JPH0652261A - Automatic wiring system - Google Patents

Abstract

PURPOSE:To provide the automatic wiring system capable of wiring while maintaining legal required wiring without lowering a wiring efficiency. CONSTITUTION:By an initializing part 1, a cost coefficient is set to a prescribed initial value, and by an intersection avoiding part 2, an intersection of a wiring path of a selected net to a wiring path of a net determined already is evaded. A path searching part 3 obtains the sum of a distance from a source of each cell and a distance from a target of each cell with respect to the selected net by evading the intersection and calculates a wiring cost of a path by using a deviation of the sum to designated line length and a cost coefficient. A cost coefficient changing part 4 changes the set cost coefficient so as to become larger successively, and makes the deviation from the designated line length smaller gradually. Also, the path searching part 3 is constituted so as to determine a path by searching repeatedly a path for reducing the wiring cost in accordance with an altered cost coefficient.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a printed circuit board and an LSI.
The present invention relates to an automatic wiring system for integrated circuits such as (large-scale integrated circuit device).

[0002]

2. Description of the Related Art Automatic wiring by a computer is used for wiring of an integrated circuit such as a printed circuit board or an LSI. In recent years, the degree of integration of integrated circuits has improved, and in the design of wiring a pattern on a printed circuit board, a method of dividing a wiring region into a plurality of cells and sequentially determining a wiring route for each divided cell is used. Has been.

In this case, in the printed circuit board and the LSI, it is necessary to wire terminal pairs (nets) having the same potential to each other without crossing nets having other potentials. Conventionally, for example, a maze method is used.

As shown in FIG. 12, this maze method divides a printed circuit board into a plurality of cells to form a terminal pair, for example, each cell "SL1" to another cell "SL2".
Up to this, labeling is sequentially performed for each cell, and when mutual labeling is performed in the same cell, a route is determined.

For example, as shown in FIG. 12, a cell "SL" is used.
"1" is written to the top, bottom, left and right around 1 ", and then" 2 "is written to the top, bottom, left and right around the written" 1 "while avoiding obstacles.

Further, "3" is written vertically and horizontally around the written "2" while avoiding an obstacle, and two cells are overlapped with each other, for example, the cell shown in FIG.
The cell passing through this (A) becomes the wiring to be obtained. Then, since the cells on the route once obtained become obstacles, the net to be routed later has to be detoured for the already determined route.

Further, recently, the amount of memory required for processing and the processing time have increased. Therefore, in order to cope with the high-speed operation of the circuit, there are increasing cases in which the line length of a terminal pair (hereinafter, referred to as a net) that determines the path and should be at the same potential is limited and designed. Conventionally, when a constraint is imposed on the line length of a net, the wiring route of the net is preferentially determined.

[0008]

However, when there are many nets with specified line lengths, even if the restricted nets are preferentially routed, as the routing process progresses There are a lot of nets that cannot maintain their length. Therefore, there is a problem that the wiring rate is reduced and the number of wiring design steps is increased.

The present invention has been made in view of the above circumstances, and an object of the present invention is not to reduce the wiring rate even when there are many nets having a specified wiring length. An object of the present invention is to provide an automatic wiring system capable of performing wiring while maintaining designated wiring.

[0010]

In order to solve the above problems and achieve the object, the present invention has the following constitution. FIG. 1 shows the principle of the present invention. The present invention will be described below with reference to FIG.

According to the automatic wiring method of the present invention, the wiring region of one wiring layer is divided into a plurality of cells, a plurality of nets that should have the same potential are selected, a source and a target are set for each net, and the source and the target are set. The wiring route is determined when the length of the wiring is designated, and is configured as follows.

The initialization unit 1 sets a cost coefficient used for calculating the wiring cost of a route to a predetermined initial value. The cross avoidance unit 2 avoids crossing of the wiring route of the selected net with the wiring route of the already determined net.

