WO1996029666A1 - Method and apparatus for parallel processing a calculation of a required amount of material - Google Patents

Abstract

Method and apparatus for parallel processing a calculation of required amounts of materials such as various types of parts and raw materials needed for products to be produced in accordance with a production plan, comprising one or a plurality of MPS reading units for reading MPS (Master Production Schedule), one or a plurality of parts explosion calculation units and a synchronous mechanism unit for guaranteeing that a parts explosion calculation for each item is carried out after a parts explosion calculation for parent items has been completed and controlling the process of a parts explosion parallel calculation for each item within the range of the guarantee, whereby making it possible to obtain calculation results within a shorter period of time which are equivalent to conventional sequential required material amount explosion calculation.

Description

 Edifice

Material plant 並列 Parallel processing method and equipment for mass calculation U

 Kibori Iri j relates to the method and apparatus for calculating the required amount of each ^ W product and raw materials required to produce the products produced by this production plan when raw l-jilif is manually input. Background art

 If one product, such as a motor, is specified, one of the methods for automatically calculating the number of rice and the number of parts required to make this product is as follows. Requirements Planning (hereinafter referred to as MRP) is known. Material requirements il method and equipment refer to the calculation method and equipment for this MR. Hereinafter, the calculation of material requirement S is referred to as MRP calculation.

 The MRP and the conventional MRP meter are described in, for example, “Illustrated MRP Term 500 Selection” (Nikkan Kogyo Shimbun, published in 1983).説明 I will explain to ijt.

 In the MR 計算 calculation, based on a planned product-level production plan (Master Production Schedule, hereafter referred to as MPS), for items such as parts and raw materials, "necessary items (items), when necessary (delivery date), Only as needed (required) "Calculate an arrangement plan for purchasing or manufacturing. This calculation requires the following three types of ecology.

 (1) M P S

Generally, it is the production plan of the item (product) at the top level in the parts exploded view. Also called standard production plan. (2) r¾ ππ

 品, デ, 特, デ ー タ, 品, 部品,., And parts, parts and raw materials デ ー タWhen ,',,',

A table consisting of two pieces of data, one with the data indicating the value of 0 and the other with the data. Goods The Π data example, the, 'eta', [or to produce I,必¾ when or ^睛買: (hereinafter referred to as lead time) become, and the like. Also, depending on the product topography, for example, what is the child product 1 necessary to manufacture a certain product [71, and how many child products are required for each WJ to be manufactured (this (Called the U number of the product). (3) 茆

For each item, the location at 现 'and the number of products in process, or the delivery time and number of items already scheduled for delivery in the future (order back information).

Based on such information, MRP calculation performs the following five calculations for each item.

 (1) Total requirement calculation

The data of the item is taken in, the required a is grouped into a certain ο 、, and the total location 別: is obtained for each period. In the following, for the sake of simplicity, the period will be one day, and daily planning will be explained.

 (2) Net required ffi calculation

Based on the calculated total requirements ¾, reserves for stocks and backlogs and calculate the required net shaking on a daily basis.

 (3) Lots

 The π-specified π ·: taste ¾ ¾ is given to ¾, and 品目 is set to a number suitable for arrangement using the set lot size.

 (4) Lead time meter

Lots of lots put together for a lead time! £ discount, the order date OR Create a order by issuing a hand u.

 (5) Necessary arrogance

Expand the created order to the 1st place,,, M, with parts ¾. To be specific, I am in front of the young order Π! Based on the child items and the number of products in the product feed data, the request for the child W1 is issued and written in the required quantity data for each item.

 The totals ^ from (1) to (5) above are collectively referred to as part expansion calculation hereinafter. MRP, juice ^ is || for all products set in MPS, and the parts development calculation of all items necessary for manufacturing the products is performed.

 At this time, note; what must be done is the evening when the component development calculation for each product I is performed. For example, in the case of a product with a configuration as shown in the parts exploded view in Fig. 2, item CL is a child item of item HC and item RC. Therefore, the total requirement of item CL is the sum of the requirements from the item [Π HC and item RC, and the parts exhibition of item CL | Must be done from.

 In this example, it was a relatively simple part configuration, but in the event of a sudden, calculations were performed in a state where hundreds to thousands of products were composed of tens of thousands of items in a complex manner. There is a need to do. Therefore, in the conventional MRP calculation, a low-level code is introduced, and a queue is controlled to control the component development calculation of each item. It is also common to use a pointer called an activity chain instead of a queue. The control method is essentially the same, so the following describes the conventional method using a queue as an example.

The MRP calculation will be described.

As shown in Fig. 2, each product can be assigned a level code based on the component composition hierarchy based on the component development diagram of the product. Depending on the eyes, there may be a case where a certain product's block enters the hierarchy of the number of exclusions, or a product of the number of exclusions urinates to the な る / ί. Bring it. Therefore, the lowest level code is the low level code. Here, the term “lower” refers to the direction of escape to the level of a more basic product on the part development diagram. Normally, the level code is set to the product level 0, and the number of level codes increases as the hierarchy progresses toward the basic product 0. Therefore, the oral level code is the largest level code of the item's level code. This also means that only one level code is defined for each item, much less for low-level codes. The low-level code for each item is set by searching for all items in the BOM before the MR II calculation starts, and is written in the item data as item-specific information.

 The conventional M R Ρ calculation uses this low-level code to determine the order of the parts expansion of each item at the level 'by-level'. Specifically, the process proceeds according to the following procedure.

 (1) Read MPS

 For the product-level items in MPS, write the requested quantity read from MPS. At this time, the item name is put in the level 0 queue. (2) Start calculation from level 0 queue

 When all the reading of the MPS is completed, one item name is taken out from the level 0 queue, and the part I expansion calculation is performed. When finished, retrieve the next item name from the queue. At this time, the child item name is put into the low-level code queue of the child item at the time of the required amount expansion in the parts expansion calculation. However, do nothing if already in the queue.

(3) Level 'By Level Exhibition ^ When the cue of a certain level becomes:, take out one product name from the cue of the next lower level, and perform the part expansion calculation of 'π', 11. Until it is empty, it keeps removing it from the queue at that level. Required for part expansion calculation At the time of expansion, one child product is put into the low-level code queue of its {nu}. However, it does nothing if it is already in the queue.

 (4) End of MRP meter

Queues at all levels? The calculation ends at the に point.

As already explained, the conventional MRP meter ^ has a mouth-level code so that the product exhibition calculation of each item is always performed after the parent product (parts of ^ Ι; ΗίΠ However, as a result, this control method sequentially performs the part expansion calculation for each ', ΓΙ one by one, and the number of products, the number of parts, When 'j is large, a huge amount of processing time is required, so this MRP calculation cannot be performed frequently. For example, if you want to re-create a production plan according to market trends, standing寐, Note also in case of change of the remaining such as, finely corresponding _c present invention can not be the part ¹ of the article IXi, to treat the expanded meter ^ in parallel, MR P; 1-: To shorten the processing time of production, and to quickly formulate production plans in response to fluctuations The target. DISCLOSURE OF THE INVENTION

In order to achieve the above object, the MRP calculation method and apparatus according to the present invention comprises one or several MPS reading units (1 to 3) for reading MPS as shown in FIG. It has one or more parts expansion calculators (4 to 6) for calculating the parts expansion of fl, and the parts ffi opening of each item ^ is the parent product [I m; m I; H meter: After that, it is guaranteed that it will be performed afterwards, and the parts expansion calculation of each LI is performed in parallel at f¾4; It is characterized by having one W mechanism (7). The MPS is read by one or more MPS reading units (1 to 3). The request for the embedded product ', S: is sent as a mountain message to the parts display of the product.

 The Doho mechanism part (7) sends a total of 1) start message to the parts development calculation part (4-6) that has the item whose MPS has been read. The parts calculation that received the calculation start message; 'il ;, 4 to 6) starts the parts development calculation for each item. However, the part expansion meter Τ: ί? 1; does not perform the parts l;, Π · ^ for the parts ^ for which the parts expansion calculation of the parent item has not been completed.

 The part development calculation section (4-6) is! In the calculation of the required quantity of the item for which the total is calculated, the request for the child product is a Yamapami message. : Send to the calculation part (4 to 6). Upon receiving the piled-up message, the parts unfolding calculation section (4 to 6) writes the required quantity data of the item.

 The synchronization mechanism (7) receives a report on the progress of the calculation from the parts exhibition Ι) ίΠΙ · ϊίίίΙ '; (4-6). Then, a calculation start or a continuation message is sent to the part deployment meter that has completed the part development calculation for the paddy. The synchronization mechanism (7) determines the end of the M R Ρ calculation based on the various messages sent. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is an overall block diagram, FIG. 2 is an exploded view of parts and an explanatory view of low-level code, 笫 3I is an explanatory view of a parts configuration, and FIG. Fig. 5 is an explanatory diagram of the concept of hierarchical parallelism, Fig. 6 is an explanatory diagram of the concept of the product ^ Nfc sequence, and Fig. 7 is an explanatory diagram of the concept of hierarchical parallelism. Fig. 8 is a diagram for explaining the concept of product group hierarchy parallelism. Fig. 9 is an explanatory diagram of other parallel / distributed computer environment I; Fig. 9 is an explanatory diagram of a data storage method; The figure is an explanatory diagram of the son i'i W method. Fig. 12 is a diagram I of the hierarchical parallel process model. The ^ 13 diagram is a flowchart of the synchronization mechanism process. Explanation [¾ |, and FIG. 14 is an explanatory diagram of a flowchart of the MPS reading process, and FIG. 15 is an explanatory diagram of a flowchart of the component generation process. Fig. 16 is an explanatory diagram of the flow chart of the parts stacking reception process, Fig. 17 is an explanatory diagram of the flow chart of the parts calculation process, and Fig. 18 is a process model of product group parallel 19 (¾ | is an explanatory diagram of a flowchart of the main synchronization mechanism process, and FIG. 20 is a subordinate diagram. Fig. 21 is an explanatory diagram of a flow chart of a synchronous mechanism process, and Fig. 21 is an explanatory view of a flow chart of a main synchronous mechanism process. Fig. 22 is a flowchart of a secondary synchronous machine 'process. It is an explanatory drawing.

 The present invention will be described in more detail with reference to the accompanying drawings.

 Three embodiments will be described with reference to FIGS. 3 to 22. FIG.

First, the basic concept of these examples is described. As an example, let us consider an MRP calculation with a parts configuration as shown in Fig. 3. In this part configuration,

Only three types, の, Β, and CC, share parts with each other, and the low-level code of の /: is 6. The MR.P calculation is to perform the part expansion calculation for all products "I" shown in 筘 in this figure. <However, for a certain product W: As shown in Fig. 4, the conventional sequential meter ^ Parts exhibition- In other words, the whole progresses at the level 'by' level, and in each level, the parts development calculation of the items belonging to that level is sequentially advanced.

 On the other hand, the basic idea of the 1 iff example is that, as shown in Fig. 5, the whole sequence proceeds at the level of the conventional sequential; juice: and M levels. In, parts development calculations for items belonging to that level are performed in parallel. Hereinafter, this method is referred to as a hierarchical parallel method. In the hierarchical parallel method, a synchronization mechanism is required in order to grasp the status of component development calculations that can be performed in parallel at each level, and to proceed with processing on a level-by-level basis.The mu! Jill- Perform a total of n items at each level appropriately assigned to each.

 The basic method of the second example is as follows: First, as shown in Fig. 6, all the items are divided into several groups, and the processing in each group proceeds by the same sequential calculation as before. Force The process of each group proceeds in parallel. Hereafter, this method will be referred to as the production and group parallel method. In the product ^ parallel method, each product group is assigned to the part expansion calculation unit, and each part expansion calculation unit proceeds with the sequential MRP calculation as in the past. However, in order to calculate the EI for an arbitrary item, the progress of each part expansion calculation unit must be grasped in order to satisfy the condition that the calculation must be performed after the calculation of the i item has been completed. Parts exhibition uj A total of n parts is required. Also, in relation to this, in the product-parallel method, it is relatively easy to perform the process a [in parallel if the grouping of the items satisfies a certain criterion. This standard will be described later.

As shown in Fig. 7 (3), the embodiment of 3G [I] is a combination of the floor parallel system and the product group parallel system. That is, in the product-parallel system, each group of items was processed by the conventional sequential MRP ill-: force. Then,; Ι · Τ). In each group is performed in a floor-parallel manner. Hereinafter, this method will be referred to as the product group floor / ^ parallel method. The above is the basic concept of the present embodiment. ; „', Floor)'; · ■; parallel force 'formula is a clause -Hi combination, and its fraud can be easily considered from the details of the pm parallel system and', 'η', η parallel system. Hereinafter, this explanation is omitted.

 Next, before going into the details of each embodiment of the hierarchical parallel method and the product ^ parallel method, the common items will be explained.

 In this embodiment, as shown in FIG. 8, the power is less than that of a loosely-coupled file non-shared type parallel computer, and the tightly-coupled parallel computer shown in FIG. Le co-hit! Total! It is also possible in a distributed environment where?: Machines, or these parallel computers and conventional sequential computers are connected by a network. In FIG. 8, each loosely coupled PE (Processing Element) is a single CPU. This is also a setting for convenience of discussion, and the present invention can be implemented even if each PE is a tightly coupled multi-PU with shared memory.

 As shown in Fig. 10 (a :), this example is a situation where the database management system is distributed and distributed S to each loosely coupled PE. This is possible even in a situation where the database system is managed by a specific PE as shown in). It is also possible to use a normal file system without using a database system.

In the following description, a set of calculations is called a process. A set of multiple processes is also a process. Multitasking operating systems allow multiple processes to run in parallel on a single PE. In this example, the process will be explained with an example of a model in which processes interact with each other on such a multitasking operating system. By providing a mechanism to execute, it is possible to easily realize a single task system in which one process always runs on one PE.

 Here, the terms "parallel" and "parallel" are defined in the following description. "Parallel J means that multiple processes can be processed independently and simultaneously. This means that it is logically possible to run them at the same time. For example, if the number of processes is running on a single PE in a time-sharing manner, such as in multitasking, it means that they are running in parallel. Can be. However, given a momentary question, if the PE is a single CPU, only one process is actually running. In contrast, “parallel” refers to a state in which multiple processes are physically proceeding simultaneously. If processes are running on multiple PEs, they can be said to be running "in parallel." “Parallel” is a concept that is larger than “parallel”. Processes running in "parallel" do not always run in "parallel", while processes running in "parallel" do not always run in "parallel". In this embodiment, a plurality of processes

—The explanation is based on a model that moves in parallel while exchanging dice. Although it is possible to assign all of these processes to one PE to proceed with the process, assigning them to multiple PEs allows the process to proceed in parallel. As mentioned earlier, multiple processes assigned to each PE may be sliced more finely and run in multitasking, but each process is executable. The whole process of the process may be moved sequentially.

The order of messages sent by a process to another process is based on the calculation model that the process goes to the process in general. I will tell. If this is also done using a calculation model in which 顺 ^ is not stored, for example, messages to a certain process are all sent with a serial number when they are sent, and the receiving side sends the serial ^ By providing a mechanism for processing a message in the tree, it is possible to carry out the same operation as in the tree embodiment.

When a process sends a message to another process, it needs to know which PE it is on. Therefore, it is possible to construct a system in a brilliant way by communicating which processes are on which PEs and communicating between the processes. There are several communication methods for this process, including a method using common software called a middleware for distributed environment. In this embodiment, there are 11 methods. As shown in the figure, we will proceed with the communication system called the “son station system”. This involves placing one PE inquiry mail process (8) that centrally manages which processes are on which PEs in one place. The PE mail process (8) receives a message addressed to a certain process and transfers it to the PE where the process exists. Each PE has one in-PE mail process (9-11), and the PE inquiry mail process (8) transfers the forwarded message to the destination process. Also, if a process in a PE sends a process message, specify the destination process name! ^1. Send to the in-PE mail process (9-11) of the PE where the I is located. The PE @ mail process (9-11) immediately transfers to the destination process if the process exists in its own PE, and to the PEiiij mail process (8) otherwise. Forward. This mechanism allows each process to exchange messages without knowing which PE has the process to which the message is sent. In particular, in the case of MRP i ^ The PE ll'.J mail process (8) knows whether it is calculated, and the mail process in each PE (9-11) knows only the item name calculated by its own PE. Even if each PE @ mail process (9-11) does not know anything at all, messages always arrive via the PE [i! J mail process (8), but the People. The communication method between processes by the post office method is a series of different methods, and the present invention can be implemented by another method.

 There are also several ways to assign which process to which device. This embodiment is an example of the allocation method, and other allocation methods are also possible.

 Based on the above circumstances, an embodiment of the hierarchical parallel method will be described below as 1 岙. First, the composition will be described with reference to FIG.

 The MPS reading unit comprises an MPS reading process (12 to 13).

 The Doho Organization Department consists of the Iho Organization Process (14). This is because there are a number of queue processes (15 to 17) that manage the input and output of queues at a certain level, and control to receive messages to the synchronization mechanism and to send messages from the synchronization mechanism. It consists of a process (18).

The part development calculation unit is composed of a part development calculation process (19 to 20). In a single component development meter process, a plurality of items to be calculated are defined, but one component process is always created for each item. In other words, the part exhibition I-kawa calculation process is composed of several part processes (2 1-2 2) and a part generation process (2 3) that generates the parts and processes. Each ^ process consists of two parts: a parts totaling process (24) that calculates the expansion of parts for a certain item, and a parts pile receiving process (25) that receives a pile-up message for that item. Consists of In a parallel machine with loosely-coupled files consisting of (N + 1) PEs (PE 0, PE, PE i, PEN), a database is stored on each PE. There is a management system and database (26-28),

 The same machine process (14), the PE inquiry mail process (29), and the PE @ mail process (30) are allocated to PE0. The synchronization mechanism process (14) communicates with the mail process (30) in the PE.

 For each of the remaining PEs, the MPS reading process (12 to 13), the part opening calculation process (19 to 20), and the mail process within the PE (31 to 32) are each one ^ ^ 7, ': Yes. The PE @ mail process (31 to 32) communicates with each process and the MPS reading process that constitute the part's and mlifJ processes in the PE. The intra-PE mail process (30 to 32) at each PE communicates with the inter-PE mail process (29) allocated to PE0.

 Hereinafter, the overall behavior will be described with reference to FIG. 12 along with the progress of the MRP calculation. Regarding the behavior of each process, ^ 51 3 shows the synchronization mechanism process, Fig. 14 shows the MPS reading process, Fig. 15 shows the component generation process, ^ 16 shows the component mountain acceptance process, and Fig. 17 The flowchart of the parts calculation process is summarized below. Each document describes how to proceed when a specific message is received.

 First, the control process (18) of the Doho mechanism process receives a start message from the operator or another device. The control process (18) of the synchronization mechanism process sends an MPS reading start message of "read-mps" to each MPS reading process (12 to 13).

The MPS reading process (12 to 13) that received the message to start reading, firstly, “mps started” in the control process (18) of the mechanism process (18) Number of products) "MPS 1

Send a start message. After sending the MPS start message, start reading MPS for each product. The MPS reading process is based on the product part ', exhibition 1) 9 ίΠ-τ, ij,', ',,', exhibition \], toward the part generation process (2 3) of the process. , Send a create message of “create (product name)”, and then send it to the part stacking acceptance process (25) of the product, 'π', and send the request _¾ Send as ffi message. The piled-up message may be sent in the form of messages of the entire plan in the form of messages Γ, Γ, or a specific period may be divided into several parts> 1. In this embodiment, the piled-up message is to be sent separately as a message “set (mps, day, request No.)”, with the plan i being smaller than 1 II. The first argument of this pile message indicates the source of this message. In this case, since it is a request J! L from MPS, it is "mps". When the reading of a certain product is completed, the MPS reading process sends an “MPs—finished” MPS reading end message “mps—finished” to the part stacking process (25) of the product. This is repeated for the number of ', and the processing ends.

Upon receiving the create message “create (item name)”, the component development process (2 3) generates a component process (2 2) that performs the component expansion calculation for that item. . However, if the parts process for the item has already been created, nothing is done. The generated parts process consists of two parts, a parts calculation process and a parts stacking reception process. In particular, first of all, when generating the parts calculation process, the parts table data is read from the database using the database management system \\], and the lead time, low-level code, child part name By using the information necessary for the calculation of the component development such as the number and the number of parts, these intense reports are stored in the component calculation process. Also, at this time, the part-completion process (23) is added to the control process (18) of the Doho mechanism process by "que-set (generated item name, row of the item). After that, the part set receiving process (25) is performed. Here, the MPS reading is performed in the part stacking receiving process (25). If messages are piled up from the process, those messages will be awaited in the PE mail process (31).

Parts mountain稅seen acceptance process (2 5) has received a pile of messages, request most of the database ^'Τ¾: to update the required amount of data in the database via the Present system. The new process is to rewrite the contents of a mountain message to the amount of data already written for each day. Upon receiving the MPS reading completion message "mps_finished", the message is forwarded to flji and then to the control process (18) of the M-th machine ^ process. Since the 顺 ^ of the message sent from the MPS reading process (12 to 13) is saved, the MPS reading end message always comes after the rice paddy message. Therefore, when the parts stacking reception process (25) and the MPS reading end message are transferred, the required ffi data for the item has already been completed. Also, after the parts generation process (23) sends the product queue set message, the parts stacking reception process (25) is generated, and the message to the synchronization mechanism process is always sent to the mail in the Ρ Ρ. Since the number of the message sent via the process (31) and sent from one process to another is saved, the synchronization mechanism process receives the message from the queue set message after the queue set message. It is guaranteed that the message that the product has finished reading the MPS will yield.

Companions mechanism Process Control Process (1 8), first MPS start menu Tsu Se - receive di but, _c keep count down preparative number of products in the I message also takes only ¾ of Kyuse' Tome message And the item name in the message, the queue process of the low-level code of the message Instruct ~ 17) to enter the queue and count the number of items received at each level. After that, the reception of the MPS reading end message starts. All MPS ¾i the end receive the M ps Hajimeme message narrowing from the process (1 2-1 3) only, and, manufacturing of屮of MPS start message, ¹ "the number of the total number and M a few minutes, Exit menu narrowing seen MPS Upon receiving the message, the control process (18) determines that the reading of all MPSs has been completed, by receiving the MPS start message from all of the PS reading processes, It is possible to know the total number of all products, and by receiving the number of MPS reading end messages at the end of the parts pile acceptance process, it is possible to guarantee that all those products have been read. If it is determined that the reading of the MPS has been completed, the control process (18) of the synchronization mechanism process issues a request to the level 0 queue process (15) for the product name in the queue. Go out Then, parts meter of each product ^ towards the process (2 to 4), send the parts calculations start message that "start_mrp".

 The part meter ^ process (24) receives the start message from the part meter ίΤ. When the start message is received, the required data is read at the end of the database system SR system, and the part expansion calculation is started. This means that the processing proceeds in parallel with the number of products in the level 0 queue. It works as a multitasking in order to encourage the part metering process in the same part expansion calculation process, but it works as a multitask, but the parts, '„',, i-process on another ρ は, The calculation proceeds in parallel.

 Each part meter ^ process (2 4) has the required 時 に [part unfolding raiii to perform the part unrolling of child items at the time of development.

A create message "create (child item name)" is sent. After that, the parts pile of the child product \\ is also sent to the "Prime receiving i" process (25) by "set (own item). ΰname.U, request £ i) "as an ill message. Parts of my item At the point where ^ total ^ was completed, the machine also entered the process control process (18) | | J, sent a part called “mi-p—fini shed” π | ^ Part ¹ , ¹ , 山 ί ί ί ί ί ί ί ί ί ί 送 り ί ί ί 終了 終了 終了 終了 ί ί 終了 終了 終了 終了 終了 終了 終了 終了 ί ί ί ί ί ί ί ί ί 終了 終了 終了 ί ί 終了 ί ί 終了 終了 終了 終了

The operations of the parts generation process (23), which receives the create message, and the part III cover receiving process (25), which receives the piled-up messages, are as described above. In addition, as a result, the part generation process (23) generates a queue set message, and the control process (18) of the synchronous mechanism process that receives the message is performed in the same manner. The parts stacking reception process (25) terminates the process when it receives an end message. The control process (18) of the _c- synchronous mechanism process starts from the parts calculation process (24) at the level where the calculation is in progress. Receives the calculation end message. When the number of parts calculation completion messages reaches the same number as the number of products 01 received by the queue process at that level, the control process (18) completes the parts development meter ^ for items at that level. Then, the next lower-level queue process (16 to 17) is sent out all the item names in the queue, and the ί 'item total process (24 ), The part calculation start message "start_mrp" is sent. If the queue is empty, go to the next lower level queue. By repeating the above processing, the MRP meter is advanced at the level 'by-level'. Then, at each level, the product at that level [; 1 parts development · is advanced in parallel. Since all items are managed collectively under the level-by-level control of the same machine, any item Γ: ΐ parts exhibition ^; π · η is the parent item part of the item It is kept after all the expansion calculations are completed. When the parts of the items in the queues at all levels are completed, the OH counter n is completed.

Determines the end of MR measurement and sends an end message to all processes. After sending, the process ends. The process that has received the end message ends the processing.

 Next, a description will be given of an embodiment of a 2 並列 [ΪΙ] parallel method. First of all, I will explain whether I figured out the ^ 18 figure. The MPS reading unit and the Doho mechanism unit are the same as in the example of the 1-story / parallel method. That is, the MPS reading section is composed of the MPS reading process (33 to 34).

 The synchronization mechanism consists of the same machine ^ process (35). This is because there are multiple queue processes (36-38) that input and output queues at each level fi-I !. It consists of a control process (39) that sends out messages.

 The part development calculation part consists of a part development meter ^ process (40 to 41). In one part development calculation process, a plurality of items to be calculated are defined, but one part process is always created for each item. In other words, the part expansion calculation process consists of a plurality of part processes (42 to 43), a part generation process (44) that generates a part ', a process, and a ^ 機構 mechanism process ( 4 5). In order to distinguish them from the Domo mechanism process (35) of the synchronization mechanism section, the synchronization mechanism process (35) of the synchronization mechanism section is the "main synchronization mechanism process", and the synchronization mechanism process (45) of the component deployment calculation section is " This is called the “sub-synchronization mechanism process”. The addition of the sub-synchronous machine process is significantly different from the embodiment of the hierarchical parallel system. Each 'product process' consists of a parts totaling process (47) that calculates the expansion of parts for a certain item, and a part that receives a pile message for that item r'. It consists of two.

In the sequence analyzer, each 計 of the loosely-coupled file non-symbols composed of (Ν + 1) Ρ Ε (Ρ Ε 0, ···. PE i, · ■ ·, PEN) On the database, there is a database: a physical system and a database (49 to 51), and Ε Ε 0 At the same time, the main domo machine send process (35), the PE inquiry mail process (52), and the mail process within PE (53) are arranged. The main synchronization mechanism process communicates with the mail process in PE.

 On each of the remaining PEs, one MPS reading process (33-34), a component deployment calculation process (40-41), and a mail process in the PE (54-55) are distributed one by one. . The mail processes (54 to 55) in the PE communicate with the processes constituting the 邙 ', the development process and the MPS reading process in the PE. The intra-PE mail process (53-55) in each PE communicates with the inter-PE mail process (52) allocated to PE0. There is one thing to note here. As described in the basic idea, in the product group parallel method, each of the grouped item groups is calculated by each part expansion ^ part, that is, the part development calculation process. Therefore, in the case of a policy of assigning one component development calculation process to one PE, it is necessary to divide the group according to the number of PEs. On the other hand, priority is given to ease of grouping, especially if the number of groups is greater than the number of PEs, it is possible to assign multiple parts deployment calculation processes to one PE. It is.

The overall behavior is described below with reference to Fig. 18 along with the progress of the MRP calculation. Regarding the operation of each process, Fig. 19 shows a flowchart of the main synchronization mechanism process, and Fig. 20 shows a flowchart of the sub synchronization mechanism process. The MPS reading process, the component generation process, the component stacking acceptance process, and the component calculation process are almost the same as those in the flowcharts from FIG. 14 to FIG. The main change is that in Fig. 14 to Fig. 17 in each case, the message that was sent by fi and Rona to the synchronous machine ^ process in each of the figures was sent to the synchronous machine ^ process. In the parts generation process, the sub-synchronization mechanism process is used. In the parts metering process, it is sent to the subordinate mechanism process.

First, before the MRP meter, it is grouped into the process of calculating the inventories ^ (40-41). ¹ "1il, and the top-level mouth-level code is top-level code. If the item [I, ',' is not in the group, the level code for the item is set to | ΰ] period level code, and these level codes are obtained for each group. This information is stored in the sub-community mechanism process control (46) of the parts deployment calculation process (46), which means that each group has one top-level code and Each part expansion calculation process has an ID number, which is stored in the control process (46) of the sub-synchronization mechanism process.

 Processing proceeds as follows. First, the control process (39) of the main synchronization process receives a message from the operator or another device to start the MRP calculation. The control process (39) of the main synchronization mechanism process sets request-level-msg as the request message to the control process (46) of the sub-IRJ stage machine of each part expansion calculation process. ¾: Send.

 In each component development calculation process, the control process (46) of the sub-synchronization mechanism process receives the request message, and moves from the top level to the main synchronization mechanism process (39) of the synchronization mechanism section. Send a level message "que-set-level (part deployment meter: part I, level code, level type)" to the level. For example, the top level of the level message is "Top" for the top level, "Top" for the sync level, and "Top" for other levels. Label “Continue”.

To control the synchronous machine process, the process (39) sends a level message. Upon receipt, the pair of the ID number of the message width and the level type is placed in the queue for the level code queue process (36-38) also in the message block. Instruct Count the number of messages received at each level. The control process (39) of the main ffij-ho mechanism process knows the total number of parts exhibitions; Ι · ^ processes (40-41), and sends a level message from all parts exhibitions ff] total ^ processes. At that point, it sends an MPS read ^ start message called "read-mps" to the MPS read process (33-34).

 The process of transmitting and receiving the level message may be performed in advance of the MRP calculation. In that case, as in the embodiment of the hierarchical parallel method, the MRP calculation starts from a message to start reading into the MPS.

 The MPS reading process (33 to 34) receiving the message to start reading first starts the "mps—started (product to be read from now on) for the control process (39) of the main synchronization mechanism process. Number) is sent. After sending the MPS start message, start reading MPS for each product. The MPS reading process sends a create message "create (product name)" to the component generation process (44) of the component expansion calculation process that performs the component expansion of the product ^ The request ¾ is sent as a pile message to the parts stacking reception process (48) of the parts process of the product. The MPS reading process (33 to 34) is called “mps_finis hed” at the point when the reading of a certain product is completed, and is directed to the process with the S (Shinamiyama) S (48) for that product. Send MPS reading end message. This is repeated for the number of products, and the process ends.

Part design: The part creation process (44) of the process (40-41) receives the create message "create (product E1 name)" when it receives the create message. A part process (43) for calculating the part development of the part is generated. However, if the parts process for the product 1 has already been performed, nothing is performed. The generated parts process (43) consists of two parts, a parts calculation process and a parts stacking process. First of all, when generating the parts calculation process, the parts list data is read from the database via the data base ^ 1 system, and the parts such as lead time, level level code, child part name, and number of parts are read. Obtain the necessary information for the deployment meter ^, and have this information stored in the parts calculation process. At this time, the part generation process (44) is directed to the control process (46) of the sub-companion mechanism process of its own part development calculation process (2). ) "Is sent. After that, a parts stacking reception process (48) is generated. Here, if a stacking message has been received from the MPS reading process (33 to 34) before the parts stacking reception process (48) is generated, the PE 內 mail process (54) is used. Those messages are waited.

 Upon receiving the queue set message, the control process (46) of the sub-synchronization mechanism process in the parts development calculation process compares the item name in the message with the queue process of the mouth-level code in the message. And ask them to queue. Count the number of items received at each level.

The parts stacking reception process (48) receives the message from the MPS reading process (33-34), and calculates the required amount via the database management system in the data base. Update the data. This update is a process I !. that adds the contents of the piled-up message to the request ffi already received every day. When receiving the MPS reading finished message "mps_finished", the message is immediately sent to the synchronous machine. Transfer to the control process (39) of the main synchronous function process of the structure. Since the message sent from the MPS reading process (33-34) is preserved, the MPS reading end message always yields after the piled-up message. Therefore, when the parts stacking reception process (48) sends the MPS reading end message, the update of the required amount data has already been completed.

 In the control process (39) of the main synchronous machine send process, first, the power to receive the MPS start message and the number of products in this message are counted. After that, the reception of the MPS reading end message starts. Finished receiving MPS start messages from all MPS reading processes (33 to 34) and received MPS reading end messages for the same number as the total number of products in the MPS start messages. At this point, the control process (39) determines that the reading of all MPS has been completed, opens the queue for the level 0 queue process, and sets the ID number of the part development calculation unit in the queue. For the type whose level type is “Top”, a level calculation start message “start_MRP” is sent to the control process of the sub-synchronization mechanism process of the part expansion calculation process with the ID ^. .

 When the control process of the sub-synchronous machine process (46) receives the level calculation start message "start_MRP", the queue process of the top-level code of the part development calculation process receives the item name in the queue. And send a "startjrp" part calculation start message to the part counting process for that part.

When the part meter ^ process (47) receives the message to start the part meter ^, the data is read from the database ^? K system Kanayama and the part expansion calculation is started. Each part calculation process (4 7) The part that performs πΐ

To (4 4), send a create message "create (child product ^ name)", and then そ の,! ! Part ^ Process parts stacking reception process

(48), send a message of [iJ, "set: (\ '\' s 11 people, I", required amount).]! Minute goods ί Parts development H of I At the end of all the process, send the "mrp-finished" parts total-end message to the subordinate iij-ho mechanism process (46) in the same parts deployment meter process, and receive the parts stack of your own item. It sends a termination message to the process and terminates its own processing. The part generation process (44) that received a create message and the component stacking reception process (48) that received a pile message. As a result, the component generation process (44) sends a queue set message, and the control of the sub-synchronization mechanism process in the component deployment calculation process that receives the message is performed. The process (46) behavior is similar.

When the control process (46) of the sub-synchronization mechanism process in the component development calculation process receives the message of the completion of the component calculation from the component calculation process (47) of the level where the calculation is in progress, it enters the queue process of that level. The next item name is extracted, and a component calculation start message "start_mrp" is sent to the component calculation process for that item. When the queue becomes empty, the control process determines that all of the items to be calculated in the part expansion calculation process have been completed for the item at that level, and the control process Part 1: For the InJ Tomo Organization process (39), "level—finished (part

If the next lower level is not the synchronization level, the control process (46) immediately starts the sub-synchronization mechanism process. Queued to that level of queue process in One part name is taken out, and a part calculation start message "start_mrp" is sent to the part calculation process (47) for the item 0. If the next lower level is the same level, check if the level calculation continuation message "continue_MRP (the same level code)" has been received from the main synchronization mechanism process of the same level. If not, processing is suspended until the message arrives. When the message arrives, or if it has already yielded, one item name in the queue of the synchronization level is taken out and the parts calculation process (47) for each item is taken. Sends the "start-mrp" part calculation start message. If the queue is empty, proceed in a similar manner at lower levels.

The control process (39) of the main synchronization mechanism process of the synchronization mechanism section opens the level 0 queue, sends the level calculation start message to the corresponding part expansion calculation process, and then receives the level end message. The end-of-level message comes with messages of various level codes. At a certain level, when the same number of end-level messages as already received are received, all items at that level are received. It is determined that all of the component deployment calculations have been completed. Due to the nature of how the level messages are created, the level at which the control process (39) of the main synchronization process determines the end must always be the level code at which the queue was last opened in the main synchronization process. And When the control process (39) determines that a certain level has ended, the control process (39) causes the queue process (37-38) of the next level in the main fellowship process to open the queue and display the parts in the queue. Regarding the ID No. of the open calculation department whose level class is “Top”, the control process (46) of the sub-synchronous mechanism process of the part expansion calculation process of the ID No. Send a level calculation start message "start-MRP". Also, if the level type is "Synchronous" Then, to the control process (46) of the sub-synchronization mechanism process of the part extension ijfj calculation process of the ID ^ ff, continue the level calculation called "continue—the MRP (the rebeno record)". send a message.

 By repeating the above processing, the MRP calculation proceeds on a level-by-level basis. Each part ^: The process (40 to 41) is basically independent, and the progress of the level.by.level realization is advanced. , Ιί ': Go and proceed. A parent item of a certain item is another part deployment meter! ?: In the process, if ,, Ϊ Ϊ Ϊ Ϊ Ϊ · · · · 部品 部品 部品 部品 部品 部品 部品 部品 部品 部品 部品 部品 部品, 部品 部品 部品 部品 部品 部品 部品 部品 部品 部品 部品Processing is suspended at the item level until completion is confirmed. With such a synchronization mechanism, it is guaranteed that the part expansion of any item! Η · ^ will be performed after all the part expansion calculations for the parent item of that item have been completed. When the control process of the synchronization unit determines that all levels have been completed; at the time ^, the synchronization unit process determines the end of the MRP calculation, sends an end message to all processes, and ends the process. . The process that has received the end message ends the processing.

In the above example of the product-parallel system, the sub-synchronization mechanism process (45) in each component deployment calculation process executes the component deployment meter for each item at a certain level one by one sequentially. In particular, at the synchronization level, all processing at that level is suspended until a calculation continuation message arrives from the main synchronization mechanism process. Here, it is also possible to achieve a speedup of 10 by interrupting only a part, not all. That is, whether there are any items in the queue of the sub-synchronization mechanism process (45) that are processed by the other parts m [] fjni-η process in the parent item; If you have escalation information and are waiting for a continuous message at the synchronization level, In this case, the control process (46) of the Deputy Dopomo Machine Process processes the parent product “: 1” from the queue at that level, and extracts one item that is not calculated by the other part expansion calculation process. ', Exhibition 1 Upon receiving the calculation end message, item ii takes out the next item that is not counted by another ί'ίΐί M10η process and proceeds. When the parent item remains in the queue, and only II, which is calculated by the other parts expansion calculation process, remains, the processing is interrupted and the calculation continuation message is sent. In this way, H.W, which previously did nothing as a waiting question during suspension, can be effectively used as a time for processing to be performed, and the processing of all holidays can be improved. It becomes possible.

 Also, in the above-mentioned product-parallel system example, depending on the method of grouping the items, each part deployment meter is often interrupted at the synchronous level in the process (40 to 41), and processing is near May be. Therefore, by dividing the items into groups according to a certain S standard, it is possible to escape the processing without interrupting each part development calculation process (40 to 41).

 Here, some terms used in the following description regarding grouping will be explained.

 (1) Top level

 O The highest level of low-level code for the items in the group o

 (2) Write level

If the parent product EJ belongs to the group and has a component development calculation outside the group, the “: 1 level code is written and called the level. That is, the claim of the item may be written from another part deployment and metering process. Certain groups have the opportunity to complete multiple harmed levels. The lowest written level among them is called the lowest written level. (3) Write level

 If “Π”, “„ ”, and“ を 持 つ ”that belong to the group have a component expansion calculation outside the group, the level code of the item“ il ”is called a write level. [There is a possibility that one part may be transferred to another part development calculation process. Some groups may have multiple write levels. The level is called the highest harm level.

 Each part expansion calculation. The ^ in the grouping that allows the process to proceed without any blocking is as follows. That is,

 Criteria: In each group: ½The lowering level must be higher than the maximum writing level.

 As an example of a special grouping that satisfies the above criteria, there is a method in which an item group connected by parent and child ra is used as a root with items of level 0 as a root. However, for items that have multiple items as a parent item, such as a common part, a group of items that are connected in a parent-child relationship with itself as a root is another group. Due to this division, the top level of each group is either a level 0 product or a shared part with multiple parent items. The eyes will be in that group. As a result, the lowest write level in each group will be the top level in that group and will always meet the above criteria.

When grouping that satisfies the above criterion is performed, the operation of the main synchronization mechanism process (35) of the synchronization mechanism section and each component deployment; the operation of the sub-synchronization mechanism process (45) of the calculation section is simplified. As a result, the component deployment calculation can proceed in parallel without any breaking of each component deployment process at the synchronization level. Below, each without interruption at the same level Explain the behavior of the process.

 First, the items that are grouped into each process> '., *, M ι total ^ process before MR p leaf n are identified, the upper low level code is set as the top level, and the low written level is started. Levels, and ask for these levels for each group. This esoteric report is stored in the control process of the sub-synchronous mechanism process of the parts deployment meter ^ process. Also, each component deployment calculation process has I, which is also stored in the control process of the Deputy Domoho process.

 When the MRP meter is started, the control process (46) of the sub-synchronization mechanism process in each part expansion calculation process is started by the control process (39) of the main synchronization mechanism process of the synchronization mechanism section. From the level to the lowest level, send the level message "que_set-level (ID number of part development calculation unit, level code, level type)". In the level type column of this level message, for example, label “Start” if the level code is the start level; otherwise, label “Continue”.

 The operation of the control process (39) of the main synchronization mechanism process when a level message is received is the same as in the previous discussion.

The operations of the MPS reading process, the component generation process of the component development calculation process, the component stacking reception process, and the operation of the component totaling process are the same as those already described. When the main control process of the long-term mechanism process (39) determines that reading of all MPSs has been completed, it opens the queue for the level 0 queue process, For the part of the ID ^ sign of the computation part that has “內” and “level” of “Start”, the part of the ID f ^ “”, the total number of the processes and the control process of the sub-community mechanism process sends the level calculation Hajimeme message that "s tart- MRP" _c sub synchronization mechanism process control process (4 6) is called "start MRP" When the bell calculation start message is received, the top-level queue process of the parts development calculation process is caused to take out one item name in the queue, and the part of the product f I; the juice calculation process (47) Send a start message of "^ _start_mrp".

When the control process (46) of the subordinate mechanism process in the parts deployment meter ^ process receives a message from the parts calculation process (47) at the level where the calculation is in progress, it receives that part calculation end message, The queue process of this item takes out the next item name, and sends a component start: " ^M start_nirp" message to the component calculation process for that item. When the queue becomes empty, The control process (46) determines that all the component expansion calculations for the item at that level among the items to be calculated in the component expansion calculation process have been completed, and the main synchronization mechanism process (3 Sends a level end message “level—finished (ID number of the part development calculation unit, finished level code)” to 5), and the control process (46) immediately starts the sub-synchronization mechanism process. Noso The queue process at the first level retrieves one item name in the queue, and sends a “start_mrp” part calculation start message to the lower-level item calculation process (47). If the queue is empty, proceed at a lower level as well.

<o

The control process (39) of the main synchronization mechanism process of the synchronization mechanism section opens the level 0 queue, sends the calculation start message to the corresponding component development calculation process, and then receives the level end message. End-of-level messages are powerful when messages of various level codes are received.At a certain level, when the same number of end-level messages that have already been received are received, the parts deployment calculation for all items at that level is performed. Judge that all has been completed. Due to the nature of the way level messages are created, The level at which the control process of the initial mechanism process determines the end is always the level code that opened the queue after S at that time in the main “π” initial mechanism process 。. The process opens the queue for the queue process at the next level of the main machine ^ process 、, and displays the parts that were in the queue [¾ The ID type of the calculation unit is “start” for the level type. If it does not, send a level start message of "start_MRP" to the control process (46) of the sub-computer process of the part expansion calculation process of the ID number.

 By repeating the above processing, the MRP measurement proceeds on a level-by-level basis. Each part unfolding process receives the calculation start message and, at its own tempo, performs the level 'by.' Level calculation, so the part unfolding calculation for each part f | Can be advanced. In the part expansion calculation process, ½Because the calculation is started after all calculations at the lower level and above the level are completed, the part expansion calculation for any item is based on the part expansion of the parent item U of that item , Π-are guaranteed to be done after all have been completed. The M R Ρ calculation ends when the control process of the synchronization mechanism determines that all levels have ended. Industrial applicability

 As described above, in the parallel processing method and apparatus for calculating the material requirement according to the present invention, the component development calculation of each item is advanced in parallel by a plurality of component extensions m # 5, and each of the ', In ΙΊ, it is possible to ensure that the calculation can be started after the parent item is completed. As a result, it is possible to shorten the conventional sequential calculation of M R Ρ calculation and calculation calculation of の and the like in a short time.

Claims

^ Scope of request

1. In the equipment that performs S-calculation, one or several MPS'-incorporated parts that read the production plan to be used as a reference, and a parts deployment meter for multiple items -One or several parts expansion calculation parts and each part expansion calculation part, n, '„', 11 ', ν,', expansion i; fj total η is the part of the parent item It is guaranteed that the calculation will be performed after the completion of the expansion calculation, and it is composed of the Im unit that instructs each part expansion calculation unit to start calculation, and each parts expansion calculation unit processes the parts expansion meter ^ in parallel A material requirement calculation device characterized by the following.

 2. In the material calculation machine S described in claim 41, the M P

The S reading unit reports the number of products to be read when reading each of the industrial and industrial vehicles to the synchronous machine ^ unit first, and then reads when each product has been read. The MPS reading unit that sends the end report to the component development calculation unit that writes the required amount data for each product.

 Each of the parts deployment calculators is a parts deployment calculation unit that sends a report of the end of the damage of required quantity data of each product to the synchronization mechanism unit,

 The synchronous mechanism unit is a synchronous grenade unit for detecting completion of reading of all planned products and writing of required amount data.

 3. Claim scope 40 1 iJi! : In the computing device, the component expansion calculation unit is a component expansion calculation unit that performs parallel processing of component expansion calculation for items of the same low-level code,

 The above-mentioned synchronizing mechanism section is a synchronizing mechanism section which controls processing by level-by-level.

4. 3 ^ 所 ΪΪ I I 装 装 装 装 装 同 E: In the equipment E, the same machine unit consists of a synchronization mechanism process with a queue of each mouth level code, The part development π I-n part specializes in updating the required quantity data of one item and calculating the part development, and one or i number of part processes and a part that generates the part process. Process and

 The synchronization mechanism process is a process for giving an instruction to the parts process in charge of each item in the parts expansion calculation unit from the upper queue to 顺 to calculate the expansion of the items in the queue. ,

 Each of the parts processes described above is applied to the parts generation process of the part development manager in charge of the child item so as to form the parts process of the child item at the time of the parts expansion calculation of the expanded item. ^ After that, instruct the part process of the child item to include the request ffi for the child item in the required data, and after the completion of the part expansion calculation, calculate the item for the previous synchronization mechanism process. Is the process of reporting that

 Each of the component generation processes, when generating a component process of the specified item, sends the item name to the same level queue as the low-level code of the item in the same peer mechanism process. Is the process of instructing

 When the synchronization mechanism process receives the calculation completion report from each part process by the number of items in the queue at that level, the synchronization mechanism process instructs to perform the expansion calculation of the items in the queue at the next level. Material requirement _ calculation device characterized by the following.

 5. Claims 1. In the equipment requirement calculation device for the 1st match,

 Divide all items to be subjected to parts development calculation into one or more groups,

Each of the component development calculation units receives one of the item groups as a processing target, performs a component development calculation of each item [: 1 at the level 'by level', and a parent item of a certain item is processed. When calculated by other parts development calculators. Is a part development calculation unit that suspends processing at the level of the item [l until the processing at the next higher level of the item is completed in all the part development calculation units.計 ίΠ の 部 す る こ す る こ す る こ す る こ 計 計 す る こ 部 す る こ 部 部 す る こ す る こ 部品 す る こ す る こ 部品 部品 す る こ す る こ す る こ す る こ 部品 部品 す る こ す る こ す る こ す る こ す る こ す る こ 部品 す る こ 部品 部品 す る こ す る こ资: Π: 特 徴: 特 徴 特 徴: 资:::: 资: 资:::: 资 资: 资 资.

 6. Claims I 5, Material requirements for LM; Si:

 The synchronization mechanism comprises a master synchronization drawing process having a queue for each low-level code;

 The parts expansion section includes a sub-synchronization mechanism process similar to that of the synchronization mechanism section, one or more parts processes specializing in calculation of new and part development of the required data of one product I, and The top-level low-level code of the item that is calculated is defined as the top level, and the parent item is the low-level code of the item that is calculated by another parts development calculation unit. The level is the same as the level. The levels other than the synchronization level are the continuation levels from the top level to the lowest level. For the queue of the main synchronization mechanism process, from the top level to the lowest level, the ID number and level of each part development calculation unit Part development calculation unit that inputs a set of

 The main synchronization mechanism process starts, in order from the upper queue, the total ^ for the part expansion calculation section having the top-level level II among the ID expansion numbers of the part expansion calculation sections in the queue. Is the process of instructing

The sub-synchronization mechanism process of each part deployment meter, which received the start of the measurement, takes out the product I in the queue first from the upper-level queue to そ の, and performs the component development calculation so that Is a process that instructs the parts process of The ^ part ', process instructs the parts generation process of the parts development calculation part in charge of the child item to compose the parts process of the child item when calculating the parts expansion of the item that was found. Then, instruct the part process of the child item to include the -request ffi for the child item in the required S data, and then, after the part exhibition [¾ ί ·: This is a process to report that the calculation of the item has been completed.

 When the part generation process generates the part process for the item indicated by ^, the part name is put in the same level queue as the low-level code of the item in the deputy I second stage machine process. This is a process that instructs the sub-synchronization mechanism process.

 When the sub-synchronous machine process receives the calculation completion report from each component process, it instructs the next item in the queue at that level to perform the expansion calculation, and when the queue becomes empty, the level becomes higher. To the main synchronization process, and if the next level is not the synchronization level, instruct the system to remove one item from the next level queue and perform a deployment calculation, If the next level is the synchronization level, the process is suspended until a measurement continuation instruction is received from the main synchronization process, and when a calculation continuation instruction is received, one item in the next level queue is extracted and expanded. A process that instructs the user to perform a calculation

The main synchronous machine send process receives a report on which level has been completed from each part expansion calculation unit, and if it receives the same number of reports as the number of parts expansion calculation units queued at a certain level, it enters the next queue. Among the ID numbers of the part development calculation part, the part development calculation part whose level type is the top level is instructed to start the calculation, and the calculation is performed to the part development calculation part of the synchronization level. Material requirement equipment that features a process to instruct to continue.

7. The required range ffl 6 ΙΓΠ, the ff material ff 成

 Previous Each part exhibition l) When the synchronous machine process of the H section is interrupted at the synchronization level, the parent item of the items at the same level is processed by another part development calculation section. A material requirement calculation device characterized by the fact that it is a process that speeds up processing by indicating digits to perform component development calculation for the component process of uncalculated product 1.

 8. In the method of material requirement, one or a plurality of MPS reading units read standard production plans, and one or a plurality of parts exhibition Pf] total units each include a plurality of items. After the parts expansion calculation of the parent item is completed, the parts expansion calculation for each item, which is π ι in the η The material requirement calculation is characterized by performing a total ^ 1 start for each part expansion calculation section and guaranteeing that each part expansion calculation section processes parts expansion calculation in parallel. Method.

 9. In the material requirement calculation method described in the claim {8},

 First, each MPS reading unit first reports the number of products to be read when each of the standard production plans are read to the Dopho mechanism unit, and when the loading of each product is completed, a report of the completion of reading is given. , And send it to the parts development calculator where the data for each product is written.

 Each of the parts deployment calculation units sends a report of the completion of fli addition of required quantity data of each product to the synchronization mechanism unit,

 A method for calculating a material requirement, wherein the synchronization mechanism detects that the reading of all products in the metabolic disease plan and the insertion of the required data into the IF have been completed.

 1 0. In the request range ffl 8th, K material I total ^ method,

The parts development unit proceeds in parallel with the parts development unit ^ of the item of the same low-level code, ijij, 材 A material requirement method, characterized in that the fHj mechanism mechanism controls processing based on level-by-level.

 1 1. ^ Requirement range iffl ^ i 0 W

 The 冋 冋 machine consists of a synchronization mechanism process with a queue of each mouth-level code,

 Each of the above parts exhibition 1 total! The part performs the update of the required quantity data of one item and the parts development meter: ι, ϊ-1 "]. · The part that performs one or more parts processes and the parts process, 'HWI : Composed of

 The fiij synchronous machine send process instructs the parts process in charge of each item in each parts expansion calculation unit to calculate the expansion of the items in the queue in order from the upper queue,

 Each of the component processes indicates a digit to the component generation process of the component development calculation unit in charge of the child item so that a component process of the child item is generated at the time of the component expansion calculation of the instructed item, After that, it instructed the part process of the child item to include the required amount for the child item in the required amount data, and after the completion of the part expansion meter, the calculation of the item was completed for the synchronization mechanism process. Report that

 Each of the component generation processes is configured such that, when a component process of an item specified is generated, the synchronization mechanism is configured to put the name of the item in the same level queue as the low-level code of the item in the previous term machine process. Instruct the process

 Upon receiving the calculation completion report from each part process by the number of items in the queue at that level, the Doho Machine 'process calculates the sum of', 11 1 in the queue at the next level. The material required to be instructed to do so is fi I '.

1 2. In the ti ffl ^ 8 Jj'i,, | Divide all items to be subjected to parts development calculation into one or more groups,

 Each of the parts decompressor ^ takes charge of one of the j'jii's 1 groups as a processing target, and performs a parts decomposer ^ of each part l of the part group at a level.by level. If the parent, 'π', LI of a certain item 1 is calculated by another part expansion HI-calculation unit, the processing at the next higher level of the item ends in all the part expansion calculation units. Until the product at the R level? 1 !.

 The synchronization mechanism unit instructs each part expansion calculation unit to start or continue counting based on the progress of each part extension n t- ^ unit, and thereby the parts in each part expansion unit ^ unit A method of calculating material requirements, characterized in that the exhibition [jfj! Π · is advanced in parallel.

 1 3. In the method of calculating the material requirement of the first and second claims,

 The synchronization mechanism unit comprises a main synchronization mechanism process having a queue of each mouth-level code,

 Each of the component development π I- units includes a sub-synchronization mechanism process similar to the synchronization mechanism unit, one or more component processes dedicated to updating required data of one item and calculating component development, and The top-level low-level code of the item calculated there is defined as the top level, and the parent product [ΞΙ is the row of the item calculated by the other parts development calculation unit. The level code is the synchronization level, and the levels other than the synchronization level are the continuation levels from the top level to the lowest level. For the queue of the main synchronization mechanism process, the ID number of each part development calculation unit from the top level to the lowest level and Enter a set of level rice,

The main synchronous machine process counts in order from the higher-order queue to the part deployment meter that has the top-level level type among the ID numbers of the part exhibitions D i '. In the queue. To start, After receiving the calculation P / j start m, each sub-synchronization mechanism process of the expansion calculation unit takes out one item in the queue from the upper-level queue to を, and performs the part expansion calculation. Instructs the part process of the item to execute the part process of the child item at the time of the part expansion calculation of the item pointed to by the child item. Instruct the part process of the child item to write the required amount for the child item to the required part of the process. After the end of the total number of parts, the secondary machine reports to the process that the calculation of the item has been completed.

 Each of the component generation processes described above, when generating a component process for the indicated item, puts the item tl name in the same level queue as the low level code of the item in the i, W ΙΓί] period machine ^ process. Instruct the secondary synchronization process to

 In addition, when receiving the calculation completion report from each component process, the sub-domo partner process instructs the next item in the queue at that level to perform the expansion calculation, and when the queue becomes empty, the level becomes lower. Is reported to the main synchronization process, and if the next level is not the synchronization level, one item u in the next level queue is taken out and the unfolding calculation is instructed. If the next level is the term level, the process is suspended from the main synchronization process until the calculation continuation instruction succumbs, and when the calculation continuation instruction arrives, one item in the next level queue is deleted. Take out, instruct to perform the unfolding calculation,

Before; ΕMain fnj m machine Each process;, ν, ', Expand ;; υ ;; ΊΊνι receives a report on which level has been completed, and calculates the deployment of parts queued at a certain level. When the same number of reports are received, the part deployment meter in the next queue! Of the part ID numbers in the? Part, the part deployment meter ^ part whose level is the top level A material method, which instructs the parts expansion calculation unit at the same level to start the measurement, and instructs it to continue the measurement.

 1 4. Claim range ffl 1 3 X i;

 When the “re-in” period process of each of the above-mentioned parts is interrupted at the synchronization level, the parent item of the items belonging to the same level is counted by the other parts expansion unit. However, by instructing the component process of the product iii to perform a component development calculation, the material processing is speeded up by processing the material.

 1 5. Claims 1 2 In the method of calculating material requirements for

 When all parts Hi for which parts are to be totaled are divided into groups, the lowest level of the level of the items belonging to the group whose parent product belongs to another group is the lowest written level, and the items belonging to that group The highest write level is the highest level of the item that the child item falls into the other group, and the lowest write level is higher than the highest write level in each group. Material calculation method characterized by the fact that the processing proceeds without interruption.