JPH05282402A - Arrangement problem solving system - Google Patents

Abstract

PURPOSE:To efficiently obtain an approximate solution nearer to the optimum solution finally by giving evalutation for an (already set refererence target) selected as a reference target even once previously on arrangement results before and after the (shift) of an arranged target. CONSTITUTION:When an arranging place decided transiently by a temporary arranging place decision part 2c shows the one near to the reference target of an arranging target, it is decided as the arranging place by an arranging place decision part 2f. When it is not, the arranged target not relating to the reference target in the neighborhood of the reference target is shifted by a shift operating part 2d, and the arranging target is intruded into a null space. Thence, the sum of (degree of approximation X distance) between the already set reference target selected as the reference target once previously and the arranged target set as the reference target is held as an evaluation reference value in step 2e, and when the evaluation reference values of the already set reference target before and after shift are not deteriorated, a result of shift is decided as the arranging place, and when it is not, an arranging state before shift is restored. Those operations after evaluation are performed by the decision part 2f.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention optimally selects an unarranged object to be arranged next, in consideration of several arranged objects, for example, an arrangement condition at the time of arranging ICs of electronic parts and VLSI chip elements. Related to the placement problem solving system.

[0002]

2. Description of the Related Art Problems such as how to arrange a plurality of ICs on a printed circuit board, how to arrange elements on a VLSI chip, and how to arrange various objects to be arranged exist in various fields. ..

This kind of placement problem is extremely difficult to solve in polynomial time when trying to find the optimum solution, so attempts have been made to efficiently find a so-called approximate solution that is close to the optimum.

In general, first, for all of the already-arranged objects that have already been arranged, the optimum arrangement object for the next arrangement is selected and decided based on the degree of proximity. After that, a place that is considered to be optimal around the already arranged arrangement area is selected, and the arrangement target selected and determined as described above is arranged at that place.

However, in order to determine the placement target and its placement location in consideration of such a procedure, it is necessary to check all the possible placement locations around the placement area each time. There is a problem that it takes time to decide the place.

[0006]

As described above, in the conventional placement problem solving system, in order to decide the placement location of the placement object selected and decided, the placement area of the already placed placement object is decided. There is a problem that it is very troublesome because it is necessary to check all the surrounding places that can be arranged each time.

The present invention has been made in consideration of such circumstances, and an object of the present invention is to reduce the number of checks of places where arrangement is possible without sacrificing the satisfaction of the arrangement result and simplify the operation. Another object of the present invention is to provide a placement problem solving system capable of solving placement problems efficiently.

[0008]

In order to achieve the above-mentioned object, the invention according to claim 1 should be arranged next in a state in which several objects to be arranged are arranged at a plurality of predetermined arrangement places. A system for selecting and deciding an optimal placement location of a non-placement object from among the non-placement locations, and showing the degree to which these two placement objects are desired to be placed close to each other for any two placement objects. A table that defines the proximity, a reference target selection unit that selects an arranged target that serves as a reference for selecting an unplaced target to be arranged next by referring to this table, and this reference target selection Placement target selecting means for selecting and determining an unplaced target to be placed closest to the reference target selected by the means by referring to the table, and an unplaced pair selected and determined by the placement target selecting means. Tentatively determining means for tentatively determining the unplaced location closest to the reference object as the placement location, and the unplaced location selected and determined by the placement target selecting means in the unplaced location tentatively determined as the placement location by the tentative determination means. Placement location determining means for placing the target,

A shift operation for performing an operation of "shifting" an unplaced object, which is tentatively determined as an arrangement location by the tentative determining means, around an already-referenced object and which is irrelevant to the reference object. Means and

As a result of performing the operation of "shifting" by the shifting operation means, the placement object is interrupted in the vacant space, and at the same time with respect to each of the existing reference objects previously selected as the reference object at least once. , The existing reference object and the sum of the products of the proximity and the distance between all the arranged objects having the reference object as the reference object are held as an evaluation reference value, and the arrangement object is displaced by the displacement operation means. If the evaluation reference value of each existing reference object does not deteriorate before and after, the shifted result is output to the placement location determining means, and if not, the evaluating means for returning the placement object to the placement state before shifting It is equipped with.

