line is reversed. . The cost function uses the weight of the corrected vertex. This element movement process is
Since the object is moved within the range in which the positional relationship of the vertices is not reversed, if there is something in which the distance between the vertices is 0 when the object cannot be moved, a constraint graph that reverses the positional relationship of the vertices, In some cases, the objects can be moved further by modifying the tree and cost function.

When the wiring shortening process is completed, a table showing the relationship between each pattern as shown in FIG. 7, the corresponding vertices, and the relative coordinates in the processing direction with respect to the coordinates of the vertices, and the coordinates of each vertex obtained from the constraint graph. The coordinate data of all the layout patterns are recalculated based on, and the IC layout data after the line shortening process is generated. In the area reduction processing and the wiring shortening processing, each time the vertex is moved, the coordinates of each pattern belonging to the object corresponding to the vertex may be obtained and the IC layout data may be updated each time.

Next, the end judging section 12 judges whether or not the end condition is satisfied (step S16). If the end condition is satisfied, the process ends. If the end condition is not satisfied, the process direction determination unit 4 is returned to, the process direction is changed, and the same compaction process is repeated. As an example of the end condition, a condition in which the specified number of repetitions have ended can be considered. For example, this is a case where the x-direction processing and the y-direction processing are alternately performed twice, and a total of four times of repetition is completed.

If the termination determining unit 12 determines that the termination condition is satisfied, the output unit 14 outputs the compaction processed IC layout data (step S).
17).

Next, a case where a movable object cannot be moved by the above basic procedure will be described with reference to the example of FIG.

Lines L1, L2 and L3 of FIG.
Respectively correspond to the vertices v1, v2 and v3 of (b) and prohibit movement involving the interchange of both ends of the line. Therefore, the value of the constraint condition between v1 and v2, v3 and v2 is There are 0 branches. At this time, due to the constraint condition, L1 and L3 cannot move to the right beyond L2. Further, when the weight of each vertex is calculated with the priority of each line set to 1, the weight described next to each vertex of (b) is calculated,
Since the weight of L2 is 2 (> 0), L2 does not move to the right by itself. On the other hand, since the sum of the weights of all vertices is negative, moving the whole to the right should shorten the wiring length, but since this graph cannot be stored in a tree, the whole is to the right. Can not be moved to (c)
The processing ends in the state shown in.

Therefore, when the length of the line is 0, or when the branches between the vertices corresponding to the left end and the right end of the line are tight branches, both ends of the line have the same vertices. I will refer to it. At this time, the weight of the vertices is made equal to the sum of the weights of the vertices that were originally referenced. Since the sum of the weights of the three vertices in (a) is -2,
As in (d), the weight of the new vertex v becomes -2.
In the case of (d), the line moves to the right side, and the wiring length can be shortened.

A wiring shortening unit 10 for executing this processing
FIG. 26 shows an example of the internal configuration of the above, and FIG. 27 shows an example of the procedure thereof. The portions other than the merge processing unit 108 are basically the same as the wiring shortening unit 10 of FIG. 3 as described above, but when the end determination unit 104 determines that the merge processing is necessary, the merge processing unit In step 108, the line and vertex merge processing, the constraint graph change, and the load correction described above are performed, and then the processing is resumed.

The determination of whether or not the merge process is necessary in the end determination unit 104 is, for example, the end determination unit 104.
A method is conceivable in which the merge process is performed instead of the line inversion process when the number of times of storage determined to perform the line inversion process is an integral multiple of the specified number. For example, assuming that the prescribed number of times is 4, if the line inversion process is determined to be performed again after the line inversion process is performed three times, the merge process is performed instead of the line inversion process.

Alternatively, when the above-mentioned end condition is satisfied, the line processing is performed, the weight determination process and the element movement process are executed again, and then the above-mentioned end condition is satisfied again, the merge process is executed. In the above, the method of executing the weight determination processing and the element movement processing again, the line inversion processing is always executed when the processing cycle is executed again as shown in FIG. 23, or the object corresponding to one tree in FIG. 24 is moved. When the line inversion process is performed every time, the method of performing the merge process instead of the line inversion process when the number of accumulations of the line inversion process is an integral multiple of the specified number of times, the process cycle is performed again as shown in FIG. When executing, line inversion processing is always executed, or line inversion processing is executed every time an object corresponding to one tree in FIG. 24 is moved. Both a method of performing the merge processing when the same conditions as the conditions for performing line inversion process described above (a, b, etc.) is satisfied, various methods are conceivable.

