1281721 07213twfl.doc/006 95.1.27 九、發明說明： 本發明是有關於一種生產排程方法，且特別是有關於 一種半導體晶圓代工廠之生產排程方法。 供應鏈管理（Supply Chain Management)對企業界而 言，最直接而明瞭的說法便是產銷協調，其目的是爲了確 保供應鏈中所有投資與供應的貨品能夠確實符合市場的需 求。換言之，市場的需求皆能確實的獲得有效的供給，然 而，有許多因素總是讓產銷協調無法達到完美的平衡狀態。 其中，有一個因素是時間，市場是動態變化的，需求 會隨著時間而改變，在一個產品的生命週期裡，根據市場 的需求預測（Forecast)所規劃的產能與原物料的供給，與 爾後實際的訂單（Sales Order)，常常有數量與時間的落差。 當需求大於供給，即造成缺貨損失；反之，當需求小於供 給，即造成存貨損失。因此，如何能夠在產品生命週期的 各階段’皆能精確掌握需求與供給，即爲獲利或虧損的分 野0 其次，另一個因素是能見度（VhMUty)，業界常說需 求預測不準是常態，即訂單與需求預測在某段時間內有一 個範圍的正負値的容許誤差，而供應商必須自負此誤差所 造成之備料損失。 第1圖所示，其繪示爲一工廠需求面之產能利用率圖； 以及第2圖所示，其繪示爲一工廠供給面之產能利用率圖。 一工廠中之總產能，由需求面（Demand)來看，如第 1圖所示’可分成已接受之訂單（0rder)之產能、預測未 1281721 07 213twf1. doc/006 95.1.27 來需求（Available to Promise，ATP)之產能以及閒置（Idle) 產能。而由供給面（Supply)來看，如第2圖所示，工廠 之總產能可分成已接受之訂單（Order)之產能與可利用之 產能（Capable to Promise，CTP)。 因此，當有一訂單進來時，生產排程計畫者（Planner) 必須計算此訂單之標準製程週期（Standard Cycle Time)並 正確計算的工廠剩餘可利用的產能，以確定是否可接受此 一訂單。倘若此訂單要求做完其他後續製程（Backend Process)，則工廠排程計畫者則必須將此後續製程之可利用 產能加入其排程之規劃中。 習知規劃生產排程是用一靜態方式，由生產排程計畫 者劃表逐一確認各工廠之產能，配合訂單之製程週期，計 算出工廠內可供給此訂單多少需求，再告知客戶正確交貨 日期與數量。然而，此種方式由接獲訂單到計算出正確出 貨之日期與數量往往需耗費數天的時間，而無法即時與客 戶確認訂單之出貨日期與數量。 因此本發明之目的爲提供一種半導體晶圓代工廠之生 產排程方法，以縮短生產排程計畫者確認訂單之時間。 本發明之另一目的爲提供一種生產排程方法，可即時 動態的提供生產排程計畫者各製程機台產能之情況，以幫 助生產排程計畫者正確的將訂單確認。 本發明提出一種半導體晶圓代工廠之生產排程方法， 此方法係當接獲一不需進行晶圓後段製程之訂單時，系統 (Planning System)將搜尋晶圓廠中之可提供之一產品數量， 5 1281721 07213twf1. doc/006 95.1.27 並取此訂單之產品數量與晶圓廠中可提供之產品數量兩者 間之一較小値爲可供給此訂單之產品數量。而當接獲之一 訂單是需要進行晶圓後段製程時，系統將先扣除一倉庫中 之產品數量後，再搜尋後段製程中之可利用產能與晶圓廠 中之可供給之晶圓數量，以提供給扣除倉庫之產品數量後 之一剩餘需求，然後取後段製程中之可利用產能所能完成 之產品數量與晶圓廠中之可供給之晶圓數量兩者間之一較 小値，爲可提供給此剩餘需求之產品數量，而倉庫中之產 品數量與此較小値之加總爲可提供此訂單之產品數量。 本發明之半導體晶圓代工廠之生產排程方法，可於數 十秒之內得知工廠可提供之產能，以確認是否可滿足訂單 之需求。 本發明之半導體晶圓代工廠之生產排程方法可提供生 產排程計畫者後段製程之可利用產能之訊息，以幫助生產 排程計畫者正確的將訂單確認。 本發明之半導體晶圓代工廠之生產排程方法可即時動 態的反應出晶圓於各生產線上之情況（數量、時間等）。 爲讓本發明之上述和其他目的、特徵、和優點能更明 顯易懂，下文特舉一較佳實施例，並配合所附圖式，作詳 細說明如下： 圖式之簡單說明： 第1圖係一工廠需求面之產能利用率圖； 第2圖係一工廠供給面之產能利用率圖； 第3圖係依照本發明一較佳實施例之半導體晶圓代工 1281721 07213twfl.doc/006 廠之生產排程方法流程圖； 第4圖係依照本發明一較佳實施例之一產品於此生產 排程方法供需流程圖；以及 第5圖係依照本發明一較佳實施例之相同製程之兩產 品於此生產排程方法供需流程圖。 圖式之步驟說明： 300 :訂單 302:判斷 304、306、308、310、312、314、316 :步驟 400、500、520 :訂單 402、502、522 :倉庫 404、504 :後段製程可利用之產能 406 :已下線生產之晶圓數量 408、508 ··保留給特定客戶之晶圓數量 410、510 :晶圓廠預測之晶圓數量 506 :已下線生產給A客戶之晶圓數量 523 :已下線生產給B客戶之晶圓數量 412、512、525 :晶圓數量加總 414、514、524 :進行後段製程之晶圓數量 416、516、526 :系統 418、518、528 :供給 實施例 第3圖，其繪示爲依照本發明一較佳實施例之一種半 導體晶圓代工廠之生產排程方法流程圖。 7 1281721 〇72l3twfl.doc/〇〇5 95.1.27 請參照第3圖，一半導體晶圓廠接獲一訂單300,首先， 先判斷此訂單3〇〇是否需進行晶圓後段製程(如步驟302)。 倘若訂單3〇〇不需進行晶圓後段製程，則系統將搜尋 晶圓廠中可供給之晶圓數量(ATP)(如步驟304)。之後，取 g了單女求之產品數量與晶圓廠可供給之產品數量兩者間之 一較小値(如步驟306)，最後，可供給之產品數量（如步驟 308)爲可供給訂單300之產品數量。意即當訂單所需求之產 品數量大於晶圓廠中可供給之產品數量時，僅能提供晶圓 廠中可供給之產品數量。相反的，當訂單所需求之產品數 量小於晶圓廠中可供給之產品數量時，則可完全的滿足訂 單需求之產品數量。 倘若訂單300需要進行晶圓後段製程，則系統會先去 搜尋倉庫中已現有之產品數量，而將訂單需求之產品數量 扣除掉倉庫中已現有之產品的數量(如步驟310)。之後，系 統將搜尋後段製程之可利用產能與晶圓廠中可提供之晶圓 數量，以供給扣除倉庫現有產品數量後之剩餘需求(如步駿 312)。之後，將訂單之產品所需耗用後段之製程的產能（分 鐘/機器）計算出’例如每1單位數量之產品於後段製程之 製程週期爲1分鐘’則1〇〇〇單位數量之晶圓需耗費後段製 程1000分鐘/機器之產能。然而，要完成1000單位數量之 產品必須由晶圓廠提供1000單位之晶圓來進行後段製程而 完成該產品之製作。因此，系統將取後段製程之可利用產 能所能完成之產品數量與晶圓廠可供給之晶圓數量兩者間 之一較小値，爲可提供給剩餘需求之產品數量（如步驟 8 1281721 07213twf1. doc/006 314)。接著，將倉庫中之產品數量加上i亥較小値即爲可供給 訂單之產品數量(如步驟316)。意即當後段製程之可利用產 能無法滿足晶圚廠可提供之晶圓數量時，僅能以後段製程 之可利用產能所能完成之產品數量，爲提供給扣除倉庫產 品數量後之剩餘需求。相反的，即使後段製程之可利用產 能足以供給該訂單在扣除倉庫中產品數量後之剩餘需求， 而晶圓廠中可供給之晶圓數量卻不足以提供給後段製程 時，僅能以晶圓廠中可供給之晶圓數量，爲供給扣除倉庫 產品數量後之剩餘需求。 第4圖，其繪不爲依照本發明一較佳實施例之一產品 於此生產排程方法供需流程圖。 請參照第4圖，一訂單400之需求數量爲4000單位， 而系統416搜尋到倉庫402中已有現成產品1000單位之數 量。之後，系統416扣除掉倉庫402中現成產品數量後， 還須3000單位數量的產品才可完全滿足訂單400之需求。 因此，系統416開始搜尋後段製程中之可利用產能 404，發現有2500分鐘/機器的產能，而假設每1單位數量 之產品於後段製程之製程週期爲1分鐘，則後段製程之2500 分鐘/機器可完成2500單位數量之產品。此時，系統416會 同時搜尋在晶圓廠中是否可提供2500單位數量之晶圓到後 續製程中完成此產品的製作。 通常在晶圓廠中可供給之晶圓會分成已下線生產之晶 圓數量406、保留給特定客戶之預測晶圓數量408以及晶圓 廠商預測之晶圓數量410，其中晶圓廠商預測之晶圓數量 9 1281721 07213twf1. doc/006 95.1.27 4 1 0是要提供給特定客戶較預期超量之需求或提供給其他 非特定客戶之需求。 假設已下線生產之晶圓數量406、保留給特定客戶之預 測晶圓數量408與晶圓廠商預測之晶圓數量410分別有 1000單位數量之晶圓。而系統416會先搜尋已下線生產之 晶圓數量406，再搜尋保留給特定客戶之預測晶圓數量 408，最後才搜尋晶圓廠商預測之晶圓數量410中之晶圓， 以進行後段製程。在本實施例中，此產品在後段製程中僅 有2500分鐘/機器的產能，因此系統便先由晶圓廠中之已下 線生產之晶圓數量406中取得1〇〇〇單位，再由保留給特定 客戶之預測晶圓數量408中取得1000單位，再由晶圓廠商 預測之晶圓數量410中取得500單位，加起來共2500單位 之晶圓數量412。而此2500單位數量之晶圓爲欲進行後段 製程之晶圓數量414，並需耗用後段製程2500分鐘/機器之 可利用產能404,以完成產品之製造。因此，由此可提供之 2500單位數量之產品加上在倉庫402中1000單位數量之現 成產品，可提供給訂單之供給418有3500單位數量之產品。 第5圖所示，其繪示爲依照本發明一較佳實施例之相 同製程之兩產品於此生產排程方法供需流程圖。 請參照第5圖，首先接獲一 A客戶之訂單500,其需求 量爲4000單位數量的A產品。而系統516搜尋到倉庫502 中有1000單位數量的A產品。之後，系統516扣除掉倉庫 502中現成A產品數量後，還須3000單位數量的A產品才 可完全滿足訂單之需求。 10 1281721 〇7213twfl.doc/006 95.1.27 因此，系統516開始搜尋後段製程中之可利用產能 504，發現有45⑻分鐘/機器的產能，而假設每1單位數量 之A產品於後段製程之製程週期爲1分鐘，則後段製程之 4500分鐘/機器之可利用產能可完成4500單位數量之A產 品。因此，系統516將後段製程中之可利用產能504扣除 3000分鐘/機器。此時，系統516會同時搜尋在晶圓廠中是 否可提供3000單位數量之晶圓到後續製程中完成A產品的 製作。 系統516搜尋到晶圓廠中已下線生產給A客戶之晶圓 數量506、保留給特定客戶之晶圓數量508分別有1000單 位數量晶圓，而晶圓廠商預測之晶圓數量510有2000單位 數量晶圓。其中晶圓廠商預測之晶圓數量510是要提供給 特定客戶較預期超量的需求或提供給其他非特定客戶。 因此，系統便先由晶圓廠中之已下線生產給A客戶之 晶圓數量506中取得1000單位，再由保留給特定客戶之晶 圓數量508中取得1〇〇〇單位，再由晶圓廠商預測之晶圓數 量510中取得1000單位，共3000單位之晶圓數量512。此 3000單位數量之晶圓爲欲進行後段製程之晶圓數量514, 並需消耗後段製程3000分鐘/機器之可利用產能504,以完 成A產品之製造。因此’由此可提供之3000單位數量之A 產品加上在倉庫502中1000單位數量之現成A產品，可提 供給A客戶之供給518有4000單位數量之A產品。 倘若有另一 B客戶之訂單520在A客戶之訂單500之 後進來，而B客戶之訂單需求爲4000單位數量之B產品， 1281721 07213twf1. doc/00 6 95.1,27 且此B產品與A產品所需進行之製程相同。此時，系統526 首先搜尋到倉庫522中有1〇〇〇單位數量的B產品。之後， 系統526扣除掉倉庫522中現成B產品數量後，還須3000 單位數量的B產品才可完全滿足訂單之需求。 因此’系統526開始搜尋後段製程中之可利用產能 504 ’發現剩餘有1500分鐘/機器的產能，而假設每1單位 數量之B產品於後段製程之製程週期爲1分鐘，則後段製 程之1500分鐘/機器之剩餘可利用產能可完成1500單位數 量之B產品。此時，系統526會同時搜尋在晶圓廠中是否 可提供1500單位數量之晶圓到後續製程中完成b產品的製 作。 此時，系統搜尋到已下線生產給B客戶之晶圓數量523 有1000單位數量晶圓，保留給特定客戶之預測晶圓數量508 在提供給A客戶後已無剩餘晶圓，而晶圓廠商預測之晶圓 數量510在提供給A客戶後剩餘有1000單位數量晶圓。雖 然晶圓廠中總共可提供給B產品之晶圓數量爲2000單位， 但在後段製程中僅剩1500分鐘/機器的可利用產能504。因 此，系統526僅先由晶圓廠中之已下線生產給B客戶之晶 圓數量524中取得1000單位，再由晶圓廠商預測之晶圓數 量510中取得500單位，共1500單位之晶圓數量525，此 1500單位晶圓爲欲進行後段製程之晶圓數量524，並需耗 用後段製程1500分鐘/機器之可利用產能504，以完成B產 品之製造。因此，由此可提供之1500單位數量之B產品加 上在倉庫522中1000單位數量之現成B產品之後，可提供 12 1281721 07213twfl.doc/006 給B客戶之供給528有2500單位數量之b產品。 本發明以一產品之供需方法與相同製程之兩產品之供 需方法說明之，本發明更可應用於多種以上之產品之規劃。 綜合以上所述，本發明具有下列優點： 1.本發明之半導體晶圓代工廠之生產排程方法，可於數 十秒之內得知工廠可提供之產能，以確認是否可滿足訂單 之需求。 2·本發明之半導體晶圓代工廠之生產排程方法可提供 生產排程計畫者各製程上可利用之產能之的訊息，以幫助 生產排程計畫者正確的將訂單確認。 3·本發明之半導體晶圓代工廠之生產排程方法可即時 動態的反應出晶圓於各生產線上之情況（數量、時間等）。 雖然本發明已以一較佳實施例揭露如上，然其並非用 以限定本發明，任何熟習此技藝者，在不脫離本發明之精 神和範圍內，當可作些許之更動與潤飾，因此本發明之保 護範圍當視後附之申請專利範圍所界定者爲準。 131281721 07213twfl.doc/006 95.1.27 IX. INSTRUCTIONS: The present invention relates to a production scheduling method, and more particularly to a production scheduling method for a semiconductor wafer foundry. Supply Chain Management For the business community, the most straightforward and straightforward statement is the coordination of production and sales. The purpose is to ensure that all the investments and supplies in the supply chain are indeed in line with market demand. In other words, the market demand can surely obtain an effective supply. However, there are many factors that always make the coordination of production and sales unable to achieve a perfect balance. Among them, one factor is time, the market is dynamic, and demand will change with time. In the life cycle of a product, the planned production capacity and raw material supply are predicted according to the market demand forecast. The actual order (Sales Order) often has a gap between quantity and time. When the demand is greater than the supply, it causes the loss of stock; on the contrary, when the demand is less than the supply, it causes inventory loss. Therefore, how can we accurately grasp the demand and supply at all stages of the product life cycle, that is, the profit or loss. Second, another factor is visibility (VhMUty). The industry often says that demand forecasting is not normal. That is, the order and demand forecast have a range of positive and negative tolerances for a certain period of time, and the supplier must bear the loss of the stock caused by the error. As shown in Fig. 1, it is shown as a capacity utilization map of a factory demand surface; and as shown in Fig. 2, it is shown as a capacity utilization map of a factory supply surface. The total capacity in a factory is determined by Demand. As shown in Figure 1, the capacity can be divided into accepted orders (0rder), and the forecast is not 1281821 07 213twf1. doc/006 95.1.27 to demand ( Available to Promise, ATP) capacity and idle (Idle) capacity. From the supply side (Supply), as shown in Figure 2, the total capacity of the plant can be divided into the capacity of the accepted order and the capacity available (Capable to Promise, CTP). Therefore, when an order comes in, the production scheduler (Planner) must calculate the standard cycle time of the order and correctly calculate the remaining available capacity of the plant to determine whether the order can be accepted. If the order requires completion of other Backend Process, the plant scheduler must add the available capacity of the subsequent process to the planning of the schedule. The planned production schedule is based on a static method. The production schedule planners confirm the production capacity of each factory one by one, and coordinate with the order process cycle to calculate how much demand can be supplied to the factory, and then inform the customer of the correct payment. Date and quantity of goods. However, this method usually takes several days from receiving the order to calculating the date and quantity of the correct delivery, and cannot immediately confirm with the customer the date and quantity of the shipment. SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide a semiconductor wafer foundry production scheduling method that reduces the time required for a production scheduler to confirm an order. Another object of the present invention is to provide a production scheduling method that can dynamically and dynamically provide the production scheduler's production capacity for each process machine to help the production scheduler correctly confirm the order. The invention proposes a production scheduling method for a semiconductor wafer foundry, which is to search for one of the products available in the fab when receiving an order that does not require a post-wafer process. Quantity, 5 1281721 07213twf1. doc/006 95.1.27 The smaller the number of products in this order and the quantity of products available in the fab, is the quantity of products available for this order. When one of the orders is received for the post-wafer process, the system will deduct the quantity of the product in the warehouse and then search for the available capacity in the back-end process and the number of wafers available in the fab. One of the smaller quantities required to provide the remaining demand after deducting the quantity of the product in the warehouse, and then the amount of product that can be completed by the available capacity in the back-end process and the number of wafers that can be supplied in the fab, The quantity of the product that can be supplied to this remaining demand, and the sum of the number of products in the warehouse and this smaller amount is the quantity of the product that can provide this order. The production scheduling method of the semiconductor foundry of the present invention can know the capacity that the factory can provide within a few tens of seconds to confirm whether the order can be met. The production scheduling method of the semiconductor foundry of the present invention provides information on the available capacity of the production scheduler's back-end process to assist the production scheduler in correctly confirming the order. The production scheduling method of the semiconductor foundry of the present invention can dynamically reflect the condition (quantity, time, etc.) of the wafer on each production line. The above and other objects, features, and advantages of the present invention will become more <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; Figure 1 is a capacity utilization map of a plant demand surface; Figure 2 is a capacity utilization map of a factory supply surface; Figure 3 is a semiconductor wafer foundry according to a preferred embodiment of the invention 1281921 07213twfl.doc/006 A flow chart of a production scheduling method; FIG. 4 is a flow chart of supply and demand of a production scheduling method according to a preferred embodiment of the present invention; and FIG. 5 is a same process according to a preferred embodiment of the present invention. The two products are in this production scheduling method supply and demand flow chart. Step Description of the Drawing: 300: Order 302: Judgment 304, 306, 308, 310, 312, 314, 316: Steps 400, 500, 520: Orders 402, 502, 522: Warehouses 404, 504: Backstage Process Available Capacity 406: Number of wafers that have been produced offline 408, 508 · Number of wafers reserved for a specific customer 410, 510: Number of wafers predicted by the fab 506: Number of wafers that have been produced offline to A customer 523 : Number of wafers produced by the lower line to the B customer 412, 512, 525: total number of wafers 414, 514, 524: number of wafers 416, 516, 526 for the subsequent process: system 418, 518, 528: supply embodiment 3 is a flow chart showing a production scheduling method of a semiconductor foundry in accordance with a preferred embodiment of the present invention. 7 1281721 〇72l3twfl.doc/〇〇5 95.1.27 Please refer to Figure 3, a semiconductor fab receives an order 300, first, first determine whether the order 3 需 need to perform the wafer back-end process (such as step 302 ). If order 3 does not require a post-wafer process, the system will search for the number of wafers (ATP) available in the fab (step 304). Thereafter, take one of the number of products requested by the single woman and the quantity of products that the fab can supply (such as step 306), and finally, the quantity of products that can be supplied (such as step 308) is a supplyable order. The number of products in 300. This means that when the quantity of products required for an order is greater than the quantity of products available at the fab, only the quantity of products available in the fab can be provided. Conversely, when the quantity of products required for an order is less than the quantity of products available in the fab, the quantity of the product required for the order can be fully met. If the order 300 requires a post-wafer process, the system first searches for the number of products already in the warehouse and deducts the number of products already in the warehouse from the quantity of products already in the warehouse (step 310). The system then searches for the available capacity of the back-end process and the number of wafers available in the fab to supply the remaining demand after deducting the existing product quantity in the warehouse (eg, Step 312). After that, the production capacity (minutes/machine) of the process in the latter stage of the product of the order is calculated. For example, for every 1 unit of product, the process period of the latter process is 1 minute, and then 1 unit of wafer. It takes up to 1000 minutes/machine capacity in the latter stage. However, to complete a 1000 unit quantity of the product, the fab must provide 1000 units of wafers for the back end process to complete the production of the product. Therefore, the system will take one of the number of products that can be completed by the available capacity of the back-end process and the number of wafers that the fab can supply, and the number of products that can be supplied to the remaining demand (eg, step 8 1281721) 07213twf1. doc/006 314). Next, add the number of products in the warehouse to the number of products that can be supplied to the order (step 316). This means that when the available capacity of the back-end process cannot meet the number of wafers that can be supplied by the wafer plant, the amount of product that can be completed only by the available capacity of the later process is provided to the remaining demand after deducting the quantity of the warehouse product. Conversely, even if the available capacity of the back-end process is sufficient to supply the remaining demand for the order after deducting the quantity of the product in the warehouse, and the number of wafers that can be supplied in the fab is not sufficient for the back-end process, only the wafer can be used. The number of wafers that can be supplied in the factory is the remaining demand after deducting the quantity of warehouse products. Figure 4 is a flow chart showing the supply and demand of the production scheduling method in accordance with one of the preferred embodiments of the present invention. Referring to Figure 4, the demand for an order 400 is 4000 units, and the system 416 searches for the number of 1000 units of ready-made products in the warehouse 402. Thereafter, after the system 416 deducts the quantity of ready-made products in the warehouse 402, it must also have 3000 units of the product to fully satisfy the demand of the order 400. Therefore, the system 416 begins to search for the available capacity 404 in the back-end process and finds that there is 2500 minutes/machine capacity, and assuming that the process cycle per 1 unit of product in the back-end process is 1 minute, then the second-stage process is 2500 minutes/machine. A total of 2,500 units of products can be completed. At this point, the system 416 will simultaneously search for the availability of 2,500 units of wafers in the fab to complete the production of the product in a subsequent process. Generally, the wafers that can be supplied in the fab are divided into the number of wafers 406 that have been produced offline, the number of predicted wafers reserved to a specific customer 408, and the number of wafers predicted by the wafer manufacturer 410, among which the wafer manufacturer predicts The number of rounds 9 1281721 07213twf1. doc/006 95.1.27 4 1 0 is the need to provide a specific customer with a larger demand than expected or to provide to other non-specific customers. It is assumed that the number of wafers 406 that have been produced offline, the number of wafers 408 reserved for a particular customer, and the number of wafers 410 predicted by the wafer manufacturer have 1000 units of wafers, respectively. The system 416 first searches for the number of wafers 406 that have been produced offline, searches for the predicted number of wafers 408 reserved for a particular customer, and finally searches for the wafers in the wafer number 410 predicted by the wafer manufacturer for the back-end process. In this embodiment, the product has a capacity of only 2,500 minutes/machine in the back-end process, so the system first obtains 1 unit from the number of wafers 406 that have been produced offline in the fab, and then retains The number of predicted wafers for a particular customer is 1000 units, and the number of wafers predicted by the wafer manufacturer 410 is 500 units, adding up to a total of 2,500 units of wafers 412. The 2,500-unit wafer is the number of wafers 414 to be processed in the latter stage, and the second-stage process of 2,500 minutes/machine available capacity of 404 is required to complete the manufacture of the product. Thus, the 2500 units of product that can be supplied thereby, plus 1000 units of ready-made products in the warehouse 402, can be supplied to the order supply 418 with 3,500 units of the product. Figure 5 is a flow chart showing the supply and demand of the production scheduling method for two products of the same process according to a preferred embodiment of the present invention. Please refer to Figure 5, first receiving an order 500 from an A customer with a demand of 4000 units of A product. The system 516 searches for a 1000 unit quantity of A products in the warehouse 502. Thereafter, after the system 516 deducts the number of ready-made A products in the warehouse 502, it must also have 3000 units of the A product to fully satisfy the order. 10 1281721 〇7213twfl.doc/006 95.1.27 Therefore, System 516 begins to search for available capacity 504 in the back-end process and finds that there is 45 (8) minutes/machine capacity, while assuming a 1 unit quantity of A product in the back-end process cycle For 1 minute, 4,500 units of A product can be completed in 4500 minutes/machine availability of the latter process. Thus, system 516 deducts 3000 minutes/machine from available capacity 504 in the back-end process. At this point, the system 516 will simultaneously search for the availability of 3,000 units of wafers in the fab to complete the production of the A product in a subsequent process. The system 516 searches for the number of wafers 506 that have been produced offline to the A customer in the fab, the number of wafers reserved for a particular customer 508 has 1000 units of wafers, and the wafer manufacturer predicts that the number of wafers 510 has 2,000 units. Quantity wafers. The number of wafers 510 predicted by the wafer manufacturer is to be provided to a specific customer in excess of the expected demand or to other non-specific customers. Therefore, the system first obtains 1000 units from the number of wafers 506 that have been produced in the fab to the A customer, and then obtains 1 unit from the number of wafers 508 reserved for the specific customer, and then the wafer. The manufacturer predicted that the number of wafers in 510 was 1000 units, and the total number of wafers in 3,000 units was 512. The 3,000-unit wafer is the number of wafers 514 to be processed in the back-end process, and the downstream process 3,000 minutes/machine available capacity 504 is consumed to complete the manufacture of the A product. Thus, the 3,000 units of A product that can be supplied thereby plus 1000 units of ready-made A products in the warehouse 502 can be supplied to the A customer's supply 518 with 4000 units of the A product. If there is another B customer's order 520 coming in after the A customer's order 500, and the B customer's order demand is 4000 units of the B product, 1281721 07213twf1. doc/00 6 95.1, 27 and this B product and A product The process to be carried out is the same. At this point, system 526 first searches for a 1 unit number of B products in warehouse 522. Thereafter, after the system 526 deducts the number of ready-made B products in the warehouse 522, it must also have 3000 units of the B product to fully satisfy the order. Therefore, the system 526 starts searching for the available capacity 504 in the back-end process and finds that there is a remaining capacity of 1500 minutes/machine, and it is assumed that the process cycle of the B product per unit unit is 1 minute in the latter process, and 1500 minutes in the latter process. / The remaining available capacity of the machine can complete 1500 units of B products. At this point, system 526 will simultaneously search for the availability of 1500 units of wafers in the fab to complete the production of the b product in subsequent processes. At this point, the system finds that the number of wafers that have been produced offline to B customers 523 has 1000 units of wafers, and the predicted number of wafers reserved for a particular customer 508 has no remaining wafers after being supplied to customer A, and the wafer manufacturer The predicted number of wafers 510 is 1000 units of wafers remaining after being supplied to the A customer. Although the total number of wafers available for B products in the fab is 2000 units, only 1500 minutes/machine available capacity 504 is left in the latter stage. Therefore, the system 526 only obtains 1000 units from the number of wafers 524 that have been produced in the fab to the B customer, and 500 units in the number of wafers 510 predicted by the wafer manufacturer, and a total of 1500 units of the wafer. The quantity 525, the 1500 unit wafer is the number of wafers 524 for the back-end process, and the downstream process 1500 minutes/machine available capacity 504 is required to complete the manufacture of the B product. Therefore, the 1500 unit quantity of B product that can be supplied thereby, plus 1000 units of the ready-made B product in the warehouse 522, can provide 12 1281721 07213 twfl.doc/006 supply to the B customer 528, 2,500 units of the product b . The present invention is described in terms of a supply and demand method for a product and a supply and demand method for two products of the same process, and the present invention is more applicable to the planning of a plurality of products. In summary, the present invention has the following advantages: 1. The production scheduling method of the semiconductor foundry of the present invention can know the capacity that the factory can provide in a few tens of seconds to confirm whether the order can be satisfied. . 2. The production scheduling method of the semiconductor foundry of the present invention provides information on the production capacity available to the scheduler for each process to assist the production scheduler in correctly confirming the order. 3. The production scheduling method of the semiconductor foundry of the present invention can dynamically and dynamically reflect the condition (quantity, time, etc.) of the wafer on each production line. Although the present invention has been described above in terms of a preferred embodiment, it is not intended to limit the invention, and it is obvious to those skilled in the art that the present invention may be modified and retouched without departing from the spirit and scope of the invention. The scope of the invention is defined by the scope of the appended claims. 13