[0011]

According to the invention corresponding to claim 1, the arranged object serving as a reference for selecting the unarranged object is determined by referring to the table, and the determined arranged object is used as the reference object. After selecting and deciding the unplaced object to be placed by referring to the table, the place where the unplaced object should be placed is decided, and the reference object around the reference object during the placement is the reference object. This "shift" is used by using a layout improvement method that shifts the placement targets that have little relation to each other, and then causes the placement targets to interrupt the empty space.
Since the "already-referenced object" that has been selected as the reference object is evaluated even once even with respect to the layout results before and after, it is possible to efficiently obtain the final approximate solution close to the optimal solution.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An arrangement problem solving system of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of one embodiment. The system of this embodiment is roughly composed of three blocks: an arrangement target data input unit 1, an arrangement problem solution processing unit 2, and an arrangement result output unit 3.

The placement target data input unit 1 relates to placement targets as a table defining, for each arbitrary two placement targets among all the placement targets, the degree of proximity indicating the degree to which those placement targets are to be placed close to each other. It is for inputting data and data on an area in which these placement objects are placed. There are many placement targets to be placed by this input data, and what placement constraint condition exists between these placement targets, that is, which placement target and which placement target should be placed close to each other. The condition is given. It also shows how the placement area is defined.

The proximity degree indicates the degree to which two placement targets are to be placed close to each other, and is defined as, for example, a positive integer value, and is given as a larger numerical value as the degree to which the placement targets are placed closer to each other is higher.

The placement problem solution processing unit 2 solves the placement problem in accordance with the input data given from the placement target data input unit 1, and the reference subject selection unit 2
a, a placement target selection unit 2b, a placement location temporary determination unit 2c, a shift operation unit (shift processing unit) 2d, an evaluation unit 2e, and a placement location determination unit 2f. These six processing functions operate in relation to each other.

Basically, first, the reference object selecting section 2a serves as a reference arrangement for selecting and deciding the following arrangement objects from the already arranged arrangement objects based on the input data. The completed target is selected as the reference target α. The method of selecting and determining the reference object α depends on the type of problem to be subjected to the placement process, for example, the problem of placing a component on a board. Specifically, for example, for each of the arranged objects, the sum of the degrees of proximity to the unarranged object is obtained, and the arranged object having the maximum sum of the degrees of proximity is selected as the reference object α. Alternatively, the proximity to the unplaced object is evaluated under an evaluation function described later, and the reference object α is selected according to the evaluation result. It should be noted that if there is no already-arranged object that has already been arranged, that is, if all the objects to be arranged are in the initial state where they are not arranged, this processing procedure is omitted.

Next, the placement target selection unit 2b is activated to select one unplaced target β having the maximum proximity to the reference target α selected as described above. When there are a plurality of unplaced objects β with the highest degree of proximity, one of them is selected according to a predetermined selection rule.
Further, at the initial stage when the reference object α is not defined, one of the unplaced objects β is selected according to a predetermined rule.
Such selection is performed in consideration of, for example, the importance, function, and role of the unallocated object β.

The tentative placement location deciding unit 2c tentatively decides the placement location of the placement target, and if the placement location is near the reference target of the placement target, the placement location determining unit 2f decides the temporary placement location as the placement location. If this is not the case, the shift operation unit 2d performs an operation of shifting the arranged objects (a plurality of) that are located around the reference object and are not very related to the reference object, and as a result, make an empty space. Interrupt the placement target. Next, in the evaluation unit 2e, for each target previously selected as the reference target at least once, that is, for the already-referenced target, between the already-referenced target and all the arranged objects having the reference target as the reference target. Always hold the sum of proximity x distance as an evaluation reference value, and if the evaluation reference value of each existing reference target does not deteriorate before and after shifting,
The post-evaluation operation of determining the shifted result as the placement location and returning to the placement state before shifting otherwise is performed by the placement location determination unit 2f. The placement determination of the placement target in the placement problem solving system configured as described above will be described in more detail.

Now, as shown in FIG. 2, ten IC parts (arrangement targets,
(Hereinafter simply referred to as “component”) has already been placed, the component K is the reference target in the reference target selection unit 2a, the component S is the selected component in the placement target selection unit 2b, and the temporary placement location determination unit 2c is It is assumed that it is temporarily arranged at the position (x4, y4). Here, (m, n) represents a column-row pair of grid points.

In order to avoid making the description complicated, it is assumed that the size of all parts is [1 cm × 2 cm], and the orientation when arranging them is that the first pin is in the upper left direction. To do. The spacing (size) of the grid in which these components are arranged is [2 cm × 3 cm]. In addition, the distance between the parts is the so-called Manhattan distance between the first pins (grid points), that is, the length of the shortest path when tracing the grid in the vertical and horizontal directions. For the evaluation degree of the placement result, the product of the distance between the grid points where the two parts are placed and the degree of coupling (the number of connections) between the two parts was found for all the parts pairs, and the sum was taken. The value. In this case, the smaller the sum of products, the better the evaluation degree.

In FIG. 2, between A and B, and K and S for simplicity.
Only the degree of coupling between is shown using a line. The degree of coupling between K and B is 0. The degree of coupling between other arranged parts is regarded as 0 for simplicity. Such a restriction does not affect the following essential explanation. Also, in FIG. 2, only K and A are
It is assumed that the component is a component (already-referenced component) previously selected as a reference target at least once.

Now, in FIG. 2, the coupling degree between the reference part object K and the selected part S is 5, and the distance is (2 cm × 2) + (3 cm
× 1) = 7 cm. It is assumed that it is determined that the distance is too large despite the high degree of coupling. It is assumed that such a criterion is given in advance in the placement target data input unit 1 together with data such as the size of the component, the orientation of the placement, and the degree of coupling.

In the above case, in the shift processing section 2d, the operation is performed to shift the arranged parts (a plurality of) arranged around the reference part K and not related to the reference part, resulting in an empty space. To interrupt the selected part. In FIG. 2, since the component B is a component that is not very related to K, the components B and C are shifted downward and the selected component S is inserted into the empty space, as shown in FIG.

Next, in the evaluation unit 2e, the part previously selected as the reference object at least once (the existing reference object, FIG.
Then, for each of K and A), recalculate the sum of (proximity) × distance (evaluation value) between the existing reference object and all the arranged objects with the reference object and hold it. Compare with an evaluation reference value before a certain "shift". If the sum of the evaluation values after shifting does not exceed the sum of the evaluation reference values before shifting, the shifted result is determined as the placement location, and if not, the placement state before shifting is restored. In the state of FIG. 2, the evaluation reference values of the parts A and K are (2 cm × 2 pieces) = 4,
(7 cm × 5 lines) = 35, and the total sum is 39. On the other hand, when the evaluation values of A and K are calculated in FIG. 3, (5 cm × 2 lines) = 10 and (3 cm × 5 lines) = 15, respectively, and the total sum is 25. Since the sum 25 of the evaluation values after the shift does not exceed the sum 35 of the evaluation reference values before the shift, the shift result (FIG. 3) is determined as the placement location.

The above method is basically possible because of the concept of reference object. This is because, in general, an arrangement object having a high degree of proximity to the reference object is arranged around the reference object, and therefore it is appropriate to perform the evaluation mainly on the reference object.

The applicant of the present application filed an application similar to the above-mentioned content in advance, but this prior application specification (Japanese Patent Application No. 2-18)
No. 2617 specification), an arranged object that is a reference for selecting an undisposed object to be arranged next is selected as a reference object, and an undisposed object having the highest degree of proximity to the selected reference object. Has been proposed, and a non-arranged location closest to the reference object is determined as an arrangement location, and the selected non-arranged object is arranged at the non-arrangement location determined as the arrangement location. According to the method of this prior application, it is possible to efficiently obtain a relatively good approximate solution in which the arrangement target is not concentrated in one place of the arrangement area, but there is a problem in obtaining an approximate solution that is closer to the optimum. is there.

As described above, in the conventional placement problem solving system, in order to determine the placement location of the placement object that has been selected and decided, the placement locations around the already placed area where the placement object has already been decided are set. Everything needs to be checked,
There was a problem that it was very troublesome. In addition, the method of solving it, that is, the method of using the concept of the reference object also has a problem in that a more optimal solution is obtained.

In this embodiment, the above problems can be eliminated. That is, in the arrangement improvement method of performing an operation of "shifting" an already-arranged object that is not related to the reference object in the vicinity of the reference object during the arrangement, and as a result causes the arrangement object to be interrupted in an empty space,
The placement results before and after the shift can be evaluated, and finally an approximate solution close to the optimal solution can be obtained.

When the embodiment of the present invention described above is used in combination with the concept of reference object selection of the prior application, in the problem of arranging a large number of objects on a practical level, the efficiency and the final arrangement result are evaluated. , It will appear as a very large effect.

According to the placement problem solving system, the reference object is selected from the placed objects, the unplaced object having the highest degree of proximity to the reference object is selected, and the unplaced location closest to the reference object is selected. Priority is given to the process of arranging in.

In the process of arranging by this arranging method, when the operation of "displaced" the arranged object in order to maintain a better arrangement state, as the evaluation method before and after the displacement, At least once, for each target selected as the reference target (existing reference target),
The sum of proximity x distance between the existing reference object and all arranged objects that are the reference object is retained as an evaluation reference value, and the evaluation reference value of each existing reference object before and after shifting. If is not deteriorated, the shifted result is determined as the placement location, and if not, the previous placement state is restored.

Since this is performed, the number of checks for determining the placement location can be reduced, and the placement can be improved efficiently during the placement, so that finally a better evaluation is obtained than in the case of the conventional system. be able to.

The present invention is not limited to the above embodiment. For example, in the above-described embodiment, the size of the ICs to be arranged is uniform, but it is needless to say that the arrangement problem can be solved in consideration of the case where the sizes of the respective objects are different. Further, it goes without saying that the orientation of the placement target is taken into consideration. Others The present invention can be variously modified and implemented within the scope of the invention.

[0035]

As described above, according to the present invention, the number of checks of places where arrangement is possible is reduced and the arrangement problem is solved easily and efficiently without sacrificing the satisfaction of the arrangement result of the arrangement object. It is possible to provide an arrangement problem solving system capable of doing so.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a placement problem solving system according to an embodiment of the present invention.

FIG. 2 is a diagram showing an intermediate process of placement of placement targets in the embodiment of FIG.

FIG. 3 is a diagram after performing a “shift” operation in the embodiment of FIG.

[Explanation of symbols]

1 ... Arrangement target data input section, 2 ... Arrangement problem solving processing section,
2a ... Reference object selection section, 2b ... Arrangement object selection section, 2c ...
Placement temporary decision unit, 2d ... Shift operation unit (shift processing unit), 2e ... Evaluation unit, 2f ... Placement determination unit, 3 ... Placement result output unit.

Claims (1)

[Claims] 1. A system for selecting and deciding an optimum placement location of a non-placement object to be placed next, in a state in which several placement objects are placed at a plurality of predetermined placement locations. Therefore, a table that defines the degree of proximity for each of two arbitrary placement targets among a plurality of placement targets and the degree of closeness of these placement targets, and the next placement is performed by referring to this table The reference object selecting means for selecting the arranged object as the reference for selecting the unarranged object as the reference object, and the unarranged object to be arranged closest to the reference object selected by the reference object selecting means. Arrangement target selecting means for making a selection decision by referring to the table, and the non-arrangement object selected and decided by the arrangement target selecting means is tentatively decided as the arrangement place at the non-arrangement location closest to the reference object. Deciding means, arranging place deciding means for arranging the unplaced object selected and decided by the arranging object selecting means in the unplaced place tentatively decided as the arranging place by the tentative deciding means, and arranging as the arranging place by the tentative deciding means The tentatively determined non-placed object has an operation means for performing an operation of "displaceing" an arranged object around the reference object and having little relation to the reference object, and "displacement" by the operation means for shifting. As a result of performing the operation, the placement target is interrupted in the vacant space, and the existing reference target and the reference target for each existing reference target that has been selected as the reference target at least once before. The sum of the products of the proximity and the distance with all the arranged objects is held as an evaluation reference value, and before the displacement of the arranged objects by the shifting operation means is performed. After that, if the evaluation reference value of each existing standard object does not deteriorate, the shifted result is output to the placement location determining means, and if not, the evaluating means for returning the placement object to the placement state before shifting A placement problem solving system comprising: