JPS60242731A - Digital pattern compressing method - Google Patents

Abstract

PURPOSE:To attain ease of compression by offsetting sequentially a digital pattern at each word, comparing the said pattern with the pattern not offset so as to replace repetitively the pattern into a prescribed word. CONSTITUTION:Suppose that the compressed digital pattern P(0) is badbdbcbbbc bdbcbdbca. At first, a repetitive word is extracted from the pattern. Then when the pattern is extracted, the step goes to the step 38, and when it is not extracted, the step goes to the step 40. In the step 38, the word kind, word position and repetitive number of times are arranged, a repetitive word is replaced into a prescribed word 0 and the pattern is compressed into a pattern P(1) of badbdbc0cbdbcbdbca. Then the repetition of the pattern being consecutive word combination is extracted from the patterf P(1). When the corresponding part exists, the step goes to the step 44 and when not existing, the compression is finished. The step 44 arranges the corresponding part, and the pattern having much compression is replaced into a prescribed word with priority. The operation is repeated similarly.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a digital
The present invention relates to a digital pattern compression method for compressing patterns in the length and depth directions.

[Background of the invention]

Most of the digital patterns for IC testers used to test digital integrated circuits (ICs) such as gate arrays are created using CAD (Computer Aided Design).
This is the output of gn). However, the digital pattern created by this CAD is extremely long, and generating this pattern using a sequential digital pattern generator requires a large capacity memory. Note that a sequential digital pattern generator stores generated digital patterns in a memory in the order of addresses, and reads out the memory in the order of the addresses.

On the other hand, there is a microprogram type digital pattern generator, and this type is rJumpJ.

You can save memory capacity by using instructions such as ``5ubroutine Ca1l J.'' Also, you can save memory capacity by using instructions such as ``5ubroutine Ca1l J.'' You can also create repeated words (successive same words) or repeated pattern parts (repetition of the same pattern part that is a combination of words) from a series of digital words. The length of the digital pattern can be compressed by extracting the
It is possible to return to the original pattern using commands such as umpJ and rsubroutine Ca1lJ. Therefore, long digital patterns created by CAD can also be compressed and generated by a microprogrammed digital pattern generator.

Conventionally, compressing digital patterns relied on the intuition and experience of the operator, which was inefficient;
This was almost impossible if the pattern was very long.

[Purpose of the invention]

Therefore, an object of the present invention is to provide a digital pattern compression method that can easily compress the length of a digital pattern.

[Summary of the invention]

According to the digital pattern compression method of the present invention, a digital pattern consisting of a plurality of digital words is sequentially offset word by word, and the offset digital pattern and the unoffset digital pattern are compared. In each comparison, if the digital words at corresponding positions in the offset digital fist pattern and the non-offset digital pattern are consecutively equal, then the pattern portion consisting of these digital words is a repeating pattern portion. The repetitive pattern portion extracted in this way is replaced with a predetermined word to compress the digital pattern.

[Embodiments of the invention]

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. FIG. 1 is a flow diagram illustrating a preferred embodiment of the invention, and FIG. 2 is a block diagram of a system utilizing the invention.

(System Utilizing the Present Invention) The system shown in FIG. 2 is composed of a computer 10 and a microprogram type digital number pattern generator 12. The computer 10 includes a central processing unit (C
PU) 14, a read-only memory (ROM) 16 that stores programs for this CPU 14, and a random access memory (RAM) 18 as a temporary storage device.
and a keyboard 20 as an input device.

These blocks 14 to 20 are interconnected via a bus 22. On the other hand, digital pattern generator 1
At 2, memory 24 stores execution control instructions (microcode) and digital words (vectors), and decoder/counter 26 decodes the execution control instructions from this memory and counts the number of repetitions. Multiplexer 30 selects the destination address from memory 24 or stack memory 28 depending on the output of decoder/counter 26.
Select the return address from. Program counter 32 loads the output of multiplexer 30 in response to the output of decoder/counter 26, or simply counts clocks and provides its output as an address signal to memory 24 and stack memory 28.

The computer IO may create a digital pattern using a CAD program, or may receive a digital pattern created by another computer or the like. Then, this computer 10 compresses the created digital working pattern using a pattern compression program and loads it into the memory 24.

This pattern compression will be explained next.

(Pattern Compression) FIG. 1 is a flowchart illustrating the digital pattern compression method of the present invention. The operation described below is performed by using the RAM 18 according to the program stored in the ROM 16.
U14 will do it. It is assumed that the digital pattern to be compressed is the following pattern p (o).

badbdbcbbbcbdbcbdbca where letters a-d are any parallel bit words, e.g. rloll
J, rlloOJ, etc.

At step 34, repeating words are extracted from the digital pattern. That is, each word of this pattern is sequentially searched from the beginning, and consecutive parts of the same word are extracted.

In the above pattern P(0), word B is repeated three times between the 8th and 10th. In step 36, it is determined whether or not a repeated word has been extracted. If a repeated word has been extracted, the process proceeds to step 38; if not (there is no repeated word), the process proceeds to step 40. Step 38
Now, the extracted repeated words are sorted, that is, the word type, word position, and number of repetitions are sorted out, and each repeated word is replaced with a predetermined word.

For example, in the above pattern P(0), the three words b from the 8th to the 10th are replaced with word 0, and compressed into the following digital pattern P(1).

badbdbcOcbdbcbdbca Therefore, steps 34-38 perform pattern compression based on repeated words. Note that depending on the capability of the digital pattern generator 12 used, it is necessary to consider how many repetitions of words should be replaced. For example, it takes one clock cycle to load the count into the counter section of block 26, one clock cycle to decrement this counter section, and furthermore, it takes zero judgment of this counter section and rJumpJ (address If it takes one clock cycle for an instruction to skip rL
At least three clock cycles are required to replace OopJ (an instruction that repeats the same word a set number of times). Therefore, in this case, unless words that are repeated four or more times are replaced, there is no compression effect due to replacement.

Next, in step 40, the (
A repeating pattern portion, that is, a repetition of a pattern portion that is a combination of consecutive words, is extracted from the (compressed) digital pattern. This stesoge 40 will be described in detail later, but in the case of the above-mentioned pattern P(1), it extracts the third to fourth and fifth to sixth pattern parts bd, and
In step 42, which extracts the 4th to 7th, 100th to 1st, and 144th to 1st pattern parts bdbc, it is determined whether or not a repeating pattern part has been extracted, and it is determined whether a repeating pattern part exists. If so, step 4
If not, the pattern compression ends. In step 44, as will be described in detail later, the extracted repetitive pattern portions are sorted, and the pattern portions with the highest amount of compression are preferentially replaced with predetermined words. In the case of pattern P(1), the fourth to sixth words are pattern part b
d and bdbc, but the pattern part bdbcO has a larger amount of compression, so it is given priority and replaced with word A. Therefore, the compressed digital pattern P(2)
becomes as follows.

badAOcAAa Thus, in steps 40 to 44, pattern compression is performed based on the repeated pattern portion.

When step 44 is completed, the process returns to step 34 and step 44 is completed.

In steps 34 to 38, pattern compression is performed again based on the repeated words of pattern P (2).

In pattern P(2), the 7th and 8th words are both word A, so two repetitions of word A are word 1.
, it is compressed into the following pattern P(3).

badAOcla This pattern P(3) does not include a repeating pattern portion, so this is the final result of compression.

(Extraction of Repeated Pattern Portion) Next, step 40 of extracting the repeated pattern portion will be described in detail with reference to FIGS. 3 and 4. FIG. 3 is a detailed flowchart of step 40, and FIG. 4 is a table showing the operation according to this flowchart, taking the above-mentioned digital pattern P(1) as an example. In step 46,
Offcent the pattern by one word. The first offset results in column 61 of the table in FIG. Note that in each column of this table, the left side is a pattern when there is no offset, and the right side is a pattern when there is an offset. Further, numbers 1 to 16 indicating columns indicate the number of offsets, and numbers 1 to 18 indicating rows indicate the order of words in a pattern without offset. Step 48
Then, the offset pattern and the non-offset pattern are compared for each corresponding position, that is, for each line, and it is determined whether or not the words match consecutively. In the first column, which is the first offset, consecutive words do not match, but if they do match, a list of matches is created. In step 50, it is determined whether or not the offset has been performed the required number of times.If the answer is yes, the extraction of the repeated pattern part is completed and the process goes to step 42, but if the answer is negative, the process goes to step 48. In the case of offset, the 5th and 6th words of the non-offset pattern match the 3rd and 4th words of the offset pattern, respectively, as shown in the second column of FIG. That is, the words match consecutively. In this case, a list of 5-6:3-4 (db) is created. This list is stored in RAM 18. Similarly, at the 4th offset, 13-17: 9-13 (cbdbc) at the 6th offset, 10-13: 4-7 (bdbc) at the 8th offset, 11-12: 3-4 ( db) At the 10th offset, create a list of 14-17: 4-7 (bdbc) and at the 12th offset, 15-16: 3-4 (ab). In the case of Nono turn P(1), the 16th offset causes the non-offset pattern and the offset pattern to overlap by only 2 words, so the 16th offset is the final offset. (Even if the offset is performed for the 17th time, since only one word overlaps, the words will not match consecutively.) Therefore, the extraction of the repetitive pattern portion is completed.