The route search unit 3 obtains the sum of the distance from the source of each cell and the distance from the target of each cell for the net selected under the avoidance of the intersection by the intersection avoidance unit 2 and is designated as described above. The wiring cost of the route is calculated using the error of the sum with respect to the line length and the cost coefficient, and the route having the smaller wiring cost is searched.

The cost coefficient changing section 4 sequentially changes the cost coefficient set by the initializing section 1 to notify the route searching section 3 of the cost, and gradually reduces the error from the designated line length. .

The route searching unit 3 is configured to repeatedly search for a route which reduces the wiring cost in accordance with the cost coefficient changed by the cost coefficient changing unit 4 and determine the route.

More preferably, a multilayer wiring layer having one or more wiring layers other than the one wiring layer is formed, and the wiring area of each wiring layer is divided into a plurality of cells. Avoids the wiring route of the already determined net as an obstacle when a net is selected between different wiring layers, and the route searching unit 3 selects between the different wiring layers under the avoidance of the intersection by the intersection avoiding unit 2. The wiring cost of the route is calculated by using the cost error and the error of the sum for the specified line length to obtain the sum of the distance from the source of each cell and the distance from the target of each cell for the specified net. It is preferable to be configured to calculate and search for a route with a small wiring cost.

Further, the route searching unit 2 performs a route search from the target to the source by using a maze method to obtain a distance from each target to each cell, and performs a route search from the source to the target to source each cell. It may be configured to obtain the distance from.

[0018]

According to the present invention, the initialization unit 1 sets the cost coefficient to a predetermined initial value, and the crossing avoiding unit 2 sets the crossing of the wiring route of the selected net to the wiring route of the already determined net. To avoid.

Then, the route search unit 3 obtains the sum of the distance from the source of each cell and the distance from the target of each cell for the selected net under the avoidance of the intersection by the intersection avoidance unit 2. Further, the route searching unit 3 calculates the wiring cost of the route using the error of the sum with respect to the designated line length and the cost coefficient, and searches for the route having the smaller wiring cost.

Next, when the route with a small wiring cost searched by the route search unit 3 does not satisfy the designated line length, the cost coefficient changing unit 4 sequentially increases the cost coefficient set by the initialization unit 1. The route is changed and gradually changed so that the route approaches the specified line length.

Then, the route search unit 3 repeatedly searches for a route which reduces the wiring cost in accordance with the greatly changed cost coefficient, and determines the route, and finally the cost is sufficiently minimized. Then, the error of the sum from the designated line length gradually becomes zero.

Therefore, the designated line length can be maintained, and
Wiring can be performed without reducing the wiring rate.

[0023]

EXAMPLES Specific examples of the present invention will be described below.
FIG. 2 is a device configuration diagram for realizing the automatic wiring system according to the present invention.

The initialization unit 1 sets a cost coefficient used for calculating the wiring cost of a route to a predetermined initial value. The cross avoidance unit 2 avoids crossing of the wiring route of the selected net with the wiring route of the already determined net.

The cross avoiding unit 2 regards the crossing of routes as a wiring cost and minimizes the wiring cost to minimize the number of crossing. The application was filed in Japanese Patent Application No. 62-296674 and is described in detail therein.

The route search unit 3 obtains the sum of the distance from the source of each cell and the distance from the target of each cell to the net selected under the avoidance of the intersection by the intersection avoidance unit 2 and is designated as described above. The wiring cost of the route is calculated using the error of the sum with respect to the line length and the cost coefficient, and the route having the smaller wiring cost is searched.

The cost coefficient changing unit 4 sequentially changes the cost coefficient set by the initialization unit 1 to notify the route searching unit 3 of the cost, and gradually reduces the error from the designated line length. .

The route searching unit 3 is configured to repeatedly search for a route which reduces the wiring cost in accordance with the cost coefficient changed by the cost coefficient changing unit 4 and determine the route.

