JPH05289867A - Multibranch branching control system - Google Patents

Abstract

PURPOSE:To set a processing time for selecting a branch destination to the same as the case when the number of branch destinations is small even in the case its number is large, in the multibranch branching control system of an information processor. CONSTITUTION:In a main storage device 2, a hash table 31 containing a key part 32 and an address part 33 is stored. In the key part 32, a value of a variable is stored in an entry shown by a hash function value corresponding to its value. In each entry of the address part 33, an address of a branch destination shown by the value of the variable stored in each entry of the corresponding key part 32 is stored. When the value of the variable for showing the branch destination at the time of execution is given, a calculating means 11 calculates the hash function value corresponding to the value of its variable, and a searching means 12 searches the hash table 31 based on a result of calculation, and derives an address of the branch destination. Accordingly, even in the case there are many branch destinations, the branch destination can be selected at a high speed.

Description

Detailed Description of the Invention

[0001]

The present invention relates to an information processing device,
The present invention relates to a multi-branch control method for selecting a branch destination based on the value of a given variable.

[0002]

2. Description of the Related Art A conventional multi-branch branch control system in an information processing apparatus is provided with data as a key for selecting a branch destination, for example, as shown in the flowchart of FIG. Data and branch destination No. 1-No. n data is sequentially compared (steps S71-1 to S71-
n), the branch destination No. where the data match. i is branched to an address corresponding to i (steps S72-1 to S7)
2-n).

In addition, as a multi-branch branch control method in a conventional information processing apparatus, in addition to this, a table including a key section 81 and an address section 82 shown in FIG. In some cases, the processing shown in the flowchart of FIG. 9 is performed by giving data as a key for selecting a branch destination.

That is, when compiling, as shown in FIG.
A key portion 81 having branch destination data key1 to keyn, and an address portion 82 having branch destination addresses A1 to An corresponding to the branch destination data keys1 to keyn.
Create a table consisting of and.

When the data key serving as a key for selecting the branch destination at the time of execution is given, the data key1 stored at the head of the key portion 81 of the table is given as shown in the flowchart of FIG. It is determined whether or not the received data key matches (steps S91 and S9).
2).

When it is determined that they match, the address A1 stored in the corresponding entry of the address section 82 of the table is set as the branch destination address (step S9).
3).

If it is determined that they do not match, it is determined whether or not the search for all entries in the key portion 81 has been completed (step S94).

If it is determined that the search for all the entries has been completed, the processing is terminated, and if it is determined that the search has not been completed, the entry to be processed in the table is set to the next entry (step S95), The same processing as described above is performed again.

[0009]

As described above, according to the conventional multi-branch branching control system, the branching is performed by sequentially comparing the key data and the branch destination data, or by sequentially searching the table. Since the destination is selected, there is a problem that the time required to select the destination increases in proportion to the number of destinations.

It is an object of the present invention to provide a multi-branch branch control system in which the processing time for selecting branch destinations does not increase even if the number of branch destinations increases.

[0011]

In order to achieve the above object, the present invention provides a multi-branch control method of an information processing apparatus which selects a branch destination according to a value of a given variable, and a hash at compile time. The value of the variable and the branch destination corresponding to the value of the variable are created using the function to create a hash table stored in the entry indicated by the hash function value, and at the time of execution, the value of the variable given using the hash function. Is calculated, and the branch destination is obtained from the hash table based on the calculation result.

[0012]

The hash table in which the value of the variable and the branch destination corresponding to the value of the variable are stored in the entry indicated by the hash function value is created at the time of compilation by using the hash function.

When the value of the variable is given at the time of execution, the hash function value is calculated, and the branch destination is obtained from the hash table based on the calculation result.

[0014]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention, which includes a central processing unit 1 and a main memory 2.

The central processing unit 1 includes a calculating means 11, a searching means 12 and a branching means 13. The calculating means 11 and the searching means 12 are realized by firmware of a micro program. By doing so, the processing time can be shortened as compared with the case of executing by a machine language program.

FIG. 2 is a diagram showing an example of a source program having a multi-branch branching control structure, and shows branching to different processing according to the value given to the variable key. That is, in FIG. 2, the values given to the variable key are 0, 1, 5, 1
In the case of 0, 100, 200, 1000, key = 0, k
ey = 1, key = 5, key = 10, key = 10
It indicates that the process branches to 0, key = 200, and key = 1000, and that if the value is other than the above, the process branches to “other”.

Next, the operation of this embodiment will be described.

At the time of compiling the source program shown in FIG. 2, an appropriate hash function hash (x) is used to generate the program shown in FIG.
The hash table 31 as shown in FIG.

The hash table 31 of FIG. 3 is a hash function ha.
sh (x) ≡x mod 6 (mod represents a remainder operation), and the key unit 32
And an address section 33.

In this case, hash (x) = x mod
6 and hash (0) = 0, hash (1) = 1,
Since hash (200) = 2 and hash (5) = 5, 0 is entered in the entry of index = 0 in the key part 32,
1 is stored in the entry of index = 1, 200 is stored in the entry of index = 2, and 5 is stored in the entry of index = 5.

Also, hash (10) = hash (10
0) = hash (1000) = 4, and when the value of the variable key is 10, 100, 1000, a collision occurs, so
Only 10 is stored in the entry of index = 4 in the key part 32, 100 is connected to the entry of index = 6 by the pointer to the entry of index = 4, and 1000 is connected to the entry of the index = 6 by the pointer. It is stored in the entry of index = 7.

In the address section 33, the index = 0 entry has the branch destination address corresponding to the key = 0 processing, and the index = 1 entry has the branch destination address corresponding to the key = 1 processing. , Index = 2, the branch destination address corresponding to the key = 200 process, and index = 4 entry, the key = 10.
The address of the branch destination corresponding to the processing of is index = 5
Entry has a branch destination address corresponding to the processing of key = 5, and an entry having index = 6 has key = 10.
The branch destination address corresponding to the processing of 0 is index =
The entry of 7 stores the address of the branch destination corresponding to the processing of key 1000.

The hash table 31 shown in FIG. 3 created at the time of compilation is stored in the main memory 2 at the time of execution as shown in FIG. 1, and the central processing unit 1 is given the value of the variable key at the time of execution. The process shown in the flowchart of FIG.

First, in the central processing unit 1, the calculation means 11 causes index = hash (key) = key mod.
Calculation 6 is performed (step S41).

Then, the search means 12 refers to the entry of the key part 32 of the hash table 31 corresponding to the above calculation result, and determines whether the value stored in the entry matches the value of the given variable key. It is determined whether or not (step S42).

When it is determined that they match, the searching means 12 passes the address stored in the entry of the address portion 33 corresponding to the entry of the key portion 32 to the branching means 13 and branches to that address (step S
43).

If it is determined in step S42 that they do not match, the searching means 12 searches the entry of the key portion 32 connected by the pointer to the entry referred to in step S42, and the value of the given variable key is stored. Search for the entry that has been made (step S44).

If there is an entry in the key section 32 that stores a value that matches the value of the given variable key (YES in step S45), the searching means 12 causes the entry of the key section 32 to be performed. The address stored in the entry of the address part 33 corresponding to
3 and branches to that address (step S4
6).

When it is determined that the data does not exist, the branching means 13 is notified of that fact, and the address is branched to the address of "other" processing (step S47).

Therefore, when the value of the given variable key is, for example, 200, the calculation result of step S41 is 2
Then, in step S42, the inde of the key part 32
The entry of x = 2 is referred to, and it is further determined whether or not the value stored in the entry and the value 2 of the given variable key match. In the case of the present embodiment, 200 is stored in the entry of index = 2 of the key part 32, and the determination result of step S42 is YES. Therefore, the process of step S43 is performed, and the in
key = 200 stored in the entry of dex = 2
A branch is made to an address corresponding to the processing.

The value of the given variable key is 0, 1, 1
In the case of 0 and 5, the same processing is performed and key = 0, k
A branch is made to an address corresponding to the processing of ey = 1, key = 10, and key = 5.

The value of the given variable key is 100.
Alternatively, in the case of 1000, that is, when the hash function value becomes 4 and the value of the variable key collides with the hash function value of 10, the following processing is performed.

The value of the given variable key is, for example, 100.
In the case of 0, the calculation result of step S41 is 4, the entry of index = 4 of the key part 32 is referred to in step S42, and the value stored in the entry and the value of the given variable key are 1000. It is determined whether and match. In the case of the present embodiment, in of the key unit 32
Since 10 is stored in the entry of dex = 4, and the determination result of step S42 is NO, step S4
In step 4, the pointer is traced to search the entry of the key part 32 in which the value 1000 is stored, and the entry of index = 7 is found. Thereafter, the process of step S46 is performed, and the address section 33
Key = stored in the index = 7 entry of
A branch is made to an address corresponding to 1000 processes.

When the value of the given variable key is 100, similar processing is performed, and branching is performed to the address corresponding to the processing of key = 100.

If the value of the given variable key is not stored in the key portion 32 of the hash table 31, the determination results of steps S42 and S45 are both NO, and the process of step S47 is performed.

By the way, although the hash function is not a perfect hash function in the above-described embodiment, the multi-branch branch control structure in the program is, as shown in FIG. 2, the number of branch destinations and the variable key corresponding to each branch destination. In many cases, the value of and is known, and in such a case, a perfect hash function can be selected as the hash function.

FIG. 5 shows a complete hash function hash (x) ≡x m when the source program shown in FIG. 2 is compiled.
The hash table 51 created using od 7 is shown.

As shown in the figure, the hash table 51 is composed of a key portion 52 and an address portion 53.

In this case, hash (x) = x mod
7 and hash (0) = 0, hash (1) = 1,
hash (100) = 2, hash (10) = 3, ha
sh (200) = 4, hash (5) = 5, hash
Since (1000) = 6, the entries of index = 0 to index = 6 of the key part 52 of the hash table 51 are 0, 1, 100, 10, 200, 5, 1000, respectively.
Is stored.

Further, the indexes 0 to i of the address section 53
In the entry of ndex6, the branch destination address corresponding to the processing of the variable key = 0, the variable key = 1, the variable key = 100, the variable key = 10, the variable key = 20.
The address of the branch destination corresponding to the processing of 0, variable key = 5, variable key = 1000 is stored.

Hash table 51 created at compile time
Are stored in the main memory 2 at the time of execution, and the central processing unit 1
When the value of the variable key is given at the time of execution, performs the processing shown in the flowchart of FIG.

When the value of the variable key is given, the central processing unit 1 causes the calculation means 11 to calculate the complete hash function hash.
Using (x) = x mod 7, index = hash
The calculation of (key) is performed (step S61).

Next, the central processing unit 1 uses the search means 12 to search the key portion 5 of the hash table 51 corresponding to the above calculation result.
The second entry is referred to, and it is determined whether the value stored in the entry and the value of the given variable key match (step S62). In the case of this embodiment, since the hash table 51 is created using the complete hash function,
When the value of the given variable key is the value of the variable key described in the source program of FIG. 2, the determination result of step S62 is always YES.

Then, the determination result of step S62 is YE.
In the case of S, that is, when the value of the supplied variable key is the value of the variable key described in the source program of FIG. 2, the search means 12 causes the address section 53 corresponding to the entry of the key section 52. The address stored in the entry is passed to the branching unit 13 and branched to that address.
(Step S63).

If the decision result in the step S62 is NO, the searching means 12 notifies the branching means 13 to that effect and branches it to the "other" processing (step S6).
4).

Therefore, as the value of the variable key, the values 0, 1, 100, 1 described in the source program of FIG.
When 0,200,5,1000 is given, the calculation results of step S61 are 0, 1, 2, 3, 4, respectively.
5 and 6, and in step S62, i of the key portion 52
It is determined that the values stored in the entries of ndex = 0, 1, 2, 3, 4, 5, 6 and the value of the given variable key match, and in step S63, the address part 53
Branch to the address stored in the entry of index = 0,1,2,3,4,5,6. As described above, by using the perfect hash function, the collision of the hash function values is eliminated, so that the selection processing at the time of branching can be simplified and the processing time can be shortened.

If a value not described in the source program of FIG. 2 is given as the value of the variable key, the result of the determination in step S62 is NO, and the process of step S64 is performed.

In addition, the source program of FIG.
If the ult processing is not described, steps S62 and S64 in the part surrounded by the broken line in FIG. 6 can be omitted.

[0050]

As described above, since the present invention uses the hashing method for the multi-branch branch control method of the information processing apparatus, the processing time for selecting the branch destinations depends on the number of branch destinations. Even if there are many branches, it can be the same as when there are few branches. Therefore, according to the present invention, even if there are many branch destinations, there is an effect that the branch destinations can be selected at high speed.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a diagram showing a part of a source program executed in the embodiment.

FIG. 3 is a diagram showing a hash table created using a hash function.

4 is a flowchart showing a processing example when selecting a branch destination using the hash table of FIG.

FIG. 5 is a diagram showing a hash table created by using a perfect hash function.

6 is a flowchart showing a processing example when selecting a branch destination using the hash table of FIG.

FIG. 7 is a flowchart for explaining a conventional technique.

FIG. 8 is a table for explaining another conventional technique.

9 is a flowchart showing a processing example when selecting a branch destination using the table of FIG.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Central processing unit 2 ... Main memory device 11 ... Calculation means 12 ... Search means 13 ... Branching means 31, 51 ... Hash table 32, 52, 81 ... Key part 33, 53, 82 ... Address part

Claims (3)

[Claims] 1. In a multi-branch branch control method of an information processing apparatus for selecting a branch destination according to a value of a given variable, at the time of compilation, a hash function is used to correspond to the value of the variable and the value of the variable. Creates a hash table whose branch destination is stored in the entry indicated by the hash function value, calculates the hash function value corresponding to the value of the given variable using the hash function at runtime, and based on the calculation result A multi-branch branch control method, wherein a branch destination is obtained from the hash table. 2. The multi-branch control method according to claim 1, wherein the hash function is a residue function. 3. The multi-branch control method according to claim 1, wherein the hash function is a perfect hash function.