(Organization and Replacement of Repeated Pattern Portion) Next, the arrangement and replacement of the repetitive pattern portion of Step 4 will be explained with reference to the flowchart of FIG. -! First, preprocessing is performed in step 52, and it is checked whether there are any words in the list created in step 48 whose left and right sides overlap.

In the list based on the fourth offset, the 13th word overlaps the left and right sides, so 13-16:
9-12 (cbdb) and 14-17: 1O-13
(bd bc) to determine which one should be adopted. In this example, since there is bdbc in another list, the fourth offset is 14-17:10-13 (bdbc). Then, the data is sorted into pattern parts having the same word structure, and the maximum compression amount α (MAX) is set to zero.

In step 54, the compression amount α of the first pattern portion is
is calculated using the following formula.

α=VXC-(C+V-1) Here, V is the number of words in the pattern portion, and C is the number of the same pattern portions. Here, the first pattern part is 3-4.
5-6. If the db is 11-'12.15-16, then α=2X4-(2+4-1)=3. In step 56, it is determined whether α(MAX) is larger than α. If affirmative, the process proceeds to step 58; if negative, the process proceeds to step 54. In the case of the example, α(MAX)=
Q, α=3, so proceed to step 58 and calculate α(MAX
)-α, that is, α(MAX), is set to 3. In step 60, it is determined whether calculations have been performed for all pattern parts. If affirmative, the process proceeds to step 62; if negative, the process returns to step 54. Since the pattern portion of bdbc has not been calculated yet, α of this pattern portion is calculated in step 54.

bdbc is 4-7.10-13.14-17, so α=
4 x 3-(4+3-1)=6. Therefore, by steps 56 and 58, the new α(MAX) is bdb
It becomes 6 of c. Since all pattern parts have been calculated, the process proceeds to step 62, where the pattern part of α(MAX) is replaced with a predetermined word. In the case of pattern P(1), if the pattern portion bdbc is replaced with word A, the compressed pattern P(2) becomes badAOcAAa, and the process proceeds to step 34. Note that even if the pattern portions extracted in step 40 overlap each other, 1. The process of steps 52 to 62 eliminates the problem.

In the digital pattern P(2), the 7th and 8th words repeat A, so by steps 34 to 38,
The final compressed digital pattern P(3) becomes badAOcla. Here, 0 = b b b, A = b
d b c , 1 = AA. Microcode (execution control instructions, e.g. rAdv) is added to these patterns.
anceJ, rcall 5ubroutineJ
.

rJumpJ, etc.) and store it in the memory 24.

(Changes to Embodiments) Although only the preferred embodiments of the present invention have been described above,
Various modifications are possible. For example, in Figure 5, the digital
If there is only one type of pattern portion, only step 62 needs to be executed. Further, the present invention is useful not only when applying a long digital pattern to a microprogram type digital pattern generator, but also to reducing the amount of digital data transmitted.

〔Effect of the invention〕

As described above, according to the present invention, digital patterns can be easily and efficiently compressed sequentially starting from the pattern portion that requires the largest amount of compression.

[Brief explanation of drawings]

1, 3 and 5 are flowcharts illustrating a preferred embodiment of the invention; FIG. 2 is a block diagram of a system utilizing the invention; and FIG. 4 is a table illustrating the invention. It is. Figure 1. Figure 3.

Claims (1)

[Claims] A digital node consisting of a plurality of digital integrated circuits is sequentially offset word by word, the offset digital pattern and the non-offset digital pattern are sequentially compared to extract a repetitive pattern portion, and the extracted repeat pattern is extracted. A digital pattern compression method characterized by replacing a pattern portion with a predetermined word.