Further, the distance calculation unit 5 performs a route search from the target to the source by using the maze method to obtain a distance from the target of each cell, and performs a route search from the source to the target to determine the distance from the source of each cell. It is designed to calculate the distance and.

Each of these parts is a CP inside the computer.
It is provided in a U (central processing unit) or a microprocessor. <Embodiment 1> FIG. 3 is a flow chart showing an automatic wiring method of the embodiment, FIG. 4 is a diagram in which a target is searched from a source by a maze method and a distance is calculated, and FIG. It is the figure which was calculated. FIG. 6 is a diagram in which costs are calculated from a source to a target as a cost function, and FIG. 7 is a diagram showing patterns of wiring results. In the example shown in FIG. 4, the target T and the source S
Wiring in the case where the distance between the nets is designated by 12 will be described.

Next, the processing of the automatic wiring method in such an embodiment will be described with reference to FIG. (A) First, in step 101, the initialization unit 1 determines a cost coefficient. The error from the specified line length is regarded as the cost, and the cost function is defined as follows.

Cost function = wiring length + a × error with respect to designated line length (1) where a is a cost coefficient, and such a cost coefficient is an initial value in the example shown in FIG. The value is set to 1. (B) Next, in step 102, a plurality of nets that should have the same potential are selected. In the example shown in FIG. 4, a net (a portion indicated by the target T and the source S) is selected from each of the cells divided on the substrate. (C) Then, in step 103, the wiring route of the current net is deleted as an obstacle in order to avoid the intersection of the wiring routes of the nets selected by the intersection avoidance unit 2 with respect to the wiring route of the current net. (D) Further, in step 104, the source S and the target T are set for the selected net. (E) In step 105, the distance calculation unit 5 calculates the distance from the target T to the source S using the maze method. In the example shown in FIG. 4, labeling is sequentially performed for each cell from the target T to the source S.

As shown in FIG. 4, the distance value "1" from the target T is written in the upper, lower, left, and right cells centering on the target T, and the obstacle is avoided around the next written distance value "1". The distance value “2” from the target T is written in the upper, lower, left and right cells. Such processing is sequentially repeated while avoiding obstacles.

Next, the distance from the source S to the target T is set by using the maze method. In the example shown in FIG. 5, labeling is sequentially performed for each cell from the source S to the target T. As shown in FIG. 5, the distance value “1” from the source S is written in the upper, lower, left, and right cells centering on the source S, and the distance value “1” written next is used as the center to avoid obstacles and move up, down, left, and right. The distance value "2" from the source S is written in the cell.
Such processing is sequentially repeated while avoiding obstacles. (F) Next, in step 106, the route searching unit 3 searches for a route whose wiring cost is minimized. Since the cost coefficient a is set to 1 from the equation (1), the cost function = wiring length + specified line length error (2).

Here, the designated line length is 12, and the designated line of the sum of the distance from the source of each cell and the distance from the target of that cell is used by using the distance value of each cell shown in FIGS. 4 and 5. The cost of the error with respect to the length is calculated by the following formula.

[Cell value in FIG. 4 + Cell value in FIG. 5−Specified line length | ... (3) For example, in the cell on the left side of the source S in FIG. 6, the cost of the distance from the source S ( The wiring length 1) 1 and the error 0 (11 + 1-12) with respect to the designated line length are added to obtain the cost 1. In the cell on the right side of the source S, the cost 1 (wiring length 1) of the distance from the source S and the error 2 (9 + 1-12) with respect to the designated line length are added to obtain the cost 3.

In the cell on the left side of the target T, the cost 9 of the distance from the source S and the error (9 + 1-12) with respect to the designated line length are added to obtain the cost 11,
In the cell on the right side of the target T, the cost 11 of the distance from the source S and the error 0 (11 + 1
-12) and are added to give a cost of 11.

Thereafter, the cost of each cell is calculated in the same manner, and the cost from the source S is written in each cell, as shown in FIG.
Labeling from the source S as shown in is obtained. Then, the minimum cost sequence finally assigned to the target T,
That is, the route shown by the thick solid line as shown in FIG. 7 is the route to be obtained. (G) Further, in step 107, the route is registered in a memory (not shown). (H) In step 108, it is judged whether or not all the nets are finished. If all the nets are not finished,
Return to step 102 and step 102 to step 1
The processing up to 07 is repeated for each net. (I) Then, when all the nets are finished, it is judged in step 109 whether or not the constraints on all the designated line lengths are satisfied. Returning to 101, the processing from step 101 to step 108 is repeated.

That is, in this step, when the route with a small wiring cost searched by the route search unit 3 does not satisfy the designated line length, the cost coefficient changing unit 4 sequentially sets the cost coefficients set by the initialization unit 1. We will make major changes, and gradually change the route so that it approaches the specified line length. Then, the route search unit 3 repeatedly searches for a route that reduces the wiring cost in accordance with the significantly changed cost coefficient, determines the route, and finally, the cost is sufficiently minimized. The error of the sum from the designated line length gradually becomes zero. (J) If all the constraints regarding the designated line length are satisfied, the process proceeds to step 110, and the process ends. In the first embodiment, the cost coefficient is set to 1 first, so
The constraint regarding the specified line length is satisfied without significantly changing the cost coefficient.

Therefore, the designated line length can be maintained, and
Wiring can be performed without reducing the wiring rate. <Embodiment 2> In the embodiment 2, a two-layer wiring having one wiring layer other than the one wiring layer of the embodiment 1 is configured, and the wiring region of each wiring layer is divided into a plurality of cells.

Although not shown, the cross avoidance unit 2 avoids the wiring route of the net that has already been determined as an obstacle when a net is selected between different wiring layers.

The route search unit 3 obtains the sum of the distance from the source of each cell and the distance from the target of each cell for the net selected between different wiring layers under the avoidance of the intersection avoiding unit 2. The wiring cost of the route is calculated using the error of the sum and the cost coefficient with respect to the designated line length, and a route with a small wiring cost is searched for.

FIG. 8 is a diagram in which the distance is calculated by searching from the target to the source in the two-layer wiring by the maze method, and FIG. 9 is a diagram in which the distance is calculated from the source in the two-layer wiring by the maze method to the target. FIG. 10 is a diagram in which the cost is calculated from the source to the target in the two-layer wiring as a cost function, and FIG. 11 is a diagram showing the pattern of the two-layer wiring result.

Next, the processing of the automatic wiring system of the second embodiment when the designated line length is 7, for example, will be described with reference to the drawings.
First, in FIG. 8, “1” is written in the upper, lower, left and right cells centering on the target T of the first layer, and “2” is written in the upper and lower cells of “1” of the left and right cells of the target T. Then, with respect to the lower right cell "2" of the target T, the right cell sequentially increases its value by one. By repeating such processing in sequence, the value of each cell in the first layer is determined.

On the other hand, the value in each cell in the second layer is set to be a value obtained by increasing the value in the corresponding cell in the first layer by one. For example, cells on the left, right, top and bottom of the target T of the first layer "
The value of the corresponding second layer cell for 1 "is"
2 ″. In this way, the value of each cell in the second layer is determined.

Next, in FIG. 9, the distance from the source S arranged on the second layer to the target T is also obtained in the same manner as the above method, but the position of the source S is the position of the target T on the first layer. Since they are symmetrical, the value of each cell of the second layer is the same as the value of each cell of the first layer symmetrically arranged.

Next, in FIG. 10, labeling from the source S is sequentially performed for each cell using the equations (2) and (3) in the first embodiment. Note that the cost coefficient is 1 here as well. For example, in the cell on the right of the target T on the first layer, the wiring length is 8, and the error from the specified length is 2 (1+
Since it is 4-7), the wiring cost is 10.

Next, in FIG. 11, the source S of the second layer
Through cells 1, 2, 3, 4 from the first layer of cells 5, 8,
The target T and the route having the designated length 7 are determined. Therefore, even in the two-layer wiring, the specified line length can be maintained and the wiring can be performed without reducing the wiring rate. Although the two-layer wiring is described in the second embodiment, the present invention can be applied to other multilayer wiring.

[0049]

According to the present invention, the sum of the distance from the source of each cell and the distance from the target of each cell is found, and the route is calculated by using the error and the cost coefficient of this sum for the specified line length. If the searched route with a small wiring cost does not meet the specified line length, the cost coefficient is changed and the wiring cost is reduced according to the changed cost coefficient. Since the route is repeatedly searched to determine the route, the error of the sum with respect to the designated line length gradually becomes smaller.

Therefore, the specified line length can be maintained and
Wiring can be performed without reducing the wiring rate.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention.

FIG. 2 is a configuration diagram of an embodiment of the present invention.

FIG. 3 is a flowchart showing an automatic wiring system according to an embodiment.

FIG. 4 is a diagram in which distances are calculated by searching from a target to a source by a maze method.

FIG. 5 is a diagram in which distances are calculated by searching a source to a target by a maze method.

FIG. 6 is a diagram in which costs are calculated from a source to a target as a cost function.

FIG. 7 is a diagram showing a pattern of wiring results.

FIG. 8 is a diagram in which distances are calculated by searching from a target to a source in a two-layer wiring by the maze method.

FIG. 9 is a diagram in which distances are calculated by searching from a source to a target in a two-layer wiring by the maze method.

FIG. 10 is a diagram in which costs are calculated from a source to a target in a two-layer wiring as a cost function.

FIG. 11 is a diagram showing a pattern of a two-layer wiring result.

FIG. 12 is a diagram for explaining a maze method.

[Explanation of Codes] 1. Initialization unit 2. Intersection avoidance unit 3. Route search unit 4. Cost coefficient change unit 5. Distance calculation unit

Claims (3)

[Claims] 1. A wiring region of one wiring layer is divided into a plurality of cells, a plurality of nets having the same potential are selected, a source and a target are set for each net, and a wiring length between the source and the target is set. In the automatic routing method that determines the routing route when specified, the initialization unit (1) that sets the cost coefficient used to calculate the routing cost of the route to a predetermined initial value, and the routing of the already determined net An intersection avoidance unit (2) for avoiding an intersection of the wiring routes of the selected net with respect to the route, and a distance from the source of each cell for the selected net under the intersection avoidance unit (2). The sum of the distance from the target of each cell is calculated, the wiring cost of the route is calculated using the error of the sum with respect to the designated line length, and the cost coefficient, and the route with the smaller wiring cost is searched. The route search unit (3) to be searched and the cost coefficient set by the initialization unit (1) are sequentially changed to notify the route search unit (3) to gradually reduce the error from the designated line length. The route searching unit (3) repeatedly searches for a route for reducing the wiring cost in accordance with the cost coefficient changed by the cost coefficient changing unit (4). An automatic wiring method characterized by being configured to determine a route. 2. A multilayer wiring layer having one or more wiring layers other than the one wiring layer, wherein a wiring region of each wiring layer is divided into a plurality of cells, and the intersection avoidance unit (2). Avoids a wiring route of a net that has already been determined as an obstacle when a net is selected between different wiring layers, and the route search unit (3) differs under the avoidance of intersection by the intersection avoidance unit (2). A route is calculated using the error of the sum and the cost coefficient for the specified line length for the net selected between the wiring layers by calculating the distance from the source of each cell and the distance from the target of each cell. 2. The automatic wiring system according to claim 1, wherein the automatic wiring system is configured to calculate a wiring cost of the above and search for a route having a small wiring cost. 3. The route search unit (2) performs a route search from the target to the source by using a maze method, obtains a distance from each target to each cell, and performs a route search from the source to the target to obtain each cell. 3. The automatic wiring method according to claim 1 or 2, wherein the distance from the source is calculated.