JPH04362785A - Combination optimizing device - Google Patents

Abstract

PURPOSE:To shorten processing time by adjusting the applicable ratio of plural arrangement conversion means based on the decision result by a conversion validity decision means deciding the validity. CONSTITUTION:In an applicable ratio adjusting means 5, plural arrangement conversion means are ranked in order of high validity based on a decision result of a conversion validity decision means 6 deciding the validity every temperature at first. Next, the arrangement conversion means are divided into higher ranking arrangement conversion means and lower ranking arrangement conversion means. As for the higher ranking arrangement conversion means, the applicable ratio is increased and as for the lower ranking arrangement means, the applicable ratio is decreased. Further, the higher the ranking becomes, the more the increase rate is increased and the lower the ranking becomes, the more the decrease rate is increased. Based on the increase and decrease rates, the applicable ratio in a next temperature is decided. Namely, the higher the validity decided by the conversion validity decision means 6 deciding the validity of each arrangement conversion means stored in a data base 1 that an arrangement conversion means has is, the more the applicable ratio is increased. Thus, the processing time can be shortened.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for high-speed execution of combinatorial optimization problems in the field of design, particularly in logic circuit design.

0002

2. Description of the Related Art Recently, combinatorial optimization devices have come into widespread use in fields such as logic circuit design. As this combination optimization device, for example, S. Kirkpa
trick, C. D. Gelatt, M. P. Ve
cchi ”Optimization by Sim
"Annealing" (Science
, Vol. 220, No. 4598, pp671-68
A configuration called simulated annealing, which is described in, for example, 0, 1983), is known.

A conventional combinatorial optimization device will be explained below with reference to FIG. In FIG. 3, thin lines indicate control signals, double lines mainly indicate data reference relationships, and thick lines mainly indicate control signals and data flows. In FIG. 3, 31
32 is a database storing a plurality of layout conversion means for generating layout data; 32 is a conversion application means for randomly selecting a layout conversion means from the database 31 and calculating the cost of new layout data generated by the layout conversion means; 33 34 is a cost evaluation means that compares the cost of the conventional layout data and the cost of the new layout data to decide whether to adopt the old layout data or the new layout data, and 34 manages and changes a parameter called temperature. Temperature management means 35 is a process termination determining means for observing fluctuations in the arrangement data and terminating the process when the fluctuations disappear; 36 is an initializing means for setting and generating initial values.

[0004] In the configuration of FIG. 3 as described above, the following is performed.
The operation will be explained with reference to the flowchart shown in FIG.

First, in step 401, the initialization means generates a combination of initial values, calculates the cost of the combination of initial values,
A parameter called temperature managed by the temperature management means 34 is initialized.

Next, in step 402, a layout conversion means stored in the database 31 is randomly selected using the conversion application means 32, and new layout data is generated by applying the selected layout conversion means. At 403, the cost evaluation means 33
The cost of the placement data generated in step 402 is calculated by.

In 404, the cost of the new layout data calculated in 403, the cost of the conventional layout data, and the temperature management means 3
Based on the temperature managed in step 4, the cost evaluation means 33 determines whether to adopt the new layout data. The method is shown below.

[0008] If the cost of the new placement data is lower than the cost of the conventional placement data, the new placement data is adopted. On the other hand, if the cost of the new placement data is higher than the cost of the conventional placement data, a random number between 0 and 1 is generated,
When the random number is greater than exp(-ΔC/T), new arrangement data is adopted. Here, △C
is the difference between the cost of the new layout data and the cost of the conventional layout data, and T is the temperature managed by the temperature management means 34.

If the new layout data is to be adopted, the layout data is updated in step 405; if not, the process jumps to step 406, leaving the existing layout data as is.

At step 406, the cost evaluation means 33 manages the number of times placement data has been generated, and when the number of times exceeds a set value, the temperature managed by the temperature management means 34 is lowered. Conversely, if the number of times is within the set value, the process returns to 402.

In step 407, the process termination determining means 35 is used to terminate the process if the final cost at each temperature is the same three times in a row. Otherwise, the process returns to 402.

0012

However, in the above method, since the arrangement conversion means are applied at random, there is a high possibility that the arrangement conversion means will be applied unnecessarily, and the processing time is increased. In view of the above problems, it is an object of the present invention to efficiently apply a layout conversion means to shorten processing time.

[0013]

[Means for Solving the Problems] In order to achieve this object, the present invention provides a conversion effectiveness determining means for determining the effectiveness of a layout conversion means, and an application ratio of a plurality of layout conversion means for determining the conversion effectiveness. and an application ratio adjusting means for adjusting based on the determination result of the means.

[0014]

[Operation] In the present invention, the application ratio adjusting means for managing the application ratio of the arrangement changing means adjusts the application ratio at the next temperature based on the data of the conversion effectiveness determining means for determining the effectiveness of the arrangement changing means for each temperature. decide. Based on this ratio, the highly effective placement conversion means is preferentially applied to the next temperature, thereby reducing the number of times the placement conversion means is applied in vain.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIG.

In FIG. 1, thin lines indicate control signals, double lines mainly indicate data reference relationships, and thick lines mainly indicate control signals and data flows. In FIG. 1, reference numeral 1 denotes a database storing a plurality of arrangement conversion means for generating arrangement data; 6, conversion effectiveness determination means for determining the effectiveness of the arrangement conversion means;
5 is an application ratio adjustment means for adjusting the application ratio of a plurality of arrangement conversion means based on the determination result of the conversion effectiveness determination means 6;
2 is a conversion application means that applies the placement conversion means selected from the database 1 based on the application ratio determined by the application ratio adjustment means 5, and calculates the cost of new placement data generated by the applied placement conversion means; 3 is a cost evaluation means that compares the cost of the conventional placement data and the cost of the new placement data to decide whether to adopt the old placement data or the new placement data, and 4 is a method for managing and changing a parameter called temperature. Reference numeral 7 indicates a temperature control means for observing the fluctuations in the arrangement data and terminates the processing when the variation disappears. Reference numeral 8 indicates an initialization means for setting and generating initial values.

In the configuration of FIG. 1 as described above, the following is performed.
The operation will be explained with reference to the flowchart shown in FIG.

First, in step 201, the initialization means 8 generates a combination of initial values, calculates the cost of the combination of initial values,
The parameter of temperature managed by temperature management means 4 and the parameter of application ratio managed by application ratio adjustment means 5 are initialized.

Next, in step 202, the conversion application means 2 selects one arrangement conversion means stored in the database 1 based on the magnitude of the application ratio managed by the application ratio adjustment means 5. The arrangement conversion means with a larger application ratio is selected preferentially.

At step 203, the arrangement conversion means selected at step 202 is similarly applied by the conversion application means 2 to generate new arrangement data. At 204, the cost evaluation means 3 calculates 20
Calculate the cost of the placement data generated in step 3.

In step 205, the cost of the new arrangement data calculated in step 204, the cost of the conventional arrangement data, and the temperature management means 4 are calculated.
Based on the temperature managed by the cost evaluation means 3, it is determined whether to adopt the new layout data. The method is shown below.

[0022] If the cost of the new placement data is lower than the cost of the conventional placement data, the new placement data is adopted. On the other hand, if the cost of the new placement data is higher than the cost of the conventional placement data, a random number between 0 and 1 is generated,
When the random number is greater than exp(-ΔC/T), new arrangement data is adopted. Here, △C
is the difference between the cost of the new placement data and the cost of the conventional placement data, and T is the temperature managed by the temperature management means 4.

If the new layout data is to be adopted, the layout data is updated in step 206; if the new layout data is not to be adopted, the process jumps to 207 with the conventional layout data unchanged.

At step 207, the cost evaluation means 3 manages the number of times placement data has been generated, and when the number of times exceeds a set value, the temperature managed by the temperature management means 4 is lowered. Conversely, if the number of times is within the set value, the process jumps to 202.

At step 208, the application ratio for each temperature is updated using the application ratio adjustment means 5 and the conversion effectiveness determination means 6.

The conversion effectiveness determining means 6 determines the effectiveness of each arrangement conversion means stored in the database 1 for each temperature. In this embodiment, it is determined that the effectiveness is higher as the ratio between the number of times each layout conversion means is applied and the number of times the cost is reduced by applying that layout conversion means is larger.

The application ratio adjustment means 5 first ranks the plurality of placement conversion means in descending order of effectiveness based on the results of the conversion effectiveness determination means 6. Next, it is divided into high ranking placement conversion means and low ranking placement conversion means. The application ratio is increased for a placement conversion means with a high rank, and the application ratio is decreased for a placement conversion means with a low rank. Further, the higher the rank, the higher the increase rate, and the lower the rank, the higher the decrease rate. Based on these rate of increase/decrease, determine the application ratio at the next temperature.

That is, the higher the effectiveness determined by the conversion effectiveness determining means, the higher the application ratio is made, and the lower the effectiveness, the lower the application ratio.

In this embodiment, the number of arrangement conversion means is N, the maximum increase/decrease rate width is W, and the priority order of the arrangement conversion means is i, (N-1
)/2 is defined as the number of divisions n, and the increase/decrease rate r(i) of the application ratio of each arrangement conversion means is determined as follows. If N is an odd number (N=2n+1), r(i) = +(W/n)×((n+1) − i)
i ≦ nr(i) = 0
i = n+1r(i)
= −(W/n)×(i − (n+1))
i > n+1 If N is an even number (N=2(n+1)), r(i) = +(W/n)×((n+1) − i
) i ≦ nr(i) = 0
i = n+1, n+
2r(i) = −(W/n)×(i − (n+2)
) i > n+2.

At step 209, the process termination determining means 7 is used to terminate the process if the final cost at each temperature is the same three times in a row. If not, the process returns to 202.

As described above, by providing the application ratio adjusting means 5 and the conversion effectiveness determining means 6 and adjusting the application ratio for each temperature, it becomes possible to effectively apply the arrangement conversion means.

[0032]

As described above, the present invention provides a conversion effectiveness determining means for determining the effectiveness of the relocation means for each temperature, and application of the relocation means at the following temperatures based on the data of the conversion effectiveness determining means. By providing the application ratio adjusting means for determining the ratio, the number of times the unnecessary arrangement conversion means is applied is reduced, and as a result, high-speed processing becomes possible.

[Brief explanation of drawings]

[Fig. 1] Configuration diagram in an embodiment of the present invention

[Figure 2] Flowchart in the example

[Figure 3] Block diagram of a conventional combinatorial optimization device

[Figure 4] Flowchart in a conventional combinatorial optimization device

[Explanation of symbols]

1 Database 2. Conversion application means 3 Cost evaluation means 4 Temperature control means 5 Applicable ratio adjustment means 6 Conversion effectiveness determination means 7 Processing end determination means 8 Initialization means

Claims (1)

[Claims] Claim 1: A plurality of arrangement conversion means for generating new arrangement data, a conversion effectiveness determination means for determining the effectiveness of each arrangement conversion means, and a ratio of applying the plurality of arrangement conversion means for determining the conversion effectiveness. Calculate the cost of new placement data generated by applying an application ratio adjustment means that adjusts based on the determination result of the means and one of a plurality of placement conversion means based on the application ratio determined by the application ratio adjustment means. A cost evaluation means that compares the cost of the old placement data and the cost of the new placement data to decide whether to adopt the old placement data or the new placement data, and a parameter called temperature. 1. A combinational optimization device comprising: a temperature management means for controlling the cost of the arrangement data; and a processing end determining means for terminating the processing when there is no longer any variation in the cost of placement data.