By the way, in the above description, the object does not include both ends of the line in the processing direction, but it is possible to specify a line to be excluded from the shortening target. In this case, the line you want to exclude from the shortening target,
It may be handled in the same manner as other non-wiring patterns.

Further, it is possible to specify a specific range of the layout which is not desired to be changed. In this case, all the lines included in the specific range may be treated in the same manner as other non-wiring patterns.

The functions described in this embodiment are as follows.
It can be realized as hardware or software. When implemented as software, it can also be implemented as a machine-readable medium that records a program for causing a computer to execute the above-described procedures or means.

The present invention is not limited to the above-described embodiments, but can be implemented with various modifications within the technical scope thereof.

[0144]

According to the present invention, a tree is formed with a plurality of objects having a predetermined positional relationship as nodes, and the load of each node is changed from the downstream side of the tree toward the upstream side of the tree. It is possible to find the group of objects that should be moved easily based on the load, and to perform the object movement that reduces the wiring cost efficiently and effectively. it can.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a configuration of a compaction processing device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an example of a processing procedure of the compaction device according to the embodiment of the present invention.

FIG. 3 is a diagram showing an example of an internal configuration of a wiring shortening unit.

FIG. 4 is a flowchart showing an example of a processing procedure of a wiring shortening unit.

FIG. 5 is a flowchart showing an example of a procedure of weight determination processing and element movement processing of a wiring shortening unit.

FIG. 6 is a diagram showing an example of a layout pattern.

FIG. 7 is a diagram showing an example of a table in which correspondences between patterns, vertices, and relative coordinates are stored.

FIG. 8 is a diagram showing another example of a layout pattern.

FIG. 9 is a diagram showing an example of a tree for explaining a load.

FIG. 10 is a diagram showing an example of an arrangement state of objects.

FIG. 11 is a diagram showing an example of an arrangement state of objects.

FIG. 12 is a diagram showing an example of an arrangement state of objects.

FIG. 13 is a diagram showing an example of an arrangement state of objects.

FIG. 14 is a diagram showing an example of a constraint graph.

FIG. 15 is a diagram showing an example of a constraint graph.

FIG. 16 is a diagram showing an example of a constraint graph.

FIG. 17 is a diagram showing an example of a constraint graph.

FIG. 18 is a diagram showing an example of a tree.

FIG. 19 is a diagram showing an example of a tree.

FIG. 20 is a diagram showing an example of a tree.

FIG. 21 is a diagram showing an example of a tree.

FIG. 22 is a flowchart showing another example of the procedure of the weight determination processing and the element movement processing of the wiring shortening unit.

FIG. 23 is a flowchart showing another example of the processing procedure of the wiring shortening unit.

FIG. 24 is a flowchart showing still another example of the processing procedure of the wiring shortening unit.

FIG. 25 is a diagram for explaining merge processing.

FIG. 26 is a diagram showing another example of the internal configuration of the wiring shortening unit.

FIG. 27 is a flowchart showing an example of a processing procedure including merge processing of a wiring shortening unit.

FIG. 28 is a diagram showing an example of a configuration of a conventional compactor.

FIG. 29 is a diagram for explaining the area reduction processing.

FIG. 30 is a diagram for explaining a constraint graph.

FIG. 31 is a diagram showing an example of a general constraint graph having a sink and a source.

FIG. 32 is a diagram showing an example of an arrangement state of objects.

FIG. 33 is a diagram showing an example of a tree for explaining accumulated values.

FIG. 34 is a diagram showing an example of another tree for explaining accumulated values.

[Explanation of symbols]

2 ... Device input section 4 ... Processing direction determination unit 6 ... Constraint graph creation unit 8 ... Area reduction unit 10 ... Wiring shortening part 12 ... Termination determination unit 14 ... Output section 100 ... Weight determination unit 102 ... Element moving unit 104 ... Termination determination unit 106 ... Line inversion unit 108 ... Merge processing unit

Front page continued (58) Fields surveyed (Int.Cl. ⁷ , DB name) H01L 21/82 G06F 17/50 H01L 21/822 H01L 27/04

Claims (14)

(57) [Claims] 1. The length of the wiring is obtained by moving a specific object in a wiring pattern formed by wiring between objects according to a load showing a performance of wiring shortening in consideration of a shortening priority set for each wiring. And a first step of forming a tree with a plurality of objects having a predetermined positional relationship as nodes, respectively, and for each tree, from the leaf side of the tree,
A second step of calculating the load of the parent node with respect to the child node based on the load of the child node; and a second step of specifying an object corresponding to a tree part whose root node is a node having a load satisfying a predetermined condition as a movement target . And the predetermined positional relationship is such that two objects are in contact with each other.
It is a relationship that is located at the closest possible distance,
How to move the object out of the two objects
Object located upstream of the parent node, downstream side
Loads a node that has a child node that is located in the node and has no child nodes including leaf nodes
Of the wiring part having one end to the node
Although the wiring is expanded when moving, from the sum of the shortening priority
Move the node among the wiring line segments that have one end at the node
Then, the wiring is shortened, but the sum of the shortening priority is reduced.
With the obtained weight value, the load of the node having the child node is set to the weight value of the node,
All positive load values of the child nodes of the node
The predetermined condition that the load should satisfy is that the load is a negative value.
The wiring shortening method is characterized in that 2. A wiring pattern formed by wiring objects.
A specific object in the
Shows the performance of wiring shortening considering the shortening priority.
The length of the wiring by moving according to the load.
This is a wiring shortening method that shortens the wiring, and
A first step of forming a tree as a node, and for each tree, from the leaf side of the tree,
Of the parent node for the child node based on the load of the child node
The second step of calculating the load and the node having the load satisfying the predetermined condition as the root node
Object corresponding to the tree part
And a third step of specifying the wiring line segment length and the wiring line segment length.
The width of the line segment and the function of the mask layer to which the wiring line segment belongs
A wiring shortening method characterized by: 3. The procedure reduces the layout area.
To be executed after moving the object for
The movement of the object for reducing the layout area is to move the object so as to be arranged on one side in the one-dimensional compaction direction as closely as possible. Is a moving target in order from an object located on the other side in the one-dimensional compaction direction, the wiring shortening method according to claim 1. 4. A condition in which all objects cannot be moved even after executing the procedure is repeated twice, a condition in which all objects cannot move even after executing the procedure, and the number of repetitions is defined. An end condition consisting of one of the conditions of exceeding the number of times and the condition of the change in cost corresponding to the weighted sum value of the lengths of the wiring lines being smaller than the specified value, or a condition defined by a predetermined combination. The wiring shortening method according to any one of claims 1 to 3, wherein the procedure is repeatedly executed until the following is satisfied. 5. The relationship between each object is expressed by a constraint graph, and the objects are moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed. The wiring shortening method described in the item. 6. The object may be moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed, and the length of the wiring line segment may become 0 when the movement of the object is completed. For example, after reversing the positional relationship between both ends of the wiring line segment, modifying the structure of the tree and re-determining the load, it corresponds to the tree part whose root node is a node having a load satisfying a predetermined condition. The wiring shortening method according to any one of claims 1 to 3, wherein all the objects to be moved are moved as a unit. 7. If the length of the wiring line segment is 0 when the object is moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed, and the end condition is satisfied. 5. The wiring shortening method according to claim 4, further comprising reversing the positional relationship between both ends of the wiring line segment and repeating the procedure until the end condition is satisfied again. 8. A wiring line whose length becomes 0 when the object is moved within a range in which the positional relationship between both ends of the wiring line segment is not reversed and the end condition is satisfied. If there is a tight branch between the vertices corresponding to both ends of the minute, merge both ends of the wiring line segment and repeat the above procedure until the end condition is satisfied again. The wiring shortening method according to claim 4. 9. A converged state by specifying a portion where the layout is changed and displaying the difference between before and after the change or by displaying a change state of at least one of the total wiring length and the cost. 5. The wiring shortening method according to claim 4, wherein the number of repetitions input by the user is used as the ending condition. 10. A wiring shorten processing method according to claim 1 or 2, characterized in that to enable set up speed priority limited to specific design rules and wiring lines fractionation to a specific part of the layout. 11. The length of the wiring is obtained by moving a specific object in a wiring pattern formed by wiring between objects according to a load showing a performance of wiring shortening in consideration of a shortening priority set for each wiring. In a wiring shortening device for shortening, a means for forming a tree with a plurality of objects having a predetermined positional relationship as nodes, and for each tree, sequentially from the leaf side of the tree,
And a means for calculating the load of the parent node with respect to the child node based on the load of the child node, and a means for specifying an object corresponding to a tree part having a node having a load satisfying a predetermined condition as a root node as a movement target. and, wherein the predetermined positional relationship, contact two objects to each other
It is a relationship that is located at the closest possible distance,
How to move the object out of the two objects
Object located upstream of the parent node, downstream side
Loads a node that has a child node that is located in the node and has no child nodes including leaf nodes
Of the wiring part having one end to the node
When moved, the 0 wiring is expanded, but is it the sum of the shortening priority?
Move the node among the wiring line segments that have one end at the node
Then, the wiring is shortened, but the sum of the shortening priority is reduced.
And the load of the node having the child node as the weight value obtained by
All positive load values of the child nodes of the node
The predetermined condition that the load should satisfy is that the load is a negative value.
Wiring shortened and wherein the at. 12. A wiring pattern formed by wiring between objects.
The specific object in the
The performance of wiring shortening considering the shortening priority
By moving according load indicating the length of the wiring
In line shortening device for shortening the respective multiple objects having a predetermined positional relationship Roh
Means for forming a tree as a node, and for each tree, from the leaf side of the tree,
Of the parent node for the child node based on the load of the child node
A means for calculating the load and a node having a load satisfying a predetermined condition as a root node
Object corresponding to the tree part
And means for specifying the shortening priority of each wiring line segment,
The width of the line segment and the function of the mask layer to which the wiring line segment belongs
A wiring shortening device characterized by: 13. The length of the wiring is obtained by moving a specific object in a wiring pattern formed by wiring between objects according to a load showing a performance of wiring shortening in consideration of a shortening priority set for each wiring. In a computer-readable storage medium storing a wiring program for shortening, a program code for forming a tree with a plurality of objects having a predetermined positional relationship as nodes, and a leaf of the tree for each tree. From the side,
Program code for calculating the load of a parent node on the child node based on the load of the child node, and program code for specifying an object corresponding to a tree part whose root node is a node having a load satisfying a predetermined condition as a movement target comprising a predetermined positional relationship, contact two objects to each other
It is a relationship that is located at the closest possible distance,
How to move the object out of the two objects
Object located upstream of the parent node, downstream side
Loads a node that has a child node that is located in the node and has no child nodes including leaf nodes
Of the wiring part having one end to the node
When moved, the 0 wiring is expanded, but is it the sum of the shortening priority?
Move the node among the wiring line segments that have one end at the node
Then, the wiring is shortened, but the sum of the shortening priority is reduced.
And the load of the node having the child node as the weight value obtained by
All positive load values of the child nodes of the node
The predetermined condition that the load should satisfy is that the load is a negative value.
Storage medium characterized by at. 14. A wiring pattern formed by wiring between objects.
The specific object in the
The performance of wiring shortening considering the shortening priority
The length of the wiring by moving according to the load shown
A computer reading that stores a wiring program that shortens
In the removable storage medium, the program stores a plurality of objects each having a predetermined positional relationship.
Program code for forming a tree as a code, and for each tree, from the leaf side of the tree,
Of the parent node for the child node based on the load of the child node
The program code that calculates the load and the node that has the load that satisfies the specified condition as the root node
Object corresponding to the tree part
And a shortening priority of each wiring line segment, the length of the wiring line segment and the wiring line segment.
The width of the line segment and the function of the mask layer to which the wiring line segment belongs
A storage medium characterized by: