JP6073761B2 - Search device and search method - Google Patents

Description

  The present invention relates to a search apparatus and method for searching for conditions that match input data.

  Many communication terminals are equipped with a search device to perform processing such as filtering and routing on received packets. This search device is a device that takes a part or all of a received packet as input (input data) and searches for a condition that matches the input data from a pre-registered condition table. The search device outputs a condition number when a matching condition is found, and outputs a search result with no matching condition (non-matching) when there is no matching condition. Then, the communication terminal discards the received packet or determines the output port according to the search result output by the search device.

  With the spread of broadband, communication terminals are required to have high throughput. Therefore, the search apparatus is also required to increase the throughput, that is, to increase the number of input data that can be searched per unit time. Furthermore, it is necessary to reduce energy consumption from the viewpoint of energy saving.

  There is a linear search method as a method for the search device to search for a condition that matches the input data. This method is a method in which the input data is compared with each condition one by one to find a condition that matches the input data (match search). FIG. 14 shows a configuration example of a search apparatus that implements this method. The search device 10 includes a condition memory 40 that stores a condition table, and a comparator 50 that compares input data with a condition. In this configuration, when input data is input at the start, the comparator transmits a read address to the condition memory, reads the conditions stored in the condition memory one by one, and compares them with the input data. When all the conditions and the input data are compared, the result is output to the outside as a search result.

  In the above method, since the search result can be output only after comparing the input data with all the conditions, it takes time to process one input data. Further, since it is necessary to access the memory as many times as the condition number, there is a problem that a large amount of power is consumed.

  In order to solve this problem, Patent Document 1 proposes a search device using a mismatch detection means. In the search device of Patent Document 1, the search device is provided with a mismatch detection means capable of detecting a condition that does not match input data in a short time, so that the number of processes for reading and comparing the condition that does not match the input data from the condition memory can be increased. This can reduce the average search time of the search device.

  FIG. 15 is a block diagram of the search device of Patent Document 1. The search device 10 includes a match search unit 30 that searches for a condition that matches input data by a method similar to the linear search method of FIG. 14, and a mismatch detection unit 20 that can detect a condition that does not match input data in a short time. Composed. In the search device 10, the registered conditions are divided into a plurality of groups. For example, if four types of conditions (condition # 0: 0000, condition # 1: 0100, condition # 2: 1000, condition # 3: 1011) are registered in the condition table, the condition # 0 and the condition # 1 are grouped. # 0 and condition # 2 and condition # 3 are group # 1.

  The input data input to the search device 10 is first input to the mismatch detection means 20. The mismatch search means 20 detects a group that does not have a condition that matches the input data (mismatch) in a short time. Then, the mismatch detection unit 20 outputs the group in which the mismatch is detected to the match search unit 30 as a mismatch detection result. Since it is not necessary to search the group in which the mismatch is detected by the mismatch detection unit 20 by the match search unit 30, the number of processes in which the condition search unit 30 reads the condition from the memory and compares it with the input data can be reduced. The average search time can be shortened.

  The mismatch detection means of Patent Document 1 detects a mismatch for each group using a condition memory. Conditions are registered in this condition memory for each group. A method for registering the conditions in the memory will be described with reference to FIG. Each bit of the condition memory corresponds to each group. For example, bit [0] in the condition memory corresponds to group # 0, and bit [1] corresponds to group # 1. Each address stores the presence / absence of a value in the condition table for each group. In an address equal to the value in the condition table, “Yes” is stored in the group to which the condition belongs, and in other areas. , “Nothing” is stored.

  For example, according to the condition # 0 (the value is 0000 and belongs to the group # 0), a symbol “present” is stored in bit [0] of the address 0000 of the condition memory. Similarly, the condition # 1 (the value is 0100) Bit [0] of condition memory address 0100 due to group # 0), bit [1] of condition memory address 1000 due to condition # 2 (value belonging to group # 1 with condition 1000), condition # 3 (value Is 1011 and belongs to group # 1), a “present” symbol is stored in bit [1] of address 1011 of the condition memory. In the other areas, the symbol “Nothing” is stored.

  In the case of detecting a mismatch for each group using this memory, an address equal to the input data may be read as shown in FIG. For example, when the input data is 0100, the address 0100 of the condition memory is read. Since the “Yes” condition is stored in bit [0] of address 0100 and the “No” condition is stored in bit [1], a mismatch detection result is output to group # 1. When a mismatch is detected in group # 1, it is not necessary to compare the input data with the conditions of group 1 in the subsequent matching search means.

  In this way, since it is possible to detect data mismatch for each group simply by reading an address equal to the input data, it is not necessary to perform a match search for the group in which the mismatch was detected in the match search process, and the number of processes to be compared with the input data Can be reduced.

Although the above configuration can detect a mismatch for each group by one memory access, the required memory capacity increases when the numerical range of the matching condition with the input data becomes large. For example, when the matching condition with the input data is a numerical range of m bits, the required memory capacity is 2 ^m bits. For this reason, in Patent Document 1, m bits are divided into blocks each of n bits, and a means for detecting a mismatch in parallel is taken to reduce the memory capacity.

  FIG. 18 is a diagram for explaining a condition table and a condition memory when a search condition table in a conventional search device is divided into two blocks. In FIG. 18, the condition table illustrated in FIG. 16 is divided into two blocks, block # 0 and block # 1, and a condition memory is prepared for each block. In each divided block, the condition table is divided into two groups. In the address of the condition memory equal to the value of each condition, “present” is stored in the group to which the condition belongs, and in other areas, “Nothing” is stored.

  For example, in the case of the upper 2-bit block # 1, the symbol “Yes” is stored in bit [0] of the address 00 of the condition memory # 2 due to the condition # 0 (value is 00 and belongs to the group # 0). Similarly, according to condition # 1 (value is 01 and belongs to group # 0), condition [bit] of address 01 in condition memory # 2 and condition # 2 (value is 10 and belongs to group # 1) A bit “1” is stored in bit [1] of address 10 of condition memory # 2 due to bit [1] of address 10 of memory # 2 and condition # 3 (value is 10 and belongs to group # 1). . In the other areas, the symbol “Nothing” is stored. Similarly, the presence or absence of a value in the condition table is also stored in the condition memory # 1 for the lower 2-bit block # 0.

  FIG. 19 is a diagram illustrating an example of a mismatch determination process in the condition memory of FIG. Since the condition memory is divided into two blocks, the input data 0100 is also divided into two blocks and mismatch detection is performed. For example, it is possible to detect the mismatch of the group # 1 by reading the value of the address 01 having the same value as the 2 bits 01.

JP2013-3644A

  As described above, in the search device using the mismatch detection means of Patent Document 1, it is possible to achieve high throughput and low power consumption in “complete match search” in which input data and conditions completely match. . However, the discrepancy presentation means of Patent Document 1 cannot cope with a Don't Care symbol whose value may be 0 or 1, and therefore cannot be applied to a Prefix search used in a router or the like. .

  The problem to be solved by the present invention is to provide a search device that includes mismatch detection means that can handle Don't Care and can be applied to Prefix search.

  In order to solve the above-described problems, a search device according to the present invention is a search device that searches for a condition that matches data input from a plurality of pre-registered conditions, and each of the conditions is A non-coincidence detection unit that belongs to one or a plurality of groups and includes a memory that stores the condition, and detects that the input data does not match the condition for each group; and the input data; A match search means for checking the condition and matching the input data with the condition, and the condition is expanded when the Don't Care symbol included in the condition is 0 and 1; Each address of the memory stores, for each group, a symbol indicating whether or not some or all of the conditions are equal to the numerical value of each address. The detecting means detects a mismatch for each group by reading a symbol for each group from the memory by using a part or all of the inputted numerical value as a read address of the memory, and a result of the detection. Means for notifying the match search means, wherein the match search means does not perform collation between the input data and the condition of the group in which the mismatch is detected.

  In addition, the non-coincidence detecting unit includes a dividing unit that divides the input numerical value into partial input numerical values composed of a plurality of blocks smaller than the number of digits or the bit number, and the partial input numerical value as a read address of the memory. A unit for detecting a mismatch for each group by reading the symbol for each group from the memory for each block, and a totaling unit for counting the detection result of the mismatch for each group for each block. Also good.

  The conditions may be grouped based on at least one of a Prefix value, the number of Don't Care symbols, and the position of the Don't Care symbols.

  Further, at least one of the conditions may belong to only one group.

  In order to solve the above-described problems, a search method according to the present invention is a search method for searching for a condition that matches data input from a plurality of previously registered conditions, and each of the conditions is A mismatch detection step for detecting for each group that the input data belongs to one or a plurality of groups and the input data does not match the condition stored in a memory; and the input data and the condition A matching search step for matching and searching for a condition that matches the input data. The condition is expanded when the Don't Care symbol included in the condition is 0 and 1, and Each address stores, for each group, a symbol indicating whether a part or all of the conditions are equal to the numerical value of each address. The step detects a mismatch for each group by reading a symbol for each group from the memory by using a part or all of the input numerical value as a read address of the memory, And a step of notifying the match search step, wherein the match search step does not perform collation between the input data and the condition of the group in which the mismatch is detected.

  The mismatch detection step includes the step of dividing the input numerical value into partial input numerical values composed of a plurality of blocks smaller than the number of digits or bits, and the partial input numerical value as a read address of the memory. You may make it provide the step which detects the mismatch with respect to each said group for every said block by reading the symbol for every said group from memory, and the step which totals the detection result of the mismatch with respect to each group for every said block.

  The conditions may be grouped based on at least one of the Prefix value, the number of Don't Care symbols, and the position of the Don't Care symbols.

  At least one of the conditions may belong to only one group.

  According to the present invention, in the detection apparatus including the mismatch detection means, it is possible to increase the mismatch detection result even when the search condition includes a Prefix value or a Don't Care symbol, and the input data in the match search means Thus, the number of comparisons of the conditions can be reduced, the throughput of the search device can be improved, and further the power consumption can be reduced.

FIG. 1 is a diagram showing an example of a condition table in the condition memory according to the embodiment of the present invention. FIG. 2 is a diagram for explaining the masking process when the Prefix value is set according to the embodiment of the present invention. FIG. 3 is a diagram for explaining the condition table when the Don't Care symbol of the condition table according to the embodiment of the present invention is expanded. FIG. 4 is a diagram for explaining a condition memory in which a condition table in which a Don't Care symbol is expanded is registered according to the embodiment of the present invention. FIG. 5 is a block diagram of the mismatch detection means according to the embodiment of the present invention. FIG. 6 is a diagram for explaining the condition table and the condition memory when the condition table according to the embodiment of the present invention is divided into two blocks. FIG. 7 is a block diagram of the mismatch detection means when the condition table according to the embodiment of the present invention is divided into two blocks. FIG. 8 is a diagram for explaining the condition table and the condition memory when the grouping by the Prefix value is not performed. FIG. 9 is a diagram for explaining a condition table and a condition memory when grouping is performed according to the Prefix value according to the embodiment of the present invention. FIG. 10 is a diagram illustrating a condition table in the case where grouping based on the position of the Don't Care symbol is not performed. FIG. 11 is a diagram for explaining a condition table and a condition memory when grouping based on the position of the Don't Care symbol according to the embodiment of the present invention is not performed and the condition table is divided into two blocks. FIG. 12 is a diagram illustrating a condition table when grouping is performed based on the position of the Don't Care symbol. FIG. 13 is a diagram for explaining a condition table and a condition memory when grouping is performed according to the position of the Don't Care symbol according to the embodiment of the present invention and the condition table is divided into two blocks. FIG. 14 is a block diagram of a conventional search device. FIG. 15 is a block diagram of a search apparatus provided with a conventional mismatch detection means. FIG. 16 is a diagram illustrating an example of a condition table and a condition memory in a conventional search device. FIG. 17 is a diagram showing an example of the mismatch determination process in the mismatch search means in the conventional search device. FIG. 18 is a diagram for explaining a condition table and a condition memory when the condition table in the conventional search device is divided into two blocks. FIG. 19 is a diagram showing an example of the mismatch determination process in the condition memory when the condition table in the conventional search device is divided into two blocks.

(First embodiment)
A first embodiment of the present invention will be described. In this embodiment, when registering a condition in the memory of the mismatch detection means, one condition including Don't Care is set to a plurality of conditions such as when the Don't Care symbol is 1 and 0. All the conditions not including the expanded Don't Care symbol are registered in the memory.

  First, Prefix search used in a router or the like will be described with reference to FIG. In the Prefix search, the condition table column includes a condition, a Prefix, and a priority. Prefix indicates how many bits from the most significant bit are not set in Don't Care (matching target) in collating the input data with the condition.

  For example, as shown in FIG. 2, when the condition is 0000 and the Prefix is 1, when input data having a value of 0010 is input, the input data and the lower 3 bits of the condition are treated as Don't Care, and are used as targets of collation. Must not. In this case, only the most significant bits of the input data 0010 and the condition 0000 are to be collated, and both the most significant bits are 0, so they match. Further, when a plurality of conditions match the input data, the priority is referred to, and it is determined that the condition matches a higher priority.

  An example of registering Prefix search conditions in the mismatch detection means will be described with reference to FIG. In this figure, condition # 0 is condition value 0000 and Prefix is 3, condition # 1 is condition value 0100 and Prefix is 4, condition # 2 is condition value 1000, Prefix is 4, condition # 3 is condition value In 1011, Prefix is set to 4. Furthermore, it is assumed that conditions # 0 and # 1 belong to group # 0, and conditions # 2 and # 3 belong to group # 1.

  First, Don't Care is set in accordance with the Prefix value of each condition (Prefix expansion). In this figure, since the Prefix for condition # 0 is 3, condition # 0 is 000 *. Next, the condition is expanded when Don't Care (*) of each condition is 1 and 0 (development of Don't Care).

  In the example of this figure, since only condition # 0 includes Don't Care, only condition # 0 is expanded. Since condition # 0 is 000 *, this condition # 0 is expanded into two conditions, 0000 and 0001. That is, it is the same as having five conditions in the condition table.

  Finally, as shown in FIG. 4, the five conditions for the Prefix search may be registered in the memory of the mismatch detection means as in the conventional case. Here, the two conditions, 0000 and 0001, which expanded Don't Care belong to the group # 0 to which the condition # 0 belonged. As described above, when the condition table includes Don't Care, it is not necessary to consider the Prefix value and Don't Care by expanding Don't Care and registering it in the memory. The operation when the input data is input can detect the mismatch as the operation of the mismatch detection means in the conventional complete match search.

  FIG. 5 shows an example of the configuration of the mismatch detection means that performs the above processing. The mismatch detection means 20 stores a delay means 21 for delaying input data and input it to the match search means, and a condition table. When the input data is input to the condition memory 22 as a read address, a mismatch detection result for each group is output. The overall configuration of the detection apparatus is the same as that shown in FIG.

  As shown in FIG. 6, even when Don't Care is expanded, it is possible to divide the condition table into a plurality of blocks each having n bits and detect inconsistencies in parallel for each block. . The configuration of the mismatch detection means in this case is shown in FIG. The mismatch detection unit 20 includes a data division unit 24 that divides input data into a plurality of blocks, condition memories 22 and 23 that output a mismatch detection result for each block, and a tabulation unit 25 that tabulates the mismatch detection results for each block. . The mismatch detection means 20 aggregates the mismatch detection results for each group and outputs them to the match search means.

  In this embodiment, four condition values are searched for in the condition table to be searched. However, the present invention can be applied to other condition values regardless of the input data, the numerical range of each condition, and the number of registered conditions. Needless to say. The grouping of conditions is not limited to two groups, and may be appropriately selected according to the numerical range of each condition and the number of registered conditions. For example, the present invention can be applied even when the number of groups is one. Moreover, it is not necessary to make the condition number of each group equal.

  In the present embodiment, an example applied to the Prefix search method in which the number of bits that do not set Don't Care is set from the upper bits according to the Prefix value is shown, but Don't Care seems to be set at random. It is applicable to any case.

  In the present embodiment, the match determination unit has a configuration in which a linear search method is implemented. However, the configuration is not limited to this, and for example, a binary search tree method, a B-tree search, and a CAM (Content addressing memory) may be used.

  In the present embodiment, an example in which the search device outputs a search result has been shown, but data (related data) to be output when each condition matches the input data is stored in the memory of the match search means, Instead of the determination result, the related data may be output. For example, related data can be output ports or output queues.

  In the present embodiment, an example in which the present invention is implemented in hardware has been described, but it goes without saying that the present invention may be implemented in software described in a program language.

(Second Embodiment)
A second embodiment of the present invention will be described with reference to FIGS. As in the first embodiment, this embodiment also expands the Don't Care symbol, but differs from the first embodiment in that conditions are grouped based on Prefix.

  In the first embodiment, the group to which each condition belongs is determined without considering the Prefix and the Don't Care symbol included in the condition. However, in this method, depending on the Prefix value of the condition belonging to each group, the probability that the mismatch detection unit can detect a mismatch may decrease, and the number of conditions searched by the match determination unit may not be reduced.

  For example, as shown in FIG. 8, condition # 0 is condition value 0000 and Prefix is 0, condition # 1 is condition value 0100 and Prefix is 4, condition # 2 is condition value 1000 and Prefix is 0, condition # 3 has a condition value of 1011 and a Prefix of 4. In this case, when Prefix is expanded, the conditions # 0 and # 2 become ***.

  When Don't Care of **** is expanded, all numerical values in the range of 0000 to 1111 can be obtained, and therefore, a “present” symbol is written in all the memories that detect mismatches as shown in FIG. Even if input data is input to the mismatch detection means, it is not possible to detect mismatches for all groups, and it is necessary to compare the input data with all conditions in the subsequent match detection means.

  In order to solve this problem, in this embodiment, as shown in FIG. 9, grouping is performed according to the value of Prefix. For example, in the case of the four types of conditions shown in FIG. 8, conditions # 0 and # 2 with a Prefix value of 0 are group # 0, and conditions # 1 and # 3 with a Prefix value of 4 are group # 1. . By performing grouping in this way, as shown in FIG. 9, the condition “***” exists only in group # 0, and “none” is present in bit [1] of the memory that detects the mismatch. The symbol can be present.

  In this way, by performing grouping according to the value of Prefix, it becomes possible to detect mismatches for group # 1, and it is possible to reduce the number of comparisons between input data and conditions in the match search means. . As a grouping method, if the grouping is performed so that the values of the respective conditions exist only in the same group as much as possible, it is possible to increase the mismatch detection probability for a specific group as described above. Further, if the grouping is performed so that the value of each condition exists in only one group, the detection probability of mismatch can be further increased.

(Third embodiment)
In the second embodiment, an example in which condition grouping is performed according to the value of Prefix has been described. However, condition grouping may be performed according to the number and position of Don't Care.

  That is, in the mismatch detection unit, it is possible to reduce the operation of comparing the input data and the condition in the match search unit by grouping the condition table so that the probability that a group that does not match can be detected becomes high. is there. When performing grouping according to the number of Don't Care, for example, by making the condition having the same number of Don't Care symbols the same group, the probability that a mismatching group can be detected can be increased. .

  A case where conditions are grouped according to the position of Don't Care will be described with reference to FIGS.

  FIG. 10 is a configuration example of a condition table when Don't Care is set at random. As shown in FIG. 10, the condition # 0 has a condition value of 000 *, and the least significant bit is set to Don't Care. Similarly, in condition # 1, the most significant bit is set to Don't Care, in condition # 2, the least significant bit is set to Don't Care, and in condition # 3, the most significant bit is set to Don't Care.

  In this condition table, when the Don't Care symbol of each condition is expanded, it is expanded into eight conditions as shown in FIG.

  In the present embodiment, a case will be described in which this condition table is divided into two blocks and a mismatch is detected in parallel for each block. FIG. 11 is a configuration example of a memory when the condition table of FIG. 10 is divided into two blocks and mismatch detection is performed.

  In this example, in each address area of the upper 2-bit block # 1 and the lower 2-bit block # 0, there are two addresses each storing “Yes” in both the group # 1 and the group # 0. For this reason, the probability that mismatch detection can be performed becomes low.

  On the other hand, as shown in FIG. 12, the case where the grouping is performed so that the condition # 0 and the condition # 2, and the condition # 1 and the condition # 3 in the same Don't Care position are in the same group, respectively. explain. FIG. 13 is a configuration example of a memory when the condition table of FIG. 12 is divided into two blocks and mismatch detection is performed.

  When the memory shown in FIG. 13 is used, in each address area of the upper 2-bit block # 1 and the lower 2-bit block # 0, an address where “present” is stored in both the group # 1 and the group # 0. Each has one address, and the probability of detection of a mismatched group can be increased compared to the case of FIG.

  Thus, by performing grouping according to the position of the Don't Care symbol, it is possible to increase the detection probability of a group that does not match, and to reduce the number of comparisons of input data and conditions in the match search means. Will be able to. As a grouping method, if the grouping is performed so that the values of the respective conditions exist only in the same group as much as possible, it is possible to increase the mismatch detection probability for a specific group as described above. Further, if the grouping is performed so that the value of each condition exists in only one group, the detection probability of mismatch can be further increased.

  In the third embodiment, an example in which the number of Don't Care is one has been described. However, the present invention is not limited to this, and the condition table is appropriately set according to the value of Prefix and the number and position of Don't Care. By performing grouping and division into a plurality of blocks, it is possible to increase the detection probability of a group that does not match, and to reduce the number of comparisons of input data and conditions in the matching search means.

  As described above, according to the embodiment of the present invention, in the detection apparatus including the mismatch detection unit, even when the search condition includes the value of Prefix or the Don't Care symbol, the mismatch detection result can be increased. This makes it possible to reduce the number of comparisons between input data and conditions in the match search means, improve the throughput in the search device, and further reduce power consumption.

  The present invention can be applied to a communication apparatus that can perform filtering and routing processing of packets received at high throughput even when the search condition includes a Prefix value or a Don't Care symbol.

  DESCRIPTION OF SYMBOLS 10 ... Search apparatus, 20 ... Mismatch detection means, 21 ... Delay means, 22, 23 ... Condition memory, 24 ... Data division means, 25 ... Counting means, 30 ... Match search means, 40 ... Condition memory, 50 ... Comparison part.

Claims (8)

A search device that searches for a condition that matches data input from a plurality of pre-registered conditions,
Each of the conditions belongs to one or more groups,
A memory for storing the conditions;
Mismatch detection means for detecting for each group that the input data does not match the condition;
A match search means for matching the input data with the condition and searching for a condition that matches the input data;
The condition is expanded when the Don't Care symbol included in the condition is 0 or 1, and whether or not a part or all of the condition is equal to the numerical value of each address in each address of the memory. A symbol representing this is stored for each group,
The mismatch detection means detects a mismatch for each group by reading a symbol for each group from the memory by using a part or all of the input numerical values as a read address of the memory, and the detection Means for notifying the match search means of the result,
The match search means does not perform collation between the input data and the condition of the group in which the mismatch is detected ;
The conditions are grouped based on at least one of the Prefix value, the number of Don't Care symbols, and the position of the Don't Care symbols, and a symbol indicating that a part or all of the conditions are equal to the numerical value of each address. A search apparatus characterized by being grouped so as to minimize the number of . The mismatch detection means includes
Dividing means for dividing the inputted numerical value into partial input numerical values composed of a plurality of blocks smaller than the number of digits or the number of bits;
Means for detecting, for each block, a mismatch for each group by reading a symbol for each group from the memory using the partial input numerical value as a read address of the memory;
The search device according to claim 1, further comprising: a counting unit that tabulates the mismatch detection results for each group for each block. The condition is whether Prefix value is the same, whether the condition after Don'tCare symbol is deployed are the same, the position of Don'tCare symbols based on at least one of either the same The search device according to claim 1, wherein the search device is divided into groups. The search device according to any one of claims 1 to 3, wherein at least one of the conditions belongs to only one group. A search method for searching for a condition that matches data input from a plurality of pre-registered conditions,
Each of the conditions belongs to one or more groups,
A mismatch detection step for detecting for each group that the input data does not match the condition stored in a memory;
A match search step of matching the input data with the condition and searching for a condition that matches the input data;
The condition is expanded when the Don't Care symbol included in the condition is 0 or 1, and whether or not a part or all of the condition is equal to the numerical value of each address in each address of the memory. A symbol representing this is stored for each group,
The mismatch detection step includes detecting a mismatch for each group by reading a symbol for each group from the memory using a part or all of the input numerical value as a read address of the memory, Notifying the result to the match search step,
The matching search step does not perform matching between the input data and the condition of the group in which the mismatch is detected ;
The conditions are grouped based on at least one of the Prefix value, the number of Don't Care symbols, and the position of the Don't Care symbols, and a symbol indicating that a part or all of the conditions are equal to the numerical value of each address. A search method characterized by grouping so that the number of items is minimized . The mismatch detection step includes:
Dividing the input numerical value into partial input numerical values composed of a plurality of blocks smaller than the number of digits or the number of bits;
Detecting a mismatch for each group by reading the symbol for each group from the memory using the partial input numerical value as a read address of the memory;
The method according to claim 5, further comprising a step of counting the mismatch detection results for each group for each block. The condition is whether Prefix value is the same, whether the condition after Don'tCare symbol is deployed are the same, the position of Don'tCare symbols based on at least one of either the same The search method according to claim 5 or 6, wherein the search method is divided into groups. The search method according to any one of claims 5 to 7, wherein at least one of the conditions belongs to only one group.

Images