TW508504B - Method and apparatus for serving data - Google Patents
Method and apparatus for serving data Download PDFInfo
- Publication number
- TW508504B TW508504B TW86117151A TW86117151A TW508504B TW 508504 B TW508504 B TW 508504B TW 86117151 A TW86117151 A TW 86117151A TW 86117151 A TW86117151 A TW 86117151A TW 508504 B TW508504 B TW 508504B
- Authority
- TW
- Taiwan
- Prior art keywords
- data
- patent application
- scope
- item
- buffer
- Prior art date
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Description
508504 A7 _ B7 五、發明說明(1 ) 本發明之背景: 本發明係關於用於資料輸送之伺服器。傳統的伺服器 之設計傾向於積極涉及資料的實體傳送。對於諸如按需式 視頻或按需式伴唱之應用來說,其需要即時地輸送高數目 的數位視頻串。數位視頻串典型上包括根據I s〇/ IEC11172 或 ISO/IEC13818 (势分別 爲一般所熟知之Μ P E G — 1標準和μ P E G — 2標準) 所壓縮的視頻資料。 在歐洲第9 5 · 2008 1 9 . 1號專利申請案中已 提出一種擴展超越僅具有單純資料輸送能力之A Τ Μ (非 同步傳送模式)系列的伺服器系統。 在P C Τ (專利合作條約)所公開之第W〇 96/ 〇 8 8 9 6號國際專利申請案已提出一種用於被建構成一 種AS I C (特殊應用積體電路)之ATM通信的串式處 理器。 ’ 歐洲第E P 0 6 6 7 7 1 3 A 2號專利申請案 中提出一種根據Μ E P G標準來記錄和再生經壓縮之視頻 資料的方法。在此情況中,經壓縮的視頻資料以包含掃描 資訊之特別方式而被記錄在磁碟中,使得特別壓縮之視頻 資料再生裝置能夠獲得V C R功能(例如快轉,快速倒帶 本發明之槪述: 本發明之目的在於改善上述的習用技術及/或提供用 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) .f 經濟部智慧財產局員工消費合作社印製 I n n n n n I I I n n i n in n If n -4 - 經濟部智慧財產局員工消費合作社印製 508504 A7 ____ Β7 _ 五、發明說明(2 ) 於未來之應用的伺服器。 本發明第一方面提供一種將A T Μ單元的V P I / V C I轉換成內部I D之方法,其包括下列步驟: 根據各V Ρ I / V C I入口的一部分而將V ρ I / V C I入口分類成在一表中的分段; 接收一 A Τ Μ單元; · 根據所接收之ATM單元之VPI/VCI的該部分 來選擇即將被搜尋的分段; 對所選擇之分段實施搜尋,以便找出對應於所接收之 ATM單元之VP Ι/VC I的入口;以及 輸出對應於所找到之入口的內部I D。 根據本發明之方法的較佳實施例則說明於附屬之子申 請專利範圍中。 本發明更提供一種將厶丁“單元之¥?1/7(:1轉 換成內部I D的-置,其包括: —表,用來儲存VP I/V C Γ入口,並且被分割成 分段; 根據各VP I/VC I入口的一部分而將VPI/ V C I入口分類成表中之分段的機構; 根據所接收之A T Μ單元之V P I / V C I的該部分 來選擇即將被搜尋之分段的機構;以及 對所選擇之分段實施搜尋,用以找出對應於所接收之 ATM單元之VP I/VC I入口和用以輸出對應於所找 到之入口的內部I D之機構。 > 本纸張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) ~ -5- (請先閱讀背面之注意事項再填寫本頁}508504 A7 _ B7 V. Description of the invention (1) Background of the present invention: The present invention relates to a server for data transmission. Traditional server designs tend to actively involve the physical transfer of data. For applications such as on-demand video or on-demand accompaniment, it is necessary to deliver a high number of digital video streams in real time. Digital video strings typically include video data compressed in accordance with ISO / IEC11172 or ISO / IEC13818 (potentially the well-known MPEG-1 standard and μPEG-2 standard, respectively). In European Patent Application No. 9 · 2008 · 19.1, a server system has been proposed that extends beyond the ATM (Asynchronous Transfer Mode) series with only pure data transmission capabilities. International Patent Application No. WO96 / 〇8 8 9 6 published in PCT (Patent Cooperation Treaty) has proposed a string type for ATM communication constructed as an AS IC (Special Application Integrated Circuit) processor. ’European Patent Application No. EP 0 6 6 7 7 1 3 A 2 proposes a method for recording and reproducing compressed video data according to the M EP G standard. In this case, the compressed video data is recorded on the magnetic disk in a special way containing scanning information, so that the specially compressed video data reproduction device can obtain the VCR function (such as fast-forward, fast-rewind the description of the present invention : The purpose of the present invention is to improve the above-mentioned conventional technology and / or provide the paper size applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (Please read the precautions on the back before filling this page) .f Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs I nnnnn III nnin in n If n -4-Printed by the Consumers’ Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 508504 A7 ____ Β7 _ V. Description of the invention (2) Servers for future applications The first aspect of the present invention provides a method for converting an VPI / VCI of an ATM unit into an internal ID, which includes the following steps: classifying the VρI / VCI entry into one according to a part of each VPI / VCI entry Segments in the table; Receive an A TM unit; · Select the segment to be searched according to the part of the VPI / VCI of the received ATM unit; The segment is searched in order to find the entry corresponding to the VP I / VC I of the received ATM unit; and to output the internal ID corresponding to the found entry. A preferred embodiment of the method according to the invention is described in The attached child is in the scope of patent application. The present invention further provides a device that converts the unit "? 1/7 (: 1 into an internal ID), which includes:-a table for storing the VP I / VC Γ entry, And is divided into segments; a mechanism that classifies VPI / VCI entries into sections in the table according to a part of each VP I / VC I entry; selects the part that will be selected based on the VPI / VCI part of the received AT MU unit The organization of the searched segment; and a search of the selected segment to find the VP I / VC I entry corresponding to the received ATM unit and the mechanism to output the internal ID corresponding to the found entry ≫ This paper size applies to Chinese National Standard (CNS) A4 (210 X 297 public love) ~ -5- (Please read the precautions on the back before filling this page}
508504 Α7 ----____ Β7 經濟部智慧財產局員工消費合作社印製 五、發明說明(3 ) 本發明第二方面提供一種裝置,其用來將資料送到 A T Μ網路並且接收來自a T Μ網路之資料,該裝置包括 (a)—匯流排介面’用來介接在一主機、一儲存裝 置和該裝置之間的匯流排支援通信; (b ) —非同步傳送模式(ATM)介面,用來介接 A T Μ網路; (c ) 一發送單元,用來將輸出資料從匯流排介面傳 送ATM界面,該發送單元包括: (1 ) 一第一 RAM介面,用來介接被用做緩衝器的 R A Μ ’此緩衝器用來暫存來自匯流排介面的輸出.資料; (2 )用來來自緩衝器之輸出資料分割成輸出Α τ Μ 單元的機構;以及 (3 ) —通信量整形器,用來與分割機構配合下控制 輸出ATM單元-ATM介面之通信量;以及 (d ) —接收單元,用來將輸入資料從A Τ Μ介面傳 送到匯流排介面,該接收單元包括: (1 )用來對輸入ATM單元實施VP I/VC I過 濾之機構, (2 )利用輸入A Τ Μ單元之酬載重組輸入資料之機 構,以及 (3 ) —第二RAM介面,用來介接被用做緩衝器之 R A Μ,此緩衝器用來暫存來自重組機構的輸入資料。 根據本發明之此裝置提供與大量用戶和分布於系統上 (請先閱讀背面之注意事項再填寫本頁) Φ 訂--- 線丨· 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -6- 508504 A7 B7 五、發明說明(4 ) 管理模組互動之日常應用管理,以及對被輸送之資料的管 理。根據本發明之伺服器提供時間或處理能量來實施較高 層次的管理任務,當主機較不積極涉及資料之實體傳送時 。根據本發明之硬體能夠在不同性能要求下即時輸送資料 ,並且相當適合這種的即時輸送。根據本發明之串式處理 器能夠支援同時與許多用戶通信,並協助視頻串式任務。 根據本發明之伺服器亦提供互動能力,例如對任何類型的 用戶提供資料之服務。即將被輸送之內容可以以多種形式 (亦即原始或未格式化之形式)儲存在根據本發明·之伺服 器中。 本發明之第三方面提供一種用來使來自儲存裝置的資 料成串的方法,該方法包括下列步驟: 爲叢發資料提供寫入到緩衝器寫入位址,至少該寫入 位址的一部分係不連續的; 經由在主k、儲存裝置與串式裝置之間的匯流排支 援通信,將叢發資料從儲存裝置傳送到緩衝器; 根據該寫入位址,將該叢發資料寫入該緩衝器中;以 及 依線性方式將資料從緩衝器中讀出。 根據本發明之方法的較佳實施例將發明附屬之子申請 專利範圍中。 本發明更提供一種用以使來自儲存裝置之資料成串的 串式裝置,其包括: 經由在主機、儲存裝置與串式裝置之間的匯流排支援 本紙張尺度適用中國國家標準(CNS)A4規格(21(^ 297公釐) (請先閱讀背面之注意事項再填寫本頁) 訂--- 線丨· 經濟部智慧財產局員工消費合作社印製 J^85〇4 A7 B7 五、發明說明(5 ) 通信’接收來自儲存裝置之叢發資料的機構; 爲該叢發資料提供寫入位址之機構,該至少寫入位址 的〜部分係不連續的;以及 一緩衝器’其根據寫入位址來儲存該叢發資料,並且 ί衣線性方式輸出從那裡來的資料。 本發明的第四方面提供一種輸送資料之方法,其包括 下列步驟: 從一主機載入至少一對位址和命令; 將該資料儲存在一緩衝器中; 根據讀取指標而從該緩衝器中讀出該資料; 若偵測到該位址與由該讀取指標所指定之位址匹配時 ’則執行該命令;以及 在該命令的執行之後輸送讀取自該緩衝器。 本發明更提供一種輸送資料之裝置,其包括: 一命令方塊\用以儲存載入自主機之一對位址和命令 ,並且偵測該位址與由暫存該資料之緩衝器的讀取指標所 指定的位址之間的匹配; 一機構,當偵測到匹配時,該機構配合該命令方塊執 行該命令;以及 一機構,在該命令的執行之後,用來輸送讀取自該緩 衝器之資料。 本發明之第五方面提供一種輸送資料之方法,其包括 下列步驟: 接收來自網路的資料; 本纸張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐)508504 Α7 ----____ Β7 Printed by the Consumer Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (3) The second aspect of the present invention provides a device for sending data to the AT Μ network and receiving data from a T The information of the M network, the device includes (a)-a bus interface 'for interfacing with a host, a storage device and the bus to support communication between the device; (b)-asynchronous transfer mode (ATM) An interface for connecting to the AT M network; (c) a sending unit for transmitting output data from the bus interface to the ATM interface, the sending unit includes: (1) a first RAM interface for connecting the RA MU used as a buffer This buffer is used to temporarily store the output data from the bus interface; (2) the mechanism used to divide the output data from the buffer into output A τ Μ units; and (3)-communication A quantity shaper is used to control the communication volume of the output ATM unit-ATM interface in cooperation with the division mechanism; and (d) —a receiving unit is used to transmit input data from the ATM interface to the bus interface. The receiving unit includes : (1) Used to lose The ATM unit implements the VP I / VC I filtering mechanism, (2) the mechanism for reorganizing the input data by using the payload of the input A TM unit, and (3) the second RAM interface, which is used to interface with the buffer used as a buffer RA M, this buffer is used to temporarily store input data from the reorganization agency. This device according to the present invention is provided with a large number of users and distributed on the system (please read the precautions on the back before filling out this page) Φ Order --- line 丨 · This paper size applies to China National Standard (CNS) A4 specification (210 X 297 mm) -6- 508504 A7 B7 V. Description of the Invention (4) Management of the daily application interaction of the management module and management of the data being transmitted. The server according to the present invention provides time or processing energy to perform higher-level management tasks when the host is less actively involved in the physical transfer of data. The hardware according to the present invention can transfer data in real time under different performance requirements, and is quite suitable for such real-time transport. The string processor according to the present invention is capable of supporting simultaneous communication with many users and assisting video string tasks. The server according to the present invention also provides interactive capabilities, such as services for providing information to any type of user. The content to be delivered can be stored in the server according to the invention in various forms (i.e., raw or unformatted). A third aspect of the present invention provides a method for stringing data from a storage device, the method comprising the steps of: providing a write-to-buffer write address for a burst of data, at least a portion of the write address Is discontinuous; sends the burst data from the storage device to the buffer via the bus support communication between the master k, the storage device and the serial device; writes the burst data according to the write address The buffer; and reading data from the buffer in a linear fashion. A preferred embodiment of the method according to the present invention claims the children of the invention as patented. The present invention further provides a string device for stringing data from a storage device, which includes: Supporting a paper standard applicable to Chinese National Standard (CNS) A4 via a bus between the host, the storage device, and the string device Specifications (21 (^ 297 mm) (Please read the precautions on the back before filling out this page) Order --- Line 丨 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs J ^ 85〇4 A7 B7 V. Description of the invention (5) Communication 'a mechanism for receiving burst data from a storage device; a mechanism for providing a write address for the burst data, at least ~ part of the write address is discontinuous; and a buffer' its basis Write the address to store the burst of data, and output the data from there linearly. The fourth aspect of the present invention provides a method for transmitting data, which includes the following steps: Loading at least one pair of bits from a host Address and command; store the data in a buffer; read the data from the buffer according to the read index; if it is detected that the address matches the address specified by the read index ' Nori The command is executed; and the data is read from the buffer after the execution of the command. The present invention further provides a device for transmitting data, which includes: a command block for storing a pair of addresses and commands loaded from the host And detect a match between the address and the address specified by the read index of the buffer that temporarily stores the data; a mechanism that, when a match is detected, the mechanism cooperates with the command block to execute the command; And a mechanism for transmitting data read from the buffer after the execution of the command. A fifth aspect of the present invention provides a method for transmitting data, which includes the following steps: receiving data from the network; Zhang scale is applicable to China National Standard (CNS) A4 (210 X 297 mm)
Ilt (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 訂—-------線丨—— -8 - 508504 A7 ________________________ B7 五、發明說明(6 ) . 當所接收之資料被傳送到儲存裝置時,偵測該資料中 至少一預設的位元型樣; 當偵測到該預設的位元型樣時,將對應於該資料中之 預設位元型樣之位置的位置資訊加進到一表列中; 將該資料儲存在該儲存裝置中;以及 根據該表列中之位置資來訊控制從該儲存裝寘到該網 路之資料的輸送。 此方法之較佳實施例說明於附屬的子申請專利範圍中 〇 本發明更提供一種輸送資料之裝置,其包括: 從網路接收資料之接收機構; 一型樣偵測器,當該資料從該接收機構傳送到儲存該 資料的儲存裝置時,用來偵測該資料中之至少一預設的位 元型樣; 一列表,當te型樁偵測器偵測到該預設的位元型樣時 ’用來將對應於該資料中之預設的位元型樣之位置資訊儲 存在該表列中;以及 --機構,其根據該表列中之位置資訊來控制從該儲存 裝置到該網路之資料的輸送。 本發明之第六方面提供一種通信量整形方法,其包括 下列步驟: 將一或多個第一串分類成一或多個類型,每一個類型 包括具有相同位元率特性之一或多個串; 設定一組參數來控制每一個類型之位元率;以及 本紙張尺度適用中國國家標準(CNS)A4規格(210 x 297公釐) (請先閱讀背面之注意事項再填寫本頁) Φ 丨丨訂---------線* 經濟部智慧財產局員工消費合作社印製 一 9- 經濟部智慧財產局員工消費合作社印製 508504 A7 ______ B7 五、發明說明(7 ) 根據該組參數來執行每一個類型之配速。 此方法之較佳實施例說明於附屬之子申請專利範圍中 〇 本發明更提供一種通信量整形器,其包括: 將一或多個第一串分類成一或多個種類之裝置,每一 個類型包括具有相同位兀率特性之一或多個串; 一儲存裝置,用來儲存一組參數以控制每一個類型之 位元率;以及 根據該組參數來執行每一個類型之配速的機構。 附圖之簡略說明: 當參照所附之圖式來閱讀下面的說明時,本發明之優 點,特徵和詳細內容將會變得淸楚。 圖1顯示交互式通信量系統之一般系統架構; 圖2顯示根齒本發明之裝置之實施例的詳細方塊圖; 圖3顯示圖2中之Tx位址轉換器的方塊圖; 圖4顯示圖3中之位址轉換器之使用的例子; 圖5 A,5 Β,5 C顯示用於傳輸控制協定網際網路 協定(T C P I P )封包化之位址轉換的個別例子; 圖6顯示圖2中之Tx速率方塊之使用的例子; 圖7顯示由圖2中之通信量整形器所達成之位元率的 情形; 圖8顯示用來說明在單一單元期間串之送出的圖形; 圖9顯示不同種類之單元的提出; 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁)Ilt (Please read the notes on the back before filling this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs ----------- Line 丨 —— -8-508504 A7 ________________________ B7 V. Description of Invention (6) When the received data is transmitted to the storage device, at least one preset bit pattern in the data is detected; when the preset bit pattern is detected, it will correspond to the pre-set bit pattern in the data. Add the location information of the location of the bit pattern to a list; store the data in the storage device; and control the information from the storage device to the network based on the location information in the list Data transfer. A preferred embodiment of this method is described in the scope of the attached sub-application patent. The present invention further provides a device for transmitting data, which includes: a receiving mechanism for receiving data from the network; a pattern detector, when the data is from When the receiving mechanism sends to the storage device storing the data, it is used to detect at least one preset bit pattern in the data; a list, when the te pile detector detects the preset bit pattern 'Style time' is used to store the position information corresponding to the preset bit pattern in the data in the list; and-the mechanism, which controls the storage device from the storage device according to the position information in the list Transfer of data to the network. A sixth aspect of the present invention provides a traffic shaping method, which includes the following steps: classifying one or more first strings into one or more types, each type including one or more strings having a same bit rate characteristic; Set a set of parameters to control the bit rate of each type; and this paper size applies the Chinese National Standard (CNS) A4 specification (210 x 297 mm) (Please read the precautions on the back before filling this page) Φ 丨 丨Order --------- line * Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 9- Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 508504 A7 ______ B7 V. Description of the invention (7) According to the parameters of this group To perform each type of pace. A preferred embodiment of this method is described in the attached sub-application patent scope. The present invention further provides a traffic shaper, which includes: a device that classifies one or more first strings into one or more types, each type including One or more strings with the same bit rate characteristics; a storage device for storing a set of parameters to control the bit rate of each type; and a mechanism for performing the pace of each type according to the set of parameters. Brief description of the drawings: When reading the following description with reference to the accompanying drawings, the advantages, features and details of the present invention will become clear. FIG. 1 shows a general system architecture of an interactive traffic system; FIG. 2 shows a detailed block diagram of an embodiment of the device of the present invention; FIG. 3 shows a block diagram of a Tx address converter in FIG. 2; FIG. 4 shows a diagram Examples of the use of the address converter in Figure 3; Figures 5 A, 5 B, and 5 C show individual examples of address translation for the transmission control protocol Internet Protocol (TCPIP) packetization; Figure 6 shows Figure 2 An example of the use of a Tx rate block; Figure 7 shows the bit rate achieved by the traffic shaper in Figure 2; Figure 8 shows the graph used to illustrate the sending of strings during a single unit; Figure 9 shows the different Proposal of types of units; This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling this page)
-10- 508504 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明說明(8 ) 圖1 0係用於圖2中之通fe重整形益之架構的方塊圖 j 圖1 1顯示圖2之命令方塊的方塊圖; 圖1 2係用來說明圖2中位元組交換器之操作的圖形 圖1 3顯示使用者網路介面(UN I )中所使用之A T Μ單元的格式; 圖1 4顯示圖2中之VP I/V C I轉換器的方塊圖 圖1 5顯示圖2中之型樣偵測器的方塊圖; 圖1 6顯示圖2中之位址轉換器的例子。 符號說明 10 伺服器 1 2 儲存i體單元 14 地方ATM轉換器 16 公共網路 18 機頂盒 20 儲存裝置 22 儲存裝置控制器 2 4 周邊元件互連 2 6 P C I 橋 2 8 主機 3 0 主 C P U ' 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) n i IK 1 MmmmmmMm m f>— tmmmmm i SI IB mmmmmm i n amM§ mmm§ emm§ ^ ^ a* ih··· μμμ Μημβ ι·>μη m··· ·§ I 9 n n·· an 1· Hi · Mmmaw tmt n 1 n· 1· emmam i MmmB emmmmm tmmmme Mmmmmm mmm§ mmmmmmm mmmmmmm Mmmmi (請先閱讀背面之注意事項再填寫本頁) -η - 508504 經濟部智慧財產局員工消費合作社印製 A7 B7_五、發明說明(9 ) 3 2 主記憶體 3 6 串式處理器 3 8 P C I介面 4 0 介面 4 2 外部R A Μ 4 4 串緩衝器 4 6 外部R A Μ 4 8 串緩衝器 50 發送路徑 5 4 Τ X位址轉換器 5 6 檢驗和方塊 5 7 檢驗和暫存器 5 8 R A Μ介面 6〇 串資料完整性(S D I )方塊 62 通信i整形器 64 位元組調換器 6 6 命令方塊 68 速率改變指示器產生器 7 0 分段方塊 8 0 接收路徑 8 4 V P I / V C I過濾方塊 8 6 重組方塊 8 8 T C P檢驗和查對方塊 90 R X RAID 或 SDI 方塊 (請先閱讀背面之注意事項再填寫本頁) -tii! I — — — — — — — — — — IIIIIII — — — 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -12- 經濟部智慧財產局員工消費合作社印製 508504 A7 B7_ 五、發明說明(10 ) 92 型樣偵測器 9 4 R A Μ介面 9 6 R X位址轉換器 102 暫存器 104 增量控制器 106 計數器 13 0 D Μ Α叢發 13 2 T C P標題 13 4 I P標題 14 2 序列 200 高優先權方塊段 2 0 2 低優先權方塊段 203 記憶體 204 更新邏輯 2 0 6 記#體 208 通信量邏輯 2 10 參考時鐘 2 12 暫存器 2 16 高優先權發送疗列 2 18 記憶體 220 通信量邏輯 222 低優先權發送佇列 300 命令產生器 3 0 2 暫存器 本紙張尺度適用申國國家標準(CNS)A4規格(210 X 297公釐) ----—--------------訂-----—»IAWI (請先閲讀背面之注意事項再填寫本頁) -13- 508504 經濟部智慧財產局員工消費合作社印製 A7 B7_五、發明說明(11 ) 304 比較器 306 環繞計數器 308 比較器 310 比較器 3 12 及閘 3 14 佇列 316 命令暫存器 3 5 0 字元 3 5 2 字元 400 雜亂函數 4 0 2 VPI/VCI 表 420 搜尋處理器 500 對齊電路 502 匹配電路 5 0 4 暫备器 506 型樣偵測控制器 600 叢發資料 3 10 位址 3 12 位址 3 0 0 位址 3 0 1 位址 3 0 2 位址 3 0 5 位址 3 0 6 位址 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) ---------訂 *-------· 丨 -14- 508504 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(12) 3 0 7 位址 1 7 0〜2 0 0 位址 2 0 4 位址 較佳實施例之詳細說明: 圖1顯不一互動式通信系統之較佳實施例的一般系統 架構,此爲實際上支援任何一種互動式多媒體應用之寬頻 系統。請特別注意到即時多媒體輸送模式應用。 一伺服器1 0作爲V 0 D (按需式視頻)伺服器、 K〇D (按需式伴唱)伺服器,及/或網際網路伺服器等 ’並且與當作客戶之STBs (機頂盒)18通信於公共 網路1 6之上。伺服器1 〇包括一地方非同步傳送模式( ATM)轉換器1 4和數個透過地方ATM轉換器1 4而 互相連接之SMUs (儲存媒體單元)12。地方ATM 轉換器1 4之主赛目的在於排定S M U s 1 2之間的資料 路線(例如從一個S M U到另一個S M U複製根據. MPEG標準所壓縮的電影),建立在伺服器1〇中以 ATM爲基礎的L AN ’以及與公共網路]_ 6介接。每個 S M U 1 2在高速以電流技術例如最高爲6 2 2 M b p s 而與地方A T Μ轉換器1 4通信,公共網路1 6爲選擇性 的,而伺服器1 〇可以直接與S T B s 1 8通信。 圖2顯示SMU 1 2的詳細方塊圖,SMU 1 2具有 儲存裝置2 0、一主機2 8和一串式處理器3 6做爲主要 的單元,這些單元係透過周邊元件互連(P C I )匯流排 (請先閱讀背面之注意事項再填寫本頁) t 訂·· -線丨·! 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 508504 A7 B7 五、發明說明(13 ) 2 4而互相連接。主機2 8中的一個主電腦3 〇和一主記 憶體3 2係以傳統的組態透過Μ I P S匯流排3 4來連接 ;在本實施例中,Μ I P S匯流排3 4係經由一 p c I橋 2 6而連接到P C I匯流排2 4,主機2 8主要是爲了與 用戶或S T B s互動之,諸如V〇D、Κ〇D、網際網路 伺服器之日常應用而設計的。 儲存裝置2 0包括一或多列硬碟,這些硬碟係經由 S C S I或光纖頻道來連接,並且儲存即時的敏感資料( 像是MP E G — 2編碼之視頻串)和資料封包的內容(像 是沒有標題之T C Ρ / I Ρ (傳輸控制協定/網際網路協 定封包的本體)。 串式處理器3 6最好是建構成單一 AS I C (特殊應 用積體電路)。串式處理器3 6使即時的敏感資料和資料 封包成串。串式處理器3 6具有一發送路徑5 0和一接收 路徑8 0 (做爲k要部分)以及一 P C I介面3 8和一介 面4 0。發送路徑5 0處理從儲存裝置2 0和主機2 8到 地方ATM轉換器1 4之輸出資料串。接收路徑8 0處理 由地方ATM轉換器1 4到儲存裝置2 0和主機2 8之輸 入資料串。發送路徑和接收路徑之高速連接和獨立性使得 在雙向同時爲6 2 2Mpb s。 PC I介面38使PC I匯流排24與發送路徑50 和接收路徑8 0介接。P C I介面3 8在發送路徑5 0上 將輸出資料串從P C I匯流排2 4傳送到 PCI F I F052 ’並且在接收路徑上將輸入資料串 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) t 訂---------線丨· 經濟部智慧財產局員工消費合作社印製 -16- 观504 A7 五、發明說明(14) 從p c 1 F I F〇9 8傳送到P c I匯流排2 4。 介面4 0使發送路徑5 0和接收路徑8 0與一個連接 到地方A T Μ轉換器1 4之外部實體層裝置(未示出)介 接。介面4 0可包括根據UT〇P I Α (用於ATM之環 球測試及操作Ρ Η Y介面)第二級標準的兩種A T Μ介面 ’其中一種爲8位元寬之資料路徑模式的U Τ〇Ρ I Α介 面’另一種爲1 6位元寬之資料路徑模式的UTOP I A 介面。 發送路徑5 0係由數個功能方塊組成,他們一起作用 來實施高速度傳輸發送。 發送路徑5 0中的第一個方塊爲Tx位址轉換器5 4 ’其將輸出資料串從PC I F I F〇52放置入配置於 一外部R A Μ (隨機存取記憶體)4 2中之串緩衝器4 4 的主機指記憶體位置。這使得受控地將資料''散播〃到不 連續記憶體之中%其對於重組所謂的R A I D (平價磁碟 之冗餘陣列)操作之操作係非常有效的,此R A I D操作 確保資料串的整體性和T C P / I P封包化。 TC Ρ / I P檢驗和方塊5 6提供硬體支援來計算 T C P / I P檢驗和,其功能係對各封包計算並保持部分 的檢驗和,直到所有的資料都已經被傳送爲止。T C P / I P檢驗和方塊5 6和Τ X位址轉換器5 4 —起工作而直 接在串緩衝器44中建立丁0?/1?封包。丁0?/ I P標題和封包之酬載係分別置於串緩衝器4 4中,且通 過保持部分的檢驗和之檢驗和方塊5 6。一旦所有的資料 本纸張尺度適用中國國家標準(CNS)A4規格(210 x 297公釐) !!——it (請先閱讀背面之注意事項再填寫本頁)-10- 508504 Printed by the Consumer Property Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention (8) Figure 1 0 is a block diagram of the structure used for reshaping benefits in Figure 2 Figure 1 1 Shows Figure 2 A block diagram of the command block; Figure 12 is a diagram for explaining the operation of the byte switch in Figure 2. Figure 13 shows the format of the AT MU unit used in the user network interface (UN I); 1 4 shows a block diagram of the VP I / VCI converter in FIG. 2. FIG. 15 shows a block diagram of the pattern detector in FIG. 2. FIG. 16 shows an example of the address converter in FIG. 2. DESCRIPTION OF SYMBOLS 10 Server 1 2 Storage unit 14 Local ATM converter 16 Public network 18 Set-top box 20 Storage device 22 Storage device controller 2 4 Peripheral component interconnection 2 6 PCI bridge 2 8 Host 3 0 Main CPU 'This paper Standards are applicable to China National Standard (CNS) A4 (210 X 297 mm) ni IK 1 MmmmmmMm m f > — tmmmmm i SI IB mmmmmm in amM§ mmm§ emm§ ^ ^ a * ih ··· μμμ Μημβ ι · > μη m ···· § I 9 nn ·· an 1 · Hi · Mmmaw tmt n 1 n · 1 · emmam i MmmB emmmmm tmmmme Mmmmmm mmm§ mmmmmmm mmmmmmm Mmmmi (Please read the precautions on the back before filling this page) -η-508504 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7_V. Invention Description (9) 3 2 Main Memory 3 6 Serial Processor 3 8 PCI Interface 4 0 Interface 4 2 External RA Μ 4 4 String Buffer 4 6 External RA M 4 8 String buffer 50 Transmission path 5 4 TX address converter 5 6 Checksum block 5 7 Checksum register 5 8 RA Μ interface 60 string data integrity (SDI) block 62 Communication i Shaper 64 Byte Converter 6 6 Command Block 68 Rate change indicator generator 7 0 Segmented block 8 0 Receive path 8 4 VPI / VCI filter block 8 6 Reassembly block 8 8 TCP check and check block 90 RX RAID or SDI block (please read the notes on the back before filling (This page) -tii! I — — — — — — — — — — — IIIIIII — — — This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -12- Consumption by Employees of Intellectual Property Bureau, Ministry of Economic Affairs Printed by the cooperative 508504 A7 B7_ V. Description of the invention (10) 92 type detector 9 4 RA Μ interface 9 6 RX address converter 102 temporary register 104 incremental controller 106 counter 13 0 D Μ Α 丛 发 13 2 TCP header 13 4 IP header 14 2 Sequence 200 High priority block 2 0 2 Low priority block 203 Memory 204 Update logic 2 0 6 Note # Body 208 Traffic logic 2 10 Reference clock 2 12 Register 2 16 High priority sending treatment queue 2 18 Memory 220 Traffic logic 222 Low priority sending queue 300 Command generator 3 0 2 Temporary register This paper standard applies to China National Standard (CNS) A4 specification (210 X 297 public) Li) ------------- ------- Order -----— »IAWI (Please read the notes on the back before filling this page) -13- 508504 Printed by the Consumers’ Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7_V. Description of the invention (11) 304 comparator 306 surround counter 308 comparator 310 comparator 3 12 and gate 3 14 queue 316 command register 3 5 0 character 3 5 2 character 400 messy function 4 0 2 VPI / VCI table 420 search Processor 500 Alignment circuit 502 Matching circuit 5 0 4 Standby 506 Pattern detection controller 600 Burst data 3 10 Address 3 12 Address 3 0 0 Address 3 0 1 Address 3 0 2 Address 3 0 5 address 3 0 6 address This paper size is applicable to China National Standard (CNS) A4 specification (210 X 297 mm) (Please read the precautions on the back before filling this page) --------- Order * ------- · 丨 -14- 508504 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (12) 3 0 7 Address 1 7 0 ~ 2 0 0 Address 2 0 4 Detailed description of the preferred embodiment of the address: FIG. 1 shows the general system architecture of the preferred embodiment of an interactive communication system, which supports virtually any type of interactive multimedia Applications of broadband systems. Please pay special attention to the application of instant multimedia delivery mode. A server 10 is used as a V 0 D (on-demand video) server, a KOD (on-demand accompaniment) server, and / or an Internet server, etc., and is used as STBs (set-top boxes) as customers. 18 communication on the public network 16. The server 10 includes a local asynchronous transfer mode (ATM) converter 14 and a plurality of SMUs (storage media units) 12 connected to each other through the local ATM converter 14. The main purpose of the local ATM converter 14 is to schedule the data route between SMU s 12 (such as copying a movie compressed according to the .MPEG standard from one SMU to another SMU), established in the server 10 to ATM-based LAN 'and interface with public network] _ 6. Each SMU 1 2 communicates with a local ATM converter 14 at a high speed with current technology such as 6 2 2 M bps. The public network 16 is selective, and the server 10 can directly communicate with STB s 1 8 通信。 8 communications. Figure 2 shows a detailed block diagram of the SMU 12. The SMU 12 has a storage device 20, a host 28, and a serial processor 36 as the main units. These units are converged through a peripheral component interconnect (PCI). Row (Please read the precautions on the back before filling this page) t Order ·· -line 丨 ·! This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 508504 A7 B7 V. Description of the invention ( 13) 2 4 and connected to each other. A host computer 30 and a main memory 32 in the host 28 are connected through a M IPS bus 34 in a conventional configuration; in this embodiment, the M IPS bus 34 is connected through a pc I The bridge 26 is connected to the PCI bus 24, and the host 2 8 is mainly designed to interact with users or STBs, such as VOD, KOD, and daily applications of Internet servers. The storage device 20 includes one or more rows of hard disks, which are connected via SCSI or fiber channels, and store real-time sensitive data (such as MP EG-2 encoded video strings) and the contents of data packets (such as Untitled TC P / IP (Transmission Control Protocol / Internet Protocol packet body). The serial processor 3 6 is best constructed to form a single AS IC (Special Application Integrated Circuit). The serial processor 3 6 Make real-time sensitive data and data packets into a string. The string processor 36 has a sending path 50 and a receiving path 80 (as part of k), and a PCI interface 38 and an interface 40. The sending path 50 processes the output data string from the storage device 20 and the host 28 to the local ATM converter 14. The receiving path 80 processes the input data string from the local ATM converter 14 to the storage device 20 and the host 28. The high-speed connection and independence of the transmission path and the reception path make it simultaneously 6 2 2Mpb s in both directions. The PC I interface 38 enables the PC I bus 24 to interface with the transmission path 50 and the reception path 80. The PCI interface 38 is in the transmission path. 5 0 will output data string from PCI The bus 2 4 is sent to PCI FI F052 'and the input data string is received on the receiving path. The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (Please read the precautions on the back before filling this page) t Order --------- line 丨 · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs-16- View 504 A7 V. Description of Invention (14) Transfer from pc 1 FIF〇 9 8 to P c I Confluence Row 2 4. Interface 40 enables the sending path 50 and receiving path 80 to interface with an external physical layer device (not shown) connected to the local AT M converter 14. The interface 40 may include an interface according to UTOPI Α (Global testing and operation of the PTM Y interface for ATM) Two AT Μ interfaces of the second-level standard 'one of which is the U TOP I 8-bit wide data path mode Α interface' the other is 1 UTOP IA interface with 6-bit wide data path mode. The transmission path 50 is composed of several functional blocks that work together to implement high-speed transmission and transmission. The first block in the transmission path 50 is Tx address conversion. Device 5 4 'It puts the output data string from the PC IFIF〇52 into the configuration The host of the string buffer 4 4 in the external RAM (random access memory) 4 2 refers to the memory location. This allows the data to be “spread” into discontinuous memory in a controlled manner. The operation of RAID (Redundant Array of Affordable Disks) operation is very effective. This RAID operation ensures the integrity of the data string and TCP / IP packetization. The TC P / IP checksum block 5 6 provides hardware support to calculate the T C P / IP checksum. Its function is to calculate and maintain part of the checksum for each packet until all the data has been transmitted. The T C P / IP checksum block 5 6 and the TX address converter 5 4 work together to establish a 0? / 1? Packet directly in the string buffer 44. The D0 / IP header and the payload of the packet are placed in the string buffer 44 respectively, and the checksum of the holding part and the box 56 are passed. Once all the information, this paper size applies the Chinese National Standard (CNS) A4 specification (210 x 297 mm) !! —— it (Please read the precautions on the back before filling this page)
I ! I I I 線丨· 經濟部智慧財產局員工消費合作社印製 508504 Α7 - Β7 五、發明說明(15 ) 均在串緩衝器4 4中,則檢驗和値被放置於T C P / I P 標題的正確位置內,而封包已可發送。 R A Μ介面5 8構成在外部R A Μ 4 2和發送路徑 5 0之間的介面,外部RAM4 2可包括雙埠SDRAM (同步動態隨機存取記憶體),此外部R A Μ 4 2包括數 個串緩衝器4 4,以便使來自儲存裝置2 0之磁碟的叢發 資料通信分離,並將所需之固定位元率資料串提供給 ATM網路1 6。每一個串緩衝器處理一個輸出資料串, 在與輸入方向的對比上,由於輸出方向上之資料流動特性 係完全可預測(控制)的,所以可預先估計緩衝器要求。 因此,串緩衝器4 4係靜態地配置於外部R A Μ 4 2之中 0 Τ X RAID或SDI (串資料整體)方塊60提 供資料冗餘的支援。T X位址轉換器5 4將所需的資料放 置在串緩衝器4、中,然後,當資料係從串緩衝器4 4中 輸出時,若儲存裝置2 0中的其中一個磁碟損壞時,貝(J Τ X R A I D方塊6 0修正錯誤資料。 通信量整形器6 2控制從串緩衝器4 4到A Τ Μ網路 1 6之輸出資料串,其設計係爲了非常精確之配速和低 CDV (單元延遲變異)。通信量整形器6 2由兩個主要 的分段所組成,其中一分段處理高優先權資料(例如視頻 通信量),另一分段則處理低優先權的一般資料通信量。 命令方塊6 6想要去除特別是即時之敏感工作之主機 2 8的負載,執行外向資料串中準確已知的位置之內容的 本紙張尺度適用細家標準麵纖格,顏^ j· --------t (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 訂---------線丨----- 5〇85〇4 A7 B7 五、發明說明(16 ) 傳輸所觸發的動作。 分段方塊7 0將從串緩衝器4 4提供進入A A L — 5 PDUs (ATM適配層—5協定資料單元)之輸出 資料串分段,並且將A A L — 5 P D U s對應到A Τ Μ 單元中。若輸出資料串爲MPEG—2 SPTS(單一 程式輸送串),分段方塊7 0能夠將 mpeg — 2 spts中之二ts封包分段成一個 A A L - 5 PDU,除非在 MPEG — 2 SPTS 中 剩下不到兩個T S封包,在後者的情況中’ A A L - 5 P D U對應到八個A Τ Μ單元。在一般的情 況中,AAL - 5分段係由PDU尺寸所控制’而每一串 中的P D U尺寸係可編程的。 接收路徑8 0具有幾個對應於發送路徑5 0之方塊的 反向操作的方塊。 —V Ρ I /\ c I (虛擬路徑識別碼/虛擬頻道識別 碼)過濾方塊8 4對輸入A Τ Μ單元實施快速且有效率之 VPΙ/VCI過濾,這是藉由在VPι/vcI表中的 入Π上將雜亂和線性搜尋函數結合而完成。 一重組方塊8 6基本上實施分段方塊7 0之相反功能 ’重組方塊8 6使用ATM.單元之酬載來重建 AAL - 5 PDUs ,然後將 AAL — 5 PDUs 對 應到上層資料(例如Μ P E G — 2 S P T S,T C P / I P封包)。 若輸入資料串係透過TC P來發送,則一TC P檢驗 本紙張尺度適用中國國家標準(CNS)A4規格(210 χ 297公釐) t (請先閱讀背面之注意事項再填寫本頁) 訂---------線丨· 經濟部智慧財產局員Η消費合作社印製 508504 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(17 ) 和查對方塊8 8查對T C P標題中之T C P檢驗和。 一型樣偵測器9 2使得有限數目之位元型樣被偵測於 輸入資料串中。一表列被建立,其準確地出指定之位元型 樣產生於串中的什麼地方,這支援某些可以快速執行之處 理任務,而否則他們必須以後處理來完成。 —R X RA I D或SD I方塊9 0將冗餘加到輸入 資料串。若一序列N個字兀被寫入一緩衝益(未不出)中 ,則這N個字元的奇偶性接著被寫入,此功能可以被開啓 /關閉。若輸入資料串將會被儲存在儲存裝置2 0之中’ 並且稍後透過發送路徑5 0而當作T C P/ I P封包被發 送,則關閉此功能。 RAM介面9 4爲介於接收路徑8 0和一外部 RAM4 6之間的介面,外部RAM4 6可包括雙埠 S D RAM,此外部RAM4 6被用來做爲數個儲存輸入 資料串的串緩衝>器4 8,每一個串緩衝器4 8處理一個輸 入資料串,輸入資料串能夠具有不可預測之性質,例如, 有些資料封包可以是非常叢發的,這意謂所需之緩衝器容 量隨串而改變並且時常改變。因此,在外部RAM4 6中 ,最好是動態緩‘衝器配置。 一 R X位址轉換器9 6將適當的讀取位址提供給串緩 衝器4 8。 以下將詳細說明串式處理器中的主要方塊。 〔T X位址轉換器〕 (請先閱讀背面之注意事項再填寫本頁) t 訂-丨 •線丨-- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -20- 508504 Α7 Β7 五、發明說明(18 ) 輸出資料串係以叢發傳輸方式在p c 1匯流排2 4上 從儲存裝置2 〇發送到串式處理器3 6 ,T X位址轉換器 5 4之目的在於將一連續DMA叢發散播在串緩衝器4 4 之適當區域內。 圖3顯示出T X位址轉換器5 4之方塊圖。在來自儲 存裝置2 0 —個連續D Μ A叢發到達之前,正確的開始位 址透過儲存裝置控制器2 2而被寫入暫存器1 0 2中。暫 存器1 0 2的內容被用做串緩衝器4 4之寫入位址。一計 數器1 06計算來自PC I F I F〇52之輸出資料串 之位元的數目。每一次由3 2位元匠組成之資料字元通過 計數器1 0 6 ,其通知增量控制器1 〇 4 —個字元已經被 傳送到串緩衝器4 4。隨著每一個新增的字元,增量控制 器1 0 4以位址—增量(ADDRESS —INCREMENT )來增加 暫存器10 2之內容,而位址_增量爲一可編程之値。若 輸出資料串爲R I D處理資料,則位址—增量之値基本 上係根據用於R A I D系統之磁碟的數目來設定。若輸出 資料串爲一 T C P / I P封包的酬載,則位址_增量之値 基本上係根據封包化參數來設定。 現在參照圖4來說明當輸出資料串爲R A I D處理資 料時位址的轉換。在此例中,R A I D或S D I系統係由 四個磁碟所組成,即磁碟0、磁碟1、磁碟2和磁碟3。 磁碟0含有即將被發送到地方A T Μ轉換器1 4之字元1 ,4,7 .........。磁碟1亦容納要發送到地方A Τ Μ轉換 器1 4之字元2,5,8 .........,磁碟2亦含有即將被發 本纸張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 丨丨--------1 (請先閱讀背面之注意事項再填寫本頁) 訂--- 線_· 經濟部智慧財產局員工消費合作社印製 -21 - 508504 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明說明(19 ) 送到地方A T Μ轉換器1 4之字元3 ,6 ’ 9……·…’磁 碟3含有用於修正錯誤之奇偶性字元元0 ,1 ,2 ......... 。每一個奇偶性字元(例如奇偶性〇 )已經在 R X RAID方塊90中由三個字元(例如字元1 ,2 和3 )所產生,此三個字元和奇偶性字元一起構成所謂的 R A I D 之帶狀單元(stripe unit )。 若其中一個磁碟(例如磁碟2 )損壞,包含奇偶性字 元之連續D Μ A叢發被傳送到T X位址轉換器5 4,爲了 方便說明,假設一連續DMA叢發之尺寸爲9 6個位元組 (24個字元),雖然實際的尺寸可以比100K個位元 組還大(視硬體及/或硬體之速率而定)。在此情況中’ 連續DAM叢發12 6包括來自磁碟〇之字元1 ,4 ’ 7 ,1 0,1 3,1 6、自來磁碟1之字元2,5,8, 1 1 ,1 4,1 7、來自磁碟2之字元3 ,6 ,9 ,1 2 ,1 5,1 8以~及來自磁碟3奇偶性字元0,1 ,2 ’ 3 ,4,5,T X位址轉換器5 4產生下面序列之位址: 178,182,186,190,194,198 (來自磁碟0之資料) 179,183,187,191,195,199 (來自磁碟1之資料) 180,184,188,192,196,2〇0 (來自磁碟2之資料). 181,185,189,193,‘197,2〇4 (來自磁碟3之資料) 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) --11.--I----I · 111----訂-------I--丨 (請先閱讀背面之注意事項再填寫本頁) -22 - 508504 A7 B7 五、發明說明(20 ) 更特別的是,在連續Μ A叢發1 2 0到達串緩衝器 4 4之前,一數値1 7 8被儲存在暫存器1 0 2中做爲起 始位址。之後,來自磁碟0之字元1被寫入串緩衝器4 4 之位址1 7 8中,當字元1通過計數器1 〇 6 ,增量控制 器1 0 4以具有對應於磁碟數目之數値的位址_增量來使 暫存器1 0 2中的數値1 7 8增加。之後,來自磁碟〇之 字元4被寫入串緩衝器4 4的位址1 8 2中,當字元4通 過計數器1 0 6 ,增量控制器1 0 4以値爲4之位址_增 童使暫存器1 0 2之値1 8 2增加。之後,來自磁碟〇之 字元7被寫入串緩衝器4 4之位址1 8 6中。同樣’地,來 自磁碟0剩餘的字元10、 13和16被寫入在串緩衝器 4 4中相隔磁碟數目之位址1 9 0、 1 9 4、 1 9 8中。 當來自磁碟0之字元16通過計數器1〇6,增量控 制器1 0 4以値爲1 9之位址_增量來使暫存器1 〇 2之 値1 9 8增加。之後,來自磁碟1之字元2被寫入串緩衝 器44之位址179中。當字元2通過計數器106,增 量控制器1 0 4以値爲4之位址_增量來使暫存器1 〇 2 中之値179增量。之後,來自磁碟1之字元5被寫入串 糸麦衝益4 4之位址1 8 3。當子兀5通過§十數器1 0 6 , 增量控制器1 0 4以値爲4之位址_增量來使暫存器 1 0 2中之値1 8 3增加。之後,來自磁碟1之字元8被 寫入串緩衝器4 4之位址1 8 7。同樣地,來自磁碟1剩 餘的字元被寫入串緩衝器4 4中相隔磁碟數目之位址,.. 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) ----------t (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 一口年 I tmt- fmmmm m I—' VI emmmmm iiflf I ϋ an am UK i i tmmt an n I n n —i n -23- 508504 A7 _______ B7 五、發明說明(21 ) 191,195,199 中。 依此方式,來自磁碟2和3之字元被寫入串緩衝器 4 4中之適當位址中。被寫入串緩衝器4 4中之字元以線 性方式來讀取,並且提供給T X R A I D方塊6 0來修 正錯誤。 當來自儲存裝置2 0之輸出資料串爲TCP/I P酬 載’位址轉換器5 4和T C P檢驗和計算方塊一起工作來 支援TCP/IP封包產生。主機28預先編程Tx位址 轉換器5 4,使得資料根據指定的封包尺寸來分配。首先 ,主機2 8需要知道所有的封包化參數,對此操作而言爲 重要的參數有TCP酬載尺寸、TCP標題尺寸、IP標 題尺寸和I P酬載尺寸,TCP標題和I P標題基本上具 有用於選擇性資料的空間,但是實際上並未使用。因此可 藉由假設用於標題的內定尺寸來引進簡化:T C P標題尺 寸爲5個字元(\ 0個位元組),且I P標題尺寸爲5個 字元(2 0個位元組)。 以下說明本機構。 主機2 8本身對T C P / I P標題之假性標題實施部 分的檢驗和計算,之後,以此値爲該T C P / I P封包初 始化T C P檢驗和方塊5 6中之T C P檢驗和暫存器5 7 。用於串緩衝器4 4之空間將保留在外部R A Μ 4 2中, 以便配合全部T C Ρ封包加上T C Ρ和I Ρ標題附加位元 〇 之後,主機2 8將會指示Τ X位址轉換器5 4中之增 本纸張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) --------t (請先閱讀背面之注意事項再填寫本頁) 訂i 線丨· 經濟部智慧財產局員工消費合作社印製 -24- 508504 A7 B7 五、發明說明(22 ) 量控制器104TCP酬載尺寸、TCP標題尺寸、IP 標題尺寸和I p酬載尺寸。之後’ T c p酬載可以在· P C I匯流排2 4上被送出當做一連續DMA叢發,並且 被放置入藉由T X位址轉換器來保留給他之串緩衝器4 4 中的區域內,且留下空間給標題。當他從P C I匯流排 2 4走到串緩衝器4 4時,檢驗和計算方塊5 6更新 T C P檢驗和暫存器5 7中之部分的檢驗和。請注意到有 了此方法,酬載(通常代表TCP/I P封包之大部分) 不需要爲了他而先從儲存裝置2 0複製到主記憶體3 2, 而後再複製到串緩衝器4 4。此舉爲主C P U 3 0節省很 有價値的匯流排頻寬和附加字元。在酬載已被寫入之後, 由主機2 8所準備之標題資訊透過位址編繹器5 4而被送 到串緩衝器4 4。如同酬載的情況,T X位址轉換器5 4 將標題放置在先前所保留的記憶體位置中。 此序列可以~被保留,藉此先寫入標題資訊再寫入酬載 〇 在任何一個狀況下,當標題和酬載已被寫入時, T C P檢驗和將會完整並且自動地複製到正確位置。 此種機構亦可用來有效地支援將一 T C P封包分段成 爲多個較小的I P封包,在此情況中,爲每一個I P封包 保留空間。T C P封包資料(標題+酬載)被分成這些封 包,並且每一個I P封包的標題由主機2 8來寫入。 除了最後一個方塊以外,所有的I P封包具有同樣的 尺寸,最後一個方塊之尺寸可能具有和其他方塊不同的尺 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -------------餐 (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 一 I n n n n n ·ϋ n I n n n n n an It n -ϋ ϋ n Hi If n ϋ t— ·ϋ n n n n n _ -25- 508504 A7 B7 五、發明說明(23) 寸,位址轉換器5 4將此列入考慮。在完整的T C P / I P封包已經形成之後,已經完成發送之準備。 圖5A、 5B和5C顯示用於TCP/IP封包化之 位址轉換的示例,在此情況中,在被當作一連續D M A叢 發1 30發送之TCP/I P酬載到達串緩衝器44之前 ,一數値3 1 〇被儲存在暫存器1 0 2中做爲起始寫入位 址,之後,第一資料的第一個字元被寫入串緩衝器4 4之 位址3 1 0中。當第一資料的第一個字元通過計數器 1 0 6時,增量控制器1 0 4以値爲1之位址_增量來使 暫存器1 Q 2中之値3 1 0增加。之後第一資料的第二個 字元被寫入串緩衝器4 4的位址3 1 1中。當第一資料的 第二個字元通過計數器1 〇 6時,增量控制器1 0 4以値 爲1之位址_增量來使暫存器1 〇 2中之値3 1 1增加。 之後’第一資料的第三個字元被寫入串緩衝器4 4的位址 3 1 2中,位址_增量値爲1之增量被重覆許多次,其對 應於I P酬載尺寸。因此,T C P / I P酬載的第一個資 料被寫入一適當區域之中。 之後,增量控制器1 0 4値對應I P標題尺寸之位址 _增量來使暫存器1 〇 2之內容增加。之後,第二資料開 始被寫入,從根據暫存器1 〇 2之內容的位址開始。因此 ,位址轉換器5 4爲酬載產生寫入位址,使得爲標題留下 空間。最後一個資料尺寸可能具有與其他資料不同的尺寸 ’最後一個資料之尺寸係藉由下面的表示式而被計算於增 量控制器1 0 4中: 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) ----------t (請先閱讀背面之注意事項再填寫本頁) 訂-------- 線丨-- 經濟部智慧財產局員工消費合作社印製 -26- 508504I! III line 丨 · Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 508504 Α7-Β7 V. Description of the invention (15) are in the string buffer 44, then the checksum is placed in the correct position of the TCP / IP header Within, and the packet is ready to be sent. The RA Μ interface 5 8 constitutes an interface between the external RA Μ 4 2 and the transmission path 50. The external RAM 4 2 may include dual-port SDRAM (Synchronous Dynamic Random Access Memory). The external RA Μ 4 2 includes several strings. The buffer 44 is used to separate the burst data communication from the magnetic disk of the storage device 20 and provide the required fixed bit rate data string to the ATM network 16. Each string buffer processes an output data string. In comparison with the input direction, since the data flow characteristics in the output direction are completely predictable (controlled), the buffer requirements can be estimated in advance. Therefore, the string buffer 4 4 is statically arranged in the external RAM 4 2 0 TX RAID or SDI (string data overall) block 60 to provide data redundancy support. The TX address converter 54 places the required data in the string buffer 4, and then when the data is output from the string buffer 44, if one of the disks in the storage device 20 is damaged, (J T XRAID block 60 0 to correct wrong data. The traffic shaper 6 2 controls the output data string from the string buffer 44 to A TM network 16 and it is designed for very accurate speed and low CDV (Unit delay variation). The traffic shaper 62 is composed of two main segments, one of which processes high-priority data (such as video traffic) and the other segment processes low-priority general data. The amount of communication. Command box 6 6 Want to remove the load of the host 2 8 which is sensitive in real time, and execute the content of the accurately known position in the outgoing data string. · -------- t (Please read the notes on the back before filling out this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs --------- line 丨 ----- 5 〇85〇4 A7 B7 V. Description of the invention (16) Action triggered by transmission. Segmented block 70 will The string buffer 44 provides segmentation of the output data string entering the AAL-5 PDUs (ATM Adaptation Layer-5 Protocol Data Unit), and maps the AAL-5 PDU s to the AT unit. If the output data string is MPEG —2 SPTS (Single Program Transport String), segmentation block 70 can segment two ts packets of mpeg-2 spts into one AAL-5 PDU, unless there are less than two TS packets left in MPEG-2 SPTS In the latter case, 'AAL-5 PDU corresponds to eight ATM units. In general, AAL-5 segmentation is controlled by PDU size' and the PDU size in each string is programmable The receiving path 80 has several blocks corresponding to the reverse operation of the blocks of the transmitting path 50. —V P I / \ c I (virtual path identifier / virtual channel identifier) filtering block 8 4 pairs of input A T The MU unit implements fast and efficient VP1 / VCI filtering, which is accomplished by combining clutter and linear search functions on the input in the VPι / vcI table. A reorganization block 8 6 basically implements a segmented block 7 0 Opposite Function 'Reorganize Block 8 6 Using ATM. Payload of Unit Reconstruct AAL-5 PDUs, and then map AAL-5 PDUs to the upper layer data (such as M PEG-2 SPTS, TCP / IP packets). If the input data string is sent through TCP, a TCP check that the paper size is applicable China National Standard (CNS) A4 Specification (210 x 297 mm) t (Please read the notes on the back before filling out this page) Order --------- line 丨 · Member of the Intellectual Property Bureau of the Ministry of Economic AffairsΗConsumer Cooperative Printed 508504 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (17) and check box 8 8 Check the TCP checksum in the TCP header. A pattern detector 92 allows a limited number of bit patterns to be detected in the input data string. A list is created that accurately indicates where the specified bit pattern is generated in the string. This supports certain processing tasks that can be performed quickly, otherwise they must be processed later to complete. —R X RA I D or SD I box 9 0 adds redundancy to the input data string. If a sequence of N characters is written into a buffer (not shown), the parity of the N characters is then written, and this function can be turned on / off. If the input data string will be stored in the storage device 20 'and later sent as a T C P / IP packet through the transmission path 50, then this function is turned off. The RAM interface 9 4 is an interface between the receiving path 80 and an external RAM 4 6. The external RAM 4 6 may include a dual-port SD RAM. The external RAM 4 6 is used as a string buffer for storing input data strings. ; Device 48, each string buffer 48 processing an input data string, the input data string can have unpredictable properties, for example, some data packets can be very bursty, which means that the required buffer capacity varies with Change from time to time and from time to time. Therefore, in the external RAM 46, it is best to use a dynamic buffer configuration. An R X address converter 9 6 provides the appropriate read address to the string buffer 4 8. The main blocks in the serial processor will be described in detail below. 〔TX Address Converter〕 (Please read the precautions on the back before filling this page) t Order-丨 • Line 丨-This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -20 -508504 Α7 Β7 V. Description of the invention (18) The output data string is sent from the storage device 2 to the serial processor 3 6 on the pc 1 bus 2 4 in a burst transmission mode, and the TX address converter 5 4 of The purpose is to spread a continuous DMA cluster in the appropriate area of the string buffer 4 4. Figure 3 shows a block diagram of the TX address converter 54. Before a serial D M A burst from the storage device 20 arrives, the correct start address is written into the register 102 through the storage device controller 2 2. The contents of the register 102 are used as the write address of the string buffer 44. A counter 106 calculates the number of bits in the output data string from PC I F I F52. Each time a 32-bit data character passes the counter 106, which informs the incremental controller 104 that characters have been transferred to the string buffer 44. With each new character, the increment controller 104 adds the content of the register 102 with address-increment (ADDRESS-INCREMENT), and the address_increment is a programmable one. . If the output data string is R ID processing data, then the address-increment range is basically set according to the number of disks used in the R ID system. If the output data string is the payload of a TCP / IP packet, the address_incremental 値 is basically set according to the packetization parameters. Referring now to FIG. 4, description will be given of address conversion when the output data string is processed by R A ID. In this example, the R A I D or S D I system is composed of four disks, namely disk 0, disk 1, disk 2 and disk 3. Disk 0 contains characters 1, 4, 7, ... which will be sent to the local AT / M converter 14. Disk 1 also contains the characters 2, 5, 8 ... to be sent to the local ATM converter 1 4. Disk 2 also contains the paper size to be issued. The Chinese standard is applicable. (CNS) A4 specification (210 X 297 mm) 丨 丨 -------- 1 (Please read the precautions on the back before filling out this page) Order --- Line _ · Staff Consumption of Intellectual Property Bureau, Ministry of Economic Affairs Printed by the cooperative -21-508504 Printed by the consumer property cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention (19) Send to the local AT Μ converter 1 zigzag 3, 6 '9 ... ·' magnetic disk 3 contains the parity characters 0, 1, 2 ......... for error correction. Each parity character (for example, parity 0) has been generated in the RX RAID block 90 by three characters (for example, characters 1, 2, and 3). These three characters and the parity character together form a so-called Stripe unit of the RAID. If one of the disks (for example, disk 2) is damaged, a continuous DMA burst containing parity characters is transmitted to the TX address converter 5 4. For the sake of explanation, it is assumed that the size of a continuous DMA burst is 9 6 bytes (24 characters), although the actual size can be larger than 100K bytes (depending on the hardware and / or the speed of the hardware). In this case, 'continuous DAM bursts 12 6 include characters 1, 4' 7, 10, 1 3, 1 6 from disk 0, characters 2, 5, 8, 1 1 from disk 1 , 1, 4, 17, 7, characters 3, 6, 9, 1, 12, 15, 18 from disk 2 and ~, and parity characters 0, 1, 2, 3 ', 4, 5 from disk 3 The TX address converter 54 generates the following sequence of addresses: 178, 182, 186, 190, 194, 198 (data from disk 0) 179, 183, 187, 191, 195, 199 (from disk 1) Information) 180, 184, 188, 192, 196, 200 (data from disk 2). 181, 185, 189, 193, '197, 204 (data from disk 3) Paper size Applicable to China National Standard (CNS) A4 specification (210 X 297 mm) --11 .-- I ---- I · 111 ---- Order ------- I-- 丨 (Please read first Note on the back, please fill out this page again) -22-508504 A7 B7 V. Description of the invention (20) More specifically, before the continuous BM bursts 1 2 0 reach the string buffer 4 4, a number of 1 7 8 It is stored in register 102 as the starting address. After that, the character 1 from the disk 0 is written into the address 1 78 of the string buffer 4 4. When the character 1 passes the counter 1 0 6, the increment controller 10 4 has a number corresponding to the number of disks. The number of addresses is incremented to increase the number 1 7 8 in the register 102. After that, the character 4 from the magnetic disk 0 is written into the address 1 8 2 of the string buffer 4 4. When the character 4 passes the counter 10 6, the increment controller 10 4 uses the address 4 as the address. _Zeng Tong increases the register 1 2 2 1 8 2. After that, the character 7 from the magnetic disk 0 is written to the address 1 186 of the string buffer 44. Similarly, the remaining characters 10, 13 and 16 from the disk 0 are written in the addresses 19, 19, 19 and 19 of the number of separated disks in the string buffer 44. When the character 16 from the disk 0 passes the counter 106, the increment controller 10 4 increments the address _19 of the register 1 102 by incrementing the address_increment of 値. After that, the character 2 from the magnetic disk 1 is written into the address 179 of the string buffer 44. When the character 2 passes the counter 106, the increment controller 104 increments the address 179 in register 1 102 by incrementing the address_increment of 4. After that, the character 5 from the magnetic disk 1 is written into the address 1 8 3 of the Omai Chongyi 4 4. When the sub-frame 5 passes through the decimal number 10 6, the increment controller 10 4 increments the address 1 8 3 in the register 1 2 with the address_increment of 4. After that, the character 8 from the disk 1 is written to the address 1 8 7 of the string buffer 44. Similarly, the remaining characters from disk 1 are written to the address of the number of separated disks in the string buffer 4 4 .. This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm)- --------- t (Please read the notes on the back before filling out this page) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economy I tmt- fmmmm m I— 'VI emmmmm iiflf I ϋ an am UK ii tmmt an n I nn —in -23- 508504 A7 _______ B7 V. Description of the invention (21) 191, 195, 199. In this way, characters from disks 2 and 3 are written to the appropriate addresses in the string buffer 44. The characters written in the string buffer 44 are read linearly and provided to the TX R A I D block 60 to correct the error. When the output data string from the storage device 20 is a TCP / IP payload, the address converter 54 and the TCP check and calculation block work together to support TCP / IP packet generation. The host 28 pre-programs the Tx address converter 54 so that the data is allocated according to the specified packet size. First, the host 28 needs to know all the packetization parameters. The parameters that are important for this operation are TCP payload size, TCP header size, IP header size, and IP payload size. TCP headers and IP headers are basically useful. Space for selective data, but it is not actually used. Therefore, simplification can be introduced by assuming an internal size for the title: the T C P title size is 5 characters (\ 0 bytes), and the IP title size is 5 characters (20 bytes). This organization is explained below. The host 2 8 performs part of the inspection and calculation of the false title of the T C P / IP header, and then uses this to initialize the T C P inspection and the T C P inspection and register 5 7 in the block 56. The space used for the string buffer 44 will be reserved in the external RAM 42 to match all TCP packets plus the additional bits of the TCP and IP headers. After that, the host 28 will instruct TX address conversion. The paper size of the paper in the printer 5 4 is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -------- t (Please read the precautions on the back before filling this page) Order i Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs -24-508504 A7 B7 V. Description of the invention (22) The quantity controller 104 TCP payload size, TCP header size, IP header size and IP payload size. After that, the TCP payload can be sent out on the PCI bus 2 4 as a continuous DMA burst, and placed in the area in the string buffer 4 4 reserved for him by the TX address converter. And leave space for the title. As he walks from the P C I bus 24 to the string buffer 44, the check and calculation block 5 6 updates the check sum of the part of the T C P check and register 5 7. Please note that with this method, the payload (usually representing a large part of the TCP / IP packet) does not need to be copied from the storage device 20 to the main memory 3 2 for him and then to the string buffer 44. This saves valuable bus bandwidth and additional characters for the main C P U 3 0. After the payload has been written, the header information prepared by the host 28 is sent to the string buffer 44 through the address compiler 54. As is the case with the payload, the TX address converter 5 4 places the title in the previously reserved memory location. This sequence can be retained, so that the header information is written first and then the payload. In any case, when the header and payload have been written, the TCP checksum will be completely and automatically copied to the correct location. . This mechanism can also be used to effectively support the segmentation of a TCP packet into multiple smaller IP packets, in which case space is reserved for each IP packet. The TCP packet information (title + payload) is divided into these packets, and the header of each IP packet is written by the host 28. Except for the last block, all IP packets have the same size. The size of the last block may be different from the size of other blocks. Paper size applies to China National Standard (CNS) A4 (210 X 297 mm) --- ---------- Meal (Please read the notes on the back before filling this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs I nnnnn · ϋ n I nnnnn an It n -ϋ ϋ n Hi If n ϋ t— · ϋ nnnnn _ -25- 508504 A7 B7 V. Description of the invention (23) inch, the address converter 5 4 takes this into consideration. After the complete TCP / IP packet has been formed, the preparation for transmission has been completed. 5A, 5B, and 5C show examples of address translation for TCP / IP packetization. In this case, before the TCP / IP payload sent as a continuous DMA burst 1 30 reaches the string buffer 44 , A number 値 3 1 〇 is stored in the temporary register 1 02 as the initial write address, after that, the first character of the first data is written to the address 3 1 of the string buffer 4 4 0 in. When the first character of the first data passes the counter 106, the increment controller 1 0 increments the address_increment of 1 to 値 3 1 0 in the register 1 Q 2. After that, the second character of the first data is written into the address 3 1 1 of the string buffer 44. When the second character of the first data passes the counter 106, the increment controller 104 increments the address_increment of 値 to 1 to increase 値 3 1 1 in the register 1 02. After that, the third character of the 'first data' is written to the address 3 1 2 of the string buffer 4 4. The increment of address_increment 値 is repeated many times, which corresponds to the IP payload. size. Therefore, the first data of the T C P / IP payload is written into an appropriate area. After that, the increment controller 10 4 値 corresponds to the address of the IP title size_increment to increase the content of the register 102. After that, the second data starts to be written, starting from the address based on the contents of the register 102. Therefore, the address converter 54 generates a write address for the payload, leaving space for the title. The last data size may have a different size from the other data. The size of the last data is calculated in the incremental controller 104 using the following expression: This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) ---------- t (Please read the notes on the back before filling out this page) Order -------- Line 丨-Staff of Intellectual Property Bureau, Ministry of Economic Affairs Printed by Consumer Cooperatives -26- 508504
發明說明(24 ) 最後的資料尺寸.= TCP酬載mod IP酬載尺寸 因此’增量的數目之控制係將最後的資料尺寸列A考 慮。依此方式’當做一連續D Μ A叢發所送出之酬載係散 播在串緩衝器4 4中的陰影區(見圖5 )。 接著,當TCP標題1 3 2當做一連續叢發而在 P C I匯流排2 4上被送出時,位址轉換器5 4爲T C p 標題而在串緩衝器4 4中產生對應於先前所保留之記憶體 位置的寫入位址。 更特別的是,在被當做一連續叢所發送出之T C P標 題到達串緩衝器4 4之前,一數値3 0 5在暫存器1 〇 2 中被設定做爲起始寫入位址,之後T C P標題的第一個字 元被寫入串緩衝器4 4的位址3 0 5中。當TC P標題的 第一個字元通過^數器1 0 6時,增量控制器1 0 4以値 爲1之位址_增量來使暫存器1 0 2中之値3 0 5增加。 之後,T C P標題的第二個字元被寫入串緩衝器4 4的位 址3 0 6中。當T C P標題的第二個字元通過計數器 1 0 6,增量控制器1 〇 4以値爲1之位址—增量來使暫 存器102中之値306增量。之後,TCP標題的第三 個字元被寫入串緩衝器4 4的位址3 0 7。位址—增量値 爲1之增量被重覆許多次,其對應於TC p標題尺寸。因 此,TC P / I P標題被寫入在圖5 B中所示之串緩衝器 4 4的陰影區。 ---------t (請先閱讀背面之注意事項再填寫本頁) 訂i 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準(CNaA4規格GW X a97公爱) -27- 508504 A7 B7 五、發明說明(25) 接著,當I P標題1 3 4當做一連續叢發而在pc 1 匯流排2 4上被送出時,位址轉換器5 4爲I P標題而在 串緩衝器4 4中產生對應於先前所保留之記憶體位置的位 址。 更特別的是,在被當做一連續叢發1 3 4所送出之 .I P標題到達串緩衝器4 4之前,一數値3 0 0在暫存器 1 0 2中被設定做爲起始寫入位址,之後第一個I P標題 的第一個字元被寫入串緩衝器4 4的位址3 〇 0中。當第 一 I P標題的第一個字元通過計數器1 0 6時,增量控制 器1 0 4以値爲1之位址_增量來使暫存器1 〇 2中之値 3 0 0增加。之後,第一個I P標題之第二個字元被寫入 串緩衝器4 4的位址3 0 1中。當第一 I P標題的第二個 字元通過計數器1 0 6時,增量控制器1 0 4以値爲1之 位址_增量來使暫存器1 0 2中之値3 0 1增量。之後, 第一 I P標題的k三個字元被寫入串緩衝器4 4的位址 3 0 2中,位址_增量値爲1之增量被重覆多次對應於 I P標題尺寸。 之後,增量控制器1 0 4以之位址_增量値其使暫存 器1 0 2之內容增加對應於TCP標題尺寸加上I P酬載 尺寸。之後,根據暫存器1 0 2之內容位址開始寫入第二 I P標題。因此,I P標題被寫入在圖5 C中所示之串緩 衝器4 4的陰影區。 接著,由TCP檢驗和方塊所5 6完成之TCP檢驗 和被複製到正確位置。 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -- -------t (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 訂-—---I-------------- -28 - 508504 A7 B7 ___ 五、發明說明(26) 依此方式,T C P / I P封包化被完成’並且可以用 線性方式從串緩衝器4 4中讀出。 在上述實施例中提及TCP/I P封包化’然而也可 能使用U D P (使用者資料小包協定)而不是Τ C P ’在 此情況中,U D P標題之內定値爲2個字元(8位元組) 〇 此外,在上述的實施例中’ τ C P標題和I P標題係 當做不同的叢發而從主機2 8送到τ X位址轉換器5 4 ’ 但是,.有可能將T C P標題和I P標題一起當做一連續叢 發而從主機2 8送到Τ X位址轉換器5 4。 [Τ X RAID 或 SDI 方塊〕 在串緩衝器4 4的字元序列中可插入奇偶性字元。此 冗餘提供一種用來修正錯誤的機構。Τ X R A I D或 5 D I方塊6 0收容一序列N+ 1個字元,其最後一個字 元是N個第一個字兀的奇偶性。右其係由硬體及/或軟體 來指示該字元m係訛誤的,例如由於磁碟損壞,則從儲存 裝置2 0檢索奇偶性字元並被用來重建該字元Μ。 例如在圖4的情況中,來自損壞的磁碟2之字元3 ’ 6 ,9 ,12,15,18 (在輸入資料142中)包括 圖6所示之錯誤,Tx RAID方塊60利用字元1 ’ 2和奇偶性字元0來重建字元3,Τ X R A .1 D方塊 6 0利用字元4,5和奇偶性字元1來重建字元6。同樣 地,Tx RAID方塊60重建3字元9, 12’15 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) ——-------Φ (請先閱讀背面之注意事項再填寫本頁) tr------—. 經濟部智慧財產局員工消費合作社印製 -29- 經濟部智慧財產局員工消費合作社印製 508504 A7 B7 五、發明說明(27 ) ,18。因此,Tx RAID方塊60執行錯誤修正, 並且毫無錯誤地輸出字元1,2,3,4之序列142。 RA I D功能可,以藉由命令方塊6 6來開啓/關閉。 〔通信量整形器〕 通信量整形器6 2係由二主要方塊段所組成,其中一 個區段處理高優先權資料,例如視頻通信量,而低優先權 區段則處理一般資料通信量。 高優先權方塊段被組織成數個通信量類型,其中一類 型爲具有相同位元率特性之一或多個串的一群組。例如, 以2Mpb s之CBR (固定位元率)的所有的串屬於同 一類型。可變位元率(V B R )型式之類型典型上僅包括 一個串,因爲不可能二可變位元率串不論什麼時候均具有 相同的頻寬型樣。每一類型有一單一組的發送參數,用以 控制位元率、提供低單元延遲變異(C D V )和精確的調 速’類型的數目係可編程的,但限制在最大値1 2 8。 每一類型有二主要的發送參數,一個是理想的排定時 間(T S )和一個是排定時間(T S )的增量(△)。其 基本機構爲當T S等於或少於參考時鐘時,一串指標被放 進發送佇列中,同時,該値T S增加數値△,發送佇列爲 一先進先出佇列,其會儘快將由該串指標所指示之串提呈 到 A T M f i f 7 2 中。 在高優先權段中,下述機構達成高準確的位元率及低 C D V 。 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) • I ί I I I n n n n n I · n n I— n n n n 一口、fl n n an (請先閱讀背面之注意事項再填寫本頁) 線丨· -30- 508504 Α7 Β7 五、發明說明(28) . — lit (請先閱讀背面之注意事項再填寫本頁) 由於參考時鐘的有限解析度,具有單一△値通常不能 給予所需之精確度。爲了達到所需之精確度,兩個△値被 交替地使用,此二値產生稍微高於及稍微低於所需之位元 率的速率。使用不同數目的單元之各個△値來補償受限的 時鐘解析度,並且可提供任意的精確度。△ Η和△ L (其中 △ L = △ Η + 1 )代衣兩個不同的增量値。Ν Η和N L參數代 表單元的數目,其對應的增量値係交替有效的。藉由此機 構,串被調變,藉此’平均位元率趨近在所需之精確度內 所需的位元率。圖7顯示藉由此機構所達成之位元率的情 形。在圖7中,Ν Η單元以△ Η來送出,而NU單元以/^來 送出。此順序被周期性地重覆。因此,由虛線所表示之平 均位元率被保持爲一長期的位元率。 經濟部智慧財產局員工消費合作社印製 低C D V之達成係藉由從不同的串中減少單元之規定 時間內的碰撞。在許多現有的通信量整形機構,碰撞的主 因在於位元率相同的串被排定在相同的時間,當串的數目 很大且獨立之位元率的數目少時’此一問題尤其嚴重。在 較佳實施例中,藉由使屬於相同類型之串的單元均勻間隔 來強調上一問題。換言之,若一串之增量應爲△,則該類 型之增量爲Α/η ,其中η爲該類型之串的數目。每次當 某一類型要被服務,資料從連續串中取出。例如,若屬於 類型0之串0的單元應該增加△,則屬於同一類型0之串 的各個單元(即串0至串η — 1 )在間隔爲△/ η的情況 下被送出’如圖8所示。 藉由上述一機構之組合’可以達成筒精確的位兀率和 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -31 - 經濟部智慧財產局員工消費合作社印製 508504 Α7 _____________ Β7 五、發明說明(29 ) 低C D v。若發送佇列不阻塞,則這些單元的傳送方式係 如圖9中所示。 高優先權區段亦.< 處理V B R通信量。通信量整形器 6 2支援發送參數之平順更新,此更新可由主機2 8,也 可由命令方塊6 6來完成。命令方塊6 6係由主機2 8來 編程,並且當一精確位置從串緩衝器4 4發送出時,其動 作即被觸發。其中一個如此之動作係取代用於通信量整形 器6 2中之一指定串的發送參數。一旦在位元率剛改變之 前送出資料,命令方塊6 6馬上更新參數。一旦其準備就 緒,此程序係自主操縱的,並且不需要與主機C P u 3 0 有任何互動。結果,主機2 8不需要準確地應該被要求互 動的時候互動。依此方式,串之即時特性得以保持,並且 主機負載保持最小。 低優先權方塊段被組織成例如固定3 2的通信量類型 ,其中一個類型係具有相同的P CR (尖峰單元速率)的 一個或多個串所組成的一群組。就一般的資料通信量來說 ,即時限制非常不顯著。低優先權區段的通信量整形之主 要目的在於限制P C R以便避免網路整頓。資料封包之通 信量整形係藉由使用一理想的排定時間(T S )和T S之 增量(△)的機構來實施,其與高優先權區段中之基本通 β量整形機構類似。然而,資料封包排程之優先權比即時 通信量還低,只有高優先權區段之發送佇列是空的,/資料 封包的串才能夠被送到A T M F I F〇7 2。 此機構係以如圖1 0所示之架構來實施,通信量整形 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) m In n m I— -1··— ϋ al tmaMm emmmmm mmmmmmm Is · n m n ϋ is n an I m fn n* (請先閱讀背面之注意事項再填寫本頁) •線丨·! -32- 508504 A7 B7 五、發明說明(30 ) 器6 2包括先前所述之高優先權區段2 〇 〇和低優先權區 段 2 0 2。 在高優先權區段2 0 0中,一記憶體2 0 3儲存一組 提供自主機2 8之用於各個等級之新的發送參數各組之新 的發送參數包括T s 1、△H i、△ L i、Ν H i、N L i和 P t 1 (其中丨si 27)。在此實施例中,P t i包括 一個或多個串指標,其指示附在類型i之後的一或多個串 ’記憶體2 0 6儲存目前的發送參數。當主機2 8或命令 方塊6 6指示一命令時,一更新邏輯2 0 4被命令所觸發 ’藉此記憶體2 0 6內之目前的發送參數更新爲記憶體 2 0 3內之新的發送參數。暫存器2 1 2儲存示來自其接 收處之主機2 8之類型的數目之參數Nr_classes,通信量邏 輯2 0 8爲各類型(從〇到N^classesj )檢查T S 1是否 等於或小於由參考時鐘2 1 0所指示之目前的時間。如果 是的話’附在此k型1之後的第一串之串指標被插入一高 優先權發送佇列2 1 6之中,並且通信量整形器2 0 8使 記憶體206中的TSi增加此類型i之AHi或· 此△ η 1或△ L 1係根據Ν η i或N L i來輪替,之後,分段 方塊7 0接收來自高優先權發送佇列2 1 6之串指標,並 且將由串指標所指示之串的A TM單元放置入 ATM FIFO 72 中。 在低優先權區段2 0 2中,記憶體2 1 8儲存一組提 供自主機2 8之用於各個等級的發送參數。在本實施例中 ,每組發送參數包括T S j、、△ j、和P t j (其中0 本紙張尺度適用中國國家標準(CNS〉A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) 訂-------- •線 -- 經濟部智慧財產局員工消費合作社印製 -33- 508504 A7 B7 五、發明說明(31 ) ^ J ^ 3 1 ) 。P t 1包括一或多個串指標,其指示附在 類型j之後的一或多個串,通信邏輯2 2 0爲各個等級( 從0到3 1 )檢查T S j是否等於或小於由參考時鐘所表 示之目前的時間’並監控高優先權發送佇列2 1 6是否爲 空的。如果是的話,附在此類型j之後的第一串之串指標 被插入低優先權發送佇列2 2 2之中,並且通信量邏輯 2 2 0使記憶體2 1 8之T S j增加此類型j之△ j 。之 後,分段方塊7 0接收低優先權發送佇列2 2 2之串指標 ,並將屬於由串指標所指示之串流的A T Μ單元放置入 ATM F I F 〇 7 2 中。 在上述實施例中,與高優先權區段2 0 0之機構類似 的一個通信量整形機構被應用於低優先權區段2 0 2。然 而,傳統漏槽(leaky bucket )機構以被應用於低優先權區 段2 0 2之通信量整形機構。 〔命令方塊〕 本纸張尺度適用中國國家標準歸娜m〇 χ π公引善 (請先閱讀背面之注音?事項再填寫本頁) 一 .δ,ν * n n n Is n m I f n n 1_ I n n m n HB flu I n an n HI If m n n 定由在在數能主多 指。。。參可,許 中性動任式有置同 串體互責串,位連 料整之的的置之 , 資之時 6 整配作列 出串及 6 調態動表 輸持證塊要靜取的 在保保方需被採-6 生便遠令中般應 6 發以永命串述中塊 及,夠是料上標方 涉的能動資如指令 能即不互出係取命 可立,些輸 4 讀將 送是任這道 4 的時 輸須責行知器 4 及 料必多執 8 衝 4 刻 資作許 ,2 緩器時 時動的中機串衝當 即些 8 例主個緩適 時這 2 施,每串在 有,機實上於在 8 置主佳則由示 2 位於較原。表機 經濟部智慧財產局員工消費合作社印製 508504 A7 B7 五、發明說明(32 ) 指示載入。適當時刻爲介在將輸出資料串載入串緩衝器 4 4的時刻和輸出資料串從串緩衝器4 4送出之時刻之間 的時刻。命令方塊6 6掃瞄串緩衝器4 4的讀取指標,若 發現與一指定的位址匹配’則與該位址連結之命令將會被 執行,並且該命令將從命令方塊6 6被淸除。 命令方塊6 6觸發離開串緩衝器4 4的資料之位址, 當讀取一串緩衝器4 4,讀取指標以環繞式處理而被逐漸 增加。各串具有〃連結表列’其包括即將根據位址而被儲 存之(位址,命令)對’一位址爲表示檔案中位址的一對 (L’M),並且與實體位址無關。L爲方塊的數目,其 方塊尺寸等於串緩衝器4 4之尺寸,Μ爲最後一個方塊的 序列號碼’各串保持一個W A C (環繞計數器),其計算 讀取指標已經被環繞的次數。若以下關係式成立,則發現 有位址匹配:Description of the invention (24) The last data size. = TCP payload mod IP payload size. Therefore, the control of the number of increments is considered in the last data size A. In this way, the payload sent as a continuous DM A burst is scattered in the shaded area in the string buffer 44 (see Figure 5). Then, when the TCP header 1 3 2 is sent out as a continuous burst on the PCI bus 24, the address converter 5 4 generates a TC p header in the string buffer 4 4 corresponding to the previously reserved one. The write address of the memory location. More specifically, before the TCP header sent as a continuous cluster reaches the string buffer 44, a number 値 3 0 5 is set as the initial write address in the register 1 02. After that, the first character of the TCP header is written to the address 3 05 of the string buffer 44. When the first character of the TC P title passes through the counter 1 0 6, the increment controller 1 0 4 uses 値 as the address _ increment to make 値 3 0 5 in the register 1 0 2 increase. After that, the second character of the TCP header is written to the address 3306 of the string buffer 44. When the second character of the TCP header passes the counter 106, the increment controller 104 increments the address 306 in the register 102 with an address of 1 as increment. After that, the third character of the TCP header is written to the address 3 0 7 of the string buffer 44. The address-increment 値 increment is repeated many times, which corresponds to the TC p header size. Therefore, the TCP / IP title is written in the shaded area of the string buffer 44 shown in Fig. 5B. --------- t (Please read the notes on the back before filling this page) Order i Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs Employee Cooperatives This paper is printed in accordance with Chinese national standards (CNaA4 GW X a97) ) -27- 508504 A7 B7 V. Description of the invention (25) Next, when the IP header 1 3 4 is sent as a continuous burst and sent on the PC 1 bus 2 4, the address converter 5 4 is the IP header and An address corresponding to a previously reserved memory location is generated in the string buffer 44. More specifically, before being sent as a continuous burst of 1 3 4. The IP header reaches the string buffer 4 4, a number of 3 0 0 is set as the initial write in the register 1 2 After entering the address, the first character of the first IP header is written into the address 300 of the string buffer 44. When the first character of the first IP header passes the counter 106, the increment controller 104 increments the address_increment of 値 to increase 値 3 0 0 in the register 1 〇2 . After that, the second character of the first IP header is written to the address 3 01 of the string buffer 44. When the second character of the first IP header passes the counter 106, the increment controller 1 0 increments the address _ by 1 to increment 値 3 0 1 in the register 1 0 2 the amount. After that, the three k characters of the first IP header are written into the address 3 0 2 of the string buffer 44, and the increment of address_increment 値 is 1 is repeated multiple times corresponding to the IP header size. After that, the increment controller 104 increments the content of the temporary register 102 by its address_increment, which corresponds to the TCP header size plus the IP payload size. After that, the second IP title is written according to the content address of the register 102. Therefore, the IP title is written in the shaded area of the string buffer 44 shown in Fig. 5C. Then, the TCP checksum completed by the TCP checksum block 56 is copied to the correct location. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)-------- t (Please read the notes on the back before filling this page) Intellectual Property Bureau, Ministry of Economic Affairs, Consumer Consumption Cooperative Printing and making ------ I -------------- -28-508504 A7 B7 ___ V. Description of the invention (26) In this way, TCP / IP packetization is completed 'and can Read out from the string buffer 44 in a linear manner. In the above embodiment, TCP / IP packetization is mentioned. However, it is also possible to use UDP (User Data Packet Protocol) instead of TCP. In this case, the UDP header is set to 2 characters (8 bytes). ) 〇 In addition, in the above-mentioned embodiment, 'τ CP header and IP header are sent to the τ X address converter 5 4 from the host 28 as different bursts. However, it is possible to send TCP headers and IP headers. Together, it is sent as a continuous burst from the host 2 8 to the TX address converter 54. [TX X RAID or SDI Block] Parity characters can be inserted in the character sequence of the string buffer 44. This redundancy provides a mechanism for correcting errors. T X R A I D or 5 D I block 60 contains a sequence of N + 1 characters whose last character is the parity of the N first characters. Right, it is indicated by hardware and / or software that the character m is wrong. For example, due to a damaged disk, the parity character is retrieved from the storage device 20 and used to reconstruct the character M. For example, in the case of FIG. 4, the characters 3'6, 9, 12, 15, 18 (in the input data 142) from the damaged disk 2 include the error shown in FIG. 6, and the Tx RAID block 60 uses characters 1 '2 and parity character 0 to reconstruct character 3, T XRA .1 D block 60 0 uses character 4, 5 and parity character 1 to reconstruct character 6. Similarly, Tx RAID box 60 rebuilds 3 characters 9, 12'15 This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) ------------ Φ (Please read the back first Note: Please fill in this page again.) Tr ------—. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economics -29 , 18. Therefore, the Tx RAID block 60 performs error correction and outputs the sequence 142 of characters 1, 2, 3, 4 without error. The RA I D function can be turned on / off by command box 66. [Traffic Shaper] The traffic shaper 62 is composed of two main blocks, one of which deals with high-priority data, such as video traffic, and the low-priority sector deals with general data traffic. High priority block segments are organized into several traffic types, one of which is a group of one or more strings with the same bit rate characteristics. For example, all strings with a CBR (fixed bit rate) of 2Mpb s belong to the same type. The type of the variable bit rate (V B R) type typically includes only one string, because it is impossible for two variable bit rate strings to have the same bandwidth pattern at all times. Each type has a single set of transmission parameters to control bit rate, provide low unit delay variation (CDV), and precise speed. The number of types is programmable, but limited to a maximum of 1 2 8. There are two main transmission parameters for each type, one is the ideal scheduling time (TS) and one is the delta (Δ) of the scheduling time (TS). The basic mechanism is that when TS is equal to or less than the reference clock, a series of indicators are placed in the transmission queue. At the same time, the 値 TS increases by 値 △ and the transmission queue is a first-in-first-out queue. The string indicated by the string index is presented to ATM fif 7 2. In the high-priority segment, the following organizations achieve high accuracy bit rates and low CDV. This paper size applies to China National Standard (CNS) A4 specification (210 X 297 mm) • I ί III nnnnn I · nn I— nnnn sip, fl nn an (Please read the precautions on the back before filling this page) Line 丨· -30- 508504 Α7 Β7 V. Description of the invention (28). — Lit (Please read the notes on the back before filling this page) Due to the limited resolution of the reference clock, a single △ 値 usually cannot give the required accuracy . To achieve the required accuracy, two Δ 値 are used alternately, and the two 値 値 produce rates slightly above and slightly below the required bit rate. Each Δ 各个 of a different number of cells is used to compensate for the limited clock resolution and can provide arbitrary accuracy. △ Η and △ L (where △ L = △ Η + 1) substitute two different increments 値. N Η and N L parameters represent the number of units, and their corresponding increments are alternately valid. By this mechanism, the string is modulated, whereby the 'average bit rate approaches the required bit rate within the required accuracy. Figure 7 shows the bit rate achieved by this mechanism. In FIG. 7, the NΗ unit is sent out as ΔΗ, and the NU unit is sent out as / ^. This sequence is repeated periodically. Therefore, the average bit rate indicated by the dotted line is maintained as a long-term bit rate. Printed by the Intellectual Property Bureau's Consumer Cooperatives of the Ministry of Economic Affairs, the achievement of a low CDV is achieved by reducing collisions within a specified period of units from different strings. In many existing traffic shaping mechanisms, the main reason for collisions is that strings with the same bit rate are scheduled at the same time. This problem is particularly serious when the number of strings is large and the number of independent bit rates is small. In the preferred embodiment, the previous problem is emphasized by uniformly spacing the cells belonging to strings of the same type. In other words, if the increment of a string should be △, then the increment of this type is A / η, where η is the number of strings of this type. Every time a certain type is to be served, the data is taken from a continuous string. For example, if the unit of string 0 belonging to type 0 should increase △, each unit belonging to a string of the same type 0 (that is, string 0 to string η — 1) is sent out with an interval of Δ / η 'as shown in Figure 8 As shown. With the combination of the above organizations, the precise bit rate and the paper size can be applied to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -31-Printed by the Intellectual Property Bureau, Ministry of Economic Affairs, Consumer Consumption Cooperative 508504 Α7 _____________ Β7 V. Description of the invention (29) Low CD v. If the transmission queue is not blocked, the transmission mode of these units is as shown in FIG. 9. The high priority sector also handles V B R traffic. The traffic shaper 6 2 supports the smooth update of the transmission parameters. This update can be performed by the host 2 8 or by the command block 6 6. The command block 6 6 is programmed by the host 2 8 and its action is triggered when an exact position is sent from the string buffer 4 4. One such action is to replace the transmission parameters for a specified string in one of the traffic shapers 62. Once the data is sent just before the bit rate is changed, command block 6 6 immediately updates the parameters. Once it is ready, this program is autonomous and does not require any interaction with the host CP30. As a result, the host 2 8 does not need to interact exactly when it should be required to interact. In this way, the real-time characteristics of the string are maintained and the host load is kept to a minimum. Low priority block segments are organized into, for example, fixed 32 traffic types, where one type is a group of one or more strings with the same P CR (Spike Cell Rate). In terms of general data traffic, real-time limits are very insignificant. The main purpose of traffic shaping in the low-priority sector is to limit PCR in order to avoid network consolidation. The traffic shaping of the data packet is implemented by using an ideal schedule time (TS) and an increment (△) of TS, which is similar to the basic flux β-shaping mechanism in the high-priority sector. However, the priority of the data packet scheduling is lower than that of the instant communication. Only when the transmission queue of the high-priority segment is empty, the string of / data packets can be sent to AT M F I F07. This mechanism is implemented with a structure as shown in Figure 10. The paper size for communication is adapted to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) m In nm I— -1 ·· — ϋ al tmaMm emmmmm mmmmmmm Is · nmn ϋ is n an I m fn n * (Please read the precautions on the back before filling out this page) • Wire 丨 ·! -32- 508504 A7 B7 V. Description of the invention (30) Device 6 2 includes the previous The high-priority segment 200 and the low-priority segment 200 are described. In the high-priority sector 2 0 0, a memory 2 0 3 stores a set of new transmission parameters provided by the host 28 for each level. The new transmission parameters of each group include T s 1, △ H i , ΔL i, NH i, NL i, and P t 1 (where si 27). In this embodiment, P t i includes one or more string indicators, which indicate that one or more strings appended to type i 'memory 2 0 6 stores the current transmission parameters. When the host 2 8 or the command block 6 6 indicates a command, an update logic 2 0 4 is triggered by the command, thereby updating the current transmission parameters in the memory 2 6 to the new transmission in the memory 2 3 parameter. The register 2 1 2 stores a parameter Nr_classes showing the number of types of the host 2 8 from its receiving place, and the traffic logic 2 0 8 for each type (from 0 to N ^ classesj) checks whether TS 1 is equal to or less than by reference The current time indicated by the clock 2 10. If yes', the string index attached to the first string after this k-type 1 is inserted into a high priority transmission queue 2 1 6 and the traffic shaper 2 0 8 increases the TSi in the memory 206 by this Type AHi or · This △ η 1 or △ L 1 is rotated in accordance with N η i or NL i. After that, the segmented block 70 receives a string of indicators from the high priority send queue 2 1 6 and will be changed by The ATM cell of the string indicated by the string index is placed in the ATM FIFO 72. In the low priority sector 202, the memory 2 1 8 stores a set of transmission parameters provided from the host 2 8 for each level. In this embodiment, each group of transmission parameters includes TS j, △ j, and P tj (where 0 this paper size applies to the Chinese national standard (CNS> A4 specification (210 X 297 mm)) (please read the note on the back first) Please fill in this page again for the order) Order -------- • Line-Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economics-33- 508504 A7 B7 V. Description of the invention (31) ^ J ^ 3 1). P t 1 includes one or more string indicators, which indicate one or more strings appended to type j, and communication logic 2 2 0 for each level (from 0 to 3 1) to check whether TS j is equal to or less than the reference clock Indicates the current time 'and monitors whether the high priority sending queue 2 1 6 is empty. If so, the string index attached to the first string after this type j is inserted into the low priority sending queue 2 2 2 Among them, and the traffic logic 2 2 0 increases the TS j of the memory 2 1 8 by △ j of this type j. After that, the segmented block 7 0 receives the low-priority send queue 2 2 2 string index, and The ATM unit belonging to the stream indicated by the string indicator is placed in the ATM FIF 〇2. In the above embodiment, the A traffic shaping mechanism similar to the mechanism of the high priority sector 2 0 0 is applied to the low priority sector 2 0. However, the traditional leaky bucket mechanism is applied to the low priority sector 2 0 The traffic shaping mechanism of 2. [Order Box] This paper size applies to the Chinese national standard Gui Na m 0χ π public guide (please read the note on the back? Matters before filling out this page) a. Δ, ν * nnn Is nm I fnn 1_ I nnmn HB flu I n an n HI If mnn is determined by the number of fingers in the number ... Refer to, Xu Neutral Mobility has a string of mutual responsibility and a bit of material. At the time of the investment, the 6 complete match list is listed and the 6 modulating table loses the certificate holders to be taken quietly. The insurer needs to be procured. It will take 6 orders to be issued forever. It is enough to describe the middle block and the energy resources involved in the bidding party are not mutually exclusive if the order can be ordered. Some of the losers will be sent to the party. The losers must be accounted for. 4 and It is expected that 8 punches and 4 ticks will be required to make more promises. 2 slow-moving middle-machine strings will be rushed immediately. 8 cases will be slow and timely. 8 Set the main good, shown by 2 is located in the original. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economics and Economics 508504 A7 B7 V. Description of the invention (32) Instructions are loaded. The appropriate time is to load the output data string into the string buffer The time between the time of 4 4 and the time when the output data string is sent from the string buffer 44. Command block 6 6 scans the read indicator of string buffer 4 4. If it finds a match with a specified address, the command linked to that address will be executed, and the command will be executed from command block 6 6. except. The command block 66 triggers the address of the data leaving the string buffer 44. When a string buffer 4 4 is read, the read index is gradually increased in a wraparound process. Each string has a 'link list' which includes (address, command) pairs to be stored according to the address', and a bit is a pair (L'M) representing the address in the file, and has nothing to do with the physical address . L is the number of blocks, the block size of which is equal to the size of the string buffer 44, and M is the sequence number of the last block. Each string holds a W AC (wrapping counter), which counts the number of times the read indicator has been wrapped. If the following relationship holds, then an address match is found:
It (請先閱讀背面之注意事項再填寫本頁)It (Please read the notes on the back before filling this page)
Μ 且 C AW II L 移 偏 器 衝 緩 I 標 指 取 線丨· 經濟部智慧財產局員工消費合作社印製 塊 方 ο 令 3 命器 示生 顯產 1 令 1 命 圖個 。 數 下括 如包 施 6 實 6 之塊 構方 機令 此命 Ο 圖 ο 6 塊令 方命 之個 6 每 適 ο 在 ο 8 3 2 器 機生 主產 , 令 令命 命個 之每 串入 料載 資令 出命 輸之 個列 每表 理一 處將 ο 時 ο 及 3 刻 器時 生的 產當 器 存 暫 內 6 偏緩 器串 衝與 緩移 存偏 儲器 2 衝 ο緩 3 -之 器中 存 2 暫 ο , 3 中器 ο存 ο 暫 3 將 器 4 生 ο 產 3 令器 令命較 命在比 的 , 中 移 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -35- ⑽5〇4 經濟部智慧財產局員工消費合作社印製 Α7 Β7 五、發明說明(33 ) 衝器4 4中之讀取指標做比較。當有一環繞發生時,讀取 指檫取得緩衝器偏移。因此,當比較器3 0 4偵測到匹配 時’ W A C (環繞計數器)3 0 6被增加。比較器3 0 8 將W a C 3 0 6的計數和提供自命令暫存器3 1 6之目前 的L値做比較,比較器3 1 0將提供自命令暫存器3 1 6 之目前的Μ値和讀取指標-緩衝器偏移做比較。當比較器 3 0 8和3 1 0偵測到匹配時及〃閘(AND gate ) 3 1 2使由一佇列3 1 4所儲存之目前的命令出佇列。每 次對應於一目前的位址(L,Μ )之目前的命令從佇列 3 1 4被輸出,對應於下一個位址之命令從命令暫存器 3 1 6被排隊入佇列3 1 4中。因此,每個命令產生器 3 〇 0根據串緩衝器4 4之讀取指標來指示命令。 即將從命令方塊6 6所指不之命令爲: 改變位元率:此命令將允許串頻寬之改變,當指示此 口口令日守,通信量整形器6 2將一串從其目刖的類型中分離 出、更新目前之類型的△値、將此串附在新的類型之後、 並且更新新的類型之連結的△値。因此,個別串的位元率 在指定串位置處改變,舉例來說,對於在V B R (可變位 元率)之Μ P E G位元串而言這是有用的。 插入RC I :此命令允許在串中指定的位置插入一 R C I (速率改變指示器),R C I能夠通知遠方端子( 例如S Τ Β 1 8 )速率在當時改變了,並幫助Μ P E G解 碼器之時鐘復原。在歐洲專利申請案第 ’ ΕΡ 0 712 250 Α2號中,RCI之詳細說 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) ' · mmmme ·ϋ 1* ϋ— ί MmaMmm ί.*I «Β1·« n Mmmmmmm 0 -30 - 508504 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明說明(34 ) 明被敘述爲、'資料速率資料〃。當指示此命令時’ R c I .產生器6 8產生RC I ’並且分段方塊7 0結束目前的分 段,而且產生用於RC 1之分開的AAL 一 5 p D u ( 一atm單元),對於VBR2MPEG位元串而言’這 是有用的。 致能R A I D :此命令爲錯誤修正設定在 T X ra I D方塊6 〇中的適當參數。 禁止r A I D :此功能爲上述致能R A I D之相反功 會b 。 執行位元組交換:此命令允許來應付介於伺服器1 〇 與S T B 1 8之間的小endian/大endian問題。當指示此 命令時,位元組調換器6 4將輸出串中的一字元3 5 0內 的位元組按照一字元3 5 2的順序重新排列’如圖1 2中 所示。 致能不同的> D U尺寸:T C P可以要求分段,一 TCP封包被分成不同的IP封包’最後一個IP封包通 常需要與先前的一個之尺寸不同的AAL—5 PDU尺 寸。當指示此命令,分段方塊7 〇改變 A A L - 5 PDU 尺寸。 中斷C P U :此爲最普通之功能,當偵測到串中的一 定位置時,其要求主C P U 3 0互動。 〔VPI/VCI過濾方塊〕 圖1 3顯示在U N I (使用者網路介面)中所使用的 本纸張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) ~ -37- --------ti — (請先閱讀背面之注意事項再填寫本頁) 訂---------線丨· 508504 Α7 Β7 五、發明說明(35) 一個A T Μ單元的格式。一 Α τ M單元係由5 3個位元組 所組成,前面的5個位元組構成A τ M標題,而其餘的 4 8個位元組攜帶酬儀,a T Μ標題的前四個位元被稱爲 G F C (總流動控制)’ A Τ Μ標題中後面的2 4個位元 被稱爲VP Ι/VC I 。實際上’ VP i/vc I包括8 位兀之VP I和1 6位元之VC I ,ATM標題中後面的 3個位元被稱爲P T (酬載型樣),A Τ Μ標題中後面的 1個位元被稱爲C L Ρ (單元損耗優先權),a Τ Μ標題 中最後的8個位元被稱爲Η E C (標題錯誤控制), VP I/VC I過濾方塊84從ATM F I F〇82檢 索這種的A Τ Μ單元。 V Ρ I / V C I過濾方塊8 4決定所接收之A Τ Μ單 元之VP I /VC I是否爲應該被接受之組νρ 1/ VC I S的一員、決定該ATM單元應屬於那個串,並過 濾〇A Μ (操作,管理,和維持)f 5單元。爲了達到此 過濾程序’在VP I/VC I過濾方塊84的VP I / V C I轉換器8 5中實施從V ρ I / V C I到內部串I d 之VPI/VC I轉換。 V Ρ I / V C I轉換機構之目的在於讓合法的V Ρ I / V C I之範圍儘可能寬’而同時協助快速轉換。最好, 所有的V Ρ I / V c I都應該是被認可的。ν Ρ I / V C I轉換可以使用傳統的二位元搜尋技術來完成。然而 ,由於時間限制,最大的可接受搜尋爲5 1 2個入口之二 位元搜尋的等級。另一方面,作用的VP Ι/VC I的最 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) Φ 經濟部智慧財產局員工消費合作社印製 一 -M n i m m K·— mmmmmmm i I n- m n -n n 1 «ϋ m -38 - 508504 A7 _— B7 五、發明說明(36) 大數目應大於5 1 2,以便支援大量用戶之同時通信。 爲了符合此目的,V P I / V C I表被分成5 1 2個 入口的區段。每一入口表示介於V P I / V C I之間的關 係,並且一內部串I D視一分配機構I D進入到某一區段 之中,並且在每一區段中,入口被排列。 當一 A T Μ單元之接收時,一旦已發現到正確的區段 ,可以對該區段上執行二位元搜尋以找出正確的入口。因 此,分配V Ρ I / V C I s的分配機構必須根據V Ρ I / V C I而允許立即索引到一區段中。況且,爲了能有效率 地使用V Ρ I / V C I表,此機構必須允許V Ρ I / V C I s有寬的分布。換言之,此機構必須使所有的入口 儘可能隨機地分布在整個V Ρ I / V C I表上。若一 V Ρ I / V C I對應到已經完全滿的V Ρ I / V C I表之 一段,他必須予以拒絕,即使在其他區段中可能還有空間 〇 滿足這些要件的一個分配機構在於簡單地使用ν c I 之較低的X位元(其中X爲整數;例如3 )'來做爲雜亂鑰 匙,以便索引到VP I / V C I表中。當有大數目的ν ρ / V C s時,低位元將是2 4位元ν Ρ I / V C I領域中 最隨機的’並且允許均勻分布’這是合理的。 使用此型樣的機構’可符合快速查閱和沒有非法或不 認可之V Ρ I / ν c 1的要求’此機構之實施如下·· 圆14顯示^?1/^(:1轉換器8 5的方塊圖。當 一新的虛擬路徑/虛擬頻道7 P/V C變成有作用時,_ 本纸張尺度適用+國國家標準(CNS)A4規格(MO X 297公爱) -- --------t (請先閱讀背面之注意事項再填寫本頁) ! i I 1 線丨· 經濟部智慧財產局員工消費合作社印製 -39- 經濟部智慧財產局員工消費合作社印製 508504 A7 B7 五、發明說明(37) 不該新的v p / v c之V P i / V C 1和對應於該V P I / V C I之內部串I D之新的入口透過一雜亂函數4 0 〇 而進入根據入v C I之較低的三個位元之區段(亦即圖 1 3中第4位元組的第7,6 ’ 5位元)。更特別的是, 若VC I之較低的3個位元爲〇〇〇 ’則入口被儲存在 V P I / V C I表4 0 2中的區段1中。若V C I之較低 的3個位元爲〇〇 1 ,則入口被儲存在VP Ι/VC I表 4 0 2中的區段2中。.若VC I之較低的3個位元爲 0 1 0,則入口被儲存在VP I/VC I表4〇2中的區 段3中。同樣地,所有新的入口根據v c I之較低的3個 位元而被儲存在適當的區段中。因此’例如有4 0 9 6個 入口的VP I/VC I表40 2被分成例如有5 1 2個入 口之8個區段(區段1至區段8 ) ’在每一段中,入口以 上升或下降的順序重新排列,以便實施二位元搜尋。 在A T Μ單元的接收時,所接收之A T Μ單元之 V Ρ I / V C I被提供給一搜尋處理器4 2 0和雜亂函數 400,雜亂函數400根據VP I/VC I之較低的3 個位元來提供一個區段索引給搜尋處理器4 2 0。之後, 搜尋處理器4 2 0對對應於該區段索引的區段實施二位元 搜尋,以便找出正確的入口。舉例言之,若所接收之 ATM單元之VC I之較低的3個位元爲〇 1 〇,雜亂函 數4 0 0提供3做爲給搜尋處理器4 2 0的區段索引。之 後,搜尋處理器4 2 0對區段3實施二位元搜尋,以便找 出正確的入口 ’並且輸出所找到之入口的內部串I D。若 本纸張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) « · 1 n 11 l_i n n i J 、a an aKm lm& i i 1 1— e I n n - -40 - 508504 Α7 Β7 五、發明說明(38) 所接收之A T Μ單元的V C I之較低的3個位元爲1 1 1 ,雜亂函數4 0 0提供8做爲給搜尋處理器4 2 0之區段 索引。之後,搜尋處理器4 2 0對區段8實施二位元搜尋 ,以便找出正確的入口,並且輸出所找到之入口的內部串 1 D,輸出內部串I D被過濾程序所使用。 在上述實施例中,V Ρ I / V C I欄之較低的三個位 元被簡單地用做區段索引。然而,對V Ρ I / V C I欄可 使用更複雜的雜亂函數來產生區段索引。 在上述實施例中,當一新的V Ρ / V C變成有作用時 ,新的入口透過雜亂函數4 0 0而進入適當的區段。然而 ,有可能在具有與雜亂函數4 0 0之機構相同的雜亂函數 之主機2 8中建立包括新的入口之新的VP i/v C I表 、將新的V Ρ I / V C I表傳送到v ρ I / V C I轉換器 85 ,並且以新的VP I/VC I表更新VP Ι/VC I 表 4 0 2。 ' 〔型樣偵測器〕 主機2 8知道在指定的V C上進來的是何種資料’主 機2 8對每個V C指示型樣偵測器9 2即將掃瞄何種型樣 。型樣偵測器9 2的目的在於偵測輸入資料串中預設的位 元型樣。每次當偵測到匹配時,型樣偵測器9 2通知主機 2 8 、、資料被偵測〃狀態。當主機2 8接收到偵測的資訊 時,他將其發生之位址加進到主記億體3 2中的一表列中 。由於偵測本身係自動完成的,所以主機2 8在此時可以 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) ---------1 (請先閱讀背面之注意事項再填寫本頁) 訂--- 線丨Φ 經濟部智慧財產局員工消費合作社印製 -41 - 508504 A7 B7 五、發明說明(39) 執行其他工作。假如預設的位元型樣被偵測到,僅需中斷 主機2 8,並且能夠採取動作。 圖1 5顯示型樣偵測器9 2的方塊圖。在輸入資料串 經由接收路徑8 0被發送之前,主機2 8對每個V C指示 型樣偵測控制器5 0 6即將掃瞄之型樣。型樣偵測控制器 5〇6可以爲每個串設定暫存器5 0 4中4個3 2個位元 寬之預先編程的位元型樣。對齊電路5 0 0執行輸入資料 的位元組對齊。匹配電路5 0 2對每個串實施與4個預先 編程位元型樣做之位元組對齊匹配。每次當偵測到匹配時 ,匹配電路5 0 2通知控制器5 0 6偵測結果。 型樣偵測器9 2之目的的示例在於找出根據Μ P E G 標準所壓縮的視頻位元串中I -圖像的位置,在Μ P E G 位元串中,緊跟在G〇Ρ標題之後的圖像總是I -圖像。 因此,有可能藉由偵測確認G Ο Ρ標題之開始的組群-起 始-碼(3 2位^ )和確認圖像標題之開始的圖像—開始 -碼(3 2位元)來找到I —圖像的位置。 例如,當電影的Μ Ρ E G位元串係傳送自另一個 S M U 1 2以便複製電影時,組群一起始一碼和圖像—開 始-碼係在暫存器5 0 4中被設定爲預設的位元型樣。型 樣偵測器9 2偵測在所接收到之Μ Ρ E G位元串中的組群 -起始-碼和圖像一開始一碼。每次當在匹配電路5 0 2 中偵測到組群-起始-碼之後’立即偵測到圖像-起始-碼,型樣偵測控制器5 0 6將偵測狀態通知給主 CPU30。主CPU30將儲存裝置20 ( I —圖像儲 度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事項再填寫本頁) •t 訂---- 經濟部智慧財產局員工消費合作社印製 -42 - 508504 A7 B7 五、發明說明(4〇 ) 存在其內)之位址加進到主記憶體3 2的〜表列中。因此 ,在Μ P E G位元串流進接收路徑8 0期間,表示I —圖 像之位置的表列被建立。 當所儲存之Μ P E G位元串被傳送到S Τ Β 1 8時 ,此表列用做V C R操作’若S Τ Β 1 8請求V C R操 作(例如快轉,快速倒帶)’則主機2 8參考此表列·,並 且指示儲存裝置控制器2 2存取及檢索ί -圖像。 利用此特色,儲存在儲存裝置2 0中之資料爲、、原始 的〃或並未指定應用而被格式化爲’這增加了伺服器系統 1 0 (圖1 )的 ''應用獨立性〃和互動能力。 〔R X位址轉換器〕 R X位址轉換器9 6之目的在於從一串緩衝器4 6中 聚集同(非連接)的字元,並且建立到P C I匯流排2 4 之叢發資料,其~基本上係Τ X位址轉換器5 4位址轉換功 能的相反功能,其差異爲在此情況中必須考慮動態緩衝器 結構,叢發資料係透過P C I匯流排而被傳送到儲存裝置 2 0或主機2 8。 圖1 6顯示應用於即將被儲存在儲存裝置2 0之磁碟 0,1 ,2 ,3中的輸入資料串之位址轉換的示例,在此 示例中,下面用於串緩衝器4 8之讀取位址的序列係藉由 R X位址轉換器9 6而被產生,以便建立叢發資料6 0 0 〇 178, 182, 13,17,1099, 1103 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) * -43- ----------t (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印I衣 d^-r^JI n n ι ϋ n I ϋ I d n n ϋ n i n n n I ϋ n n I n n ·ϋ i— n I n ϋ ' 508504 A7 B7 五、發明說明(41 ) ο 碟 磁 於 用Μ and C AW II L deflector offset buffer I Marking line 丨 · Printed by the Consumer Consumption Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. Count the number of block formulas that include 6 and 6 to order this command. Figure ο 6 blocks of order and order 6 each suitable ο 8 3 2 machines to produce the main production, order each order of command and order Each column of the order of the material load order will lose ο ο ο ο ο ο ο ο ο ο 3 engraving time when the production time is stored in the storage 6 6 serial buffers and slow storage partial storage 2 red ο 3 -The device stores 2 temporary ο, 3 medium ο ο stored ο temp 3 will produce 4 orders ο produce 3 orders to make life less than the order of life, China Mobile's paper standards apply Chinese National Standard (CNS) A4 specifications (210 X 297 mm) -35- ⑽504. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the Invention (33) The reading index in the punch 4 4 is compared. When a wraparound occurs, the read pointer gets the buffer offset. Therefore, when the comparator 3 0 4 detects a match, the 'W A C (surround counter) 3 0 6 is incremented. Comparator 3 0 8 compares the count of Wa C 3 0 6 with the current L 値 provided from command register 3 1 6 and comparator 3 1 0 will provide the current value from command register 3 1 6 Μ 値 is compared with the read index-buffer offset. When the comparators 308 and 3 1 0 detect a match, the AND gate 3 1 2 dequeues the current command stored by a queue 3 1 4. Each time a current command corresponding to a current address (L, M) is output from the queue 3 1 4 and a command corresponding to the next address is queued into the queue 3 1 6 4 in. Therefore, each command generator 300 instructs the command according to the read index of the string buffer 44. The commands that will not be referred to from command block 6 6 are: Change the bit rate: This command will allow the string bandwidth to be changed. When this port password is instructed to keep watch, the traffic shaper 6 2 will remove a string from its destination. In the type, Δ 値 of the current type is updated, the string is appended to the new type, and Δ 値 of the link of the new type is updated. Therefore, the bit rate of an individual string is changed at a specified string position, which is useful, for example, for a MPEG bit string at V B R (variable bit rate). Insert RC I: This command allows an RCI (Rate Change Indicator) to be inserted at the specified position in the string. The RCI can notify remote terminals (such as S T B 1 8) that the rate has changed at that time and help the clock of the MPEG decoder. recovery. In European Patent Application No. 'EP 0 712 250 Α2, the details of the RCI are that the paper size applies to the Chinese National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling out this page ) '· Mmmme · ϋ 1 * ϋ— ί MmaMmm ί. * I «Β1 ·« n Mmmmmmm 0 -30-508504 Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention (34) The statement is described as , 'Data rate data'. When this command is instructed, 'R c I. Generator 6 8 generates RC I' and segmentation block 7 0 ends the current segmentation, and generates a separate AAL for RC 1-5 p D u (one atm unit) 'This is useful for VBR2MPEG bit strings. Enable R A I D: This command sets the appropriate parameters in the T X ra I D block 6 for error correction. Disable r A I D: This function is the opposite function of enabling R A I D above. Perform byte exchange: This command allows to deal with small endian / large endian problems between server 10 and ST B 18. When this command is instructed, the byte swapper 64 rearranges the bytes within one character 3 50 in the output string in the order of one character 3 5 2 'as shown in FIG. 12. Enable different > D U size: TCP can require segmentation, a TCP packet is divided into different IP packets. The last IP packet usually requires an AAL-5 PDU size that is different from the previous one. When this command is instructed, the segmentation block 70 changes the A A L-5 PDU size. Interrupt C P U: This is the most common function. When a certain position in the string is detected, it requires the main C P U 30 to interact. [VPI / VCI filter block] Figure 1 3 shows that the paper size used in UNI (User Network Interface) applies to China National Standard (CNS) A4 (210 X 297 mm) ~ -37-- ------ ti — (Please read the notes on the back before filling out this page) Order --------- line 丨 508508 Α7 Β7 V. Description of the invention (35) Format of an AT Μ unit . A A τ M unit is composed of 53 bytes. The first 5 bytes form the A τ M title, while the remaining 48 bytes carry the meter, the first four of the a T Μ title. The bits are called GFC (Total Flow Control). The next 24 bits in the ATM header are called VP I / VC I. In fact, 'VP i / vc I includes 8-bit VP I and 16-bit VC I. The last three bits in the ATM header are called PT (payload type), and the latter are in the A TM header. 1 bit is called CL P (Unit Loss Priority), the last 8 bits in a TM title are called Η EC (Title Error Control), VP I / VC I filter block 84 from ATM FIF 〇82 retrieves this AT unit. The V P I / VCI filter block 8 4 determines whether the VP I / VC I of the received A TM unit is a member of the group νρ 1 / VC IS that should be accepted, determines which string the ATM unit should belong to, and filters. AM (operation, management, and maintenance) f 5 units. To achieve this filtering procedure, a VPI / VC I conversion from V ρ I / V C I to the internal string I d is implemented in the VP I / V C I converter 85 of the VP I / VC I filtering block 84. The purpose of the V Pl / V C I conversion mechanism is to make the range of legal V Pl / V C I as wide as possible 'while assisting in rapid conversion. Preferably, all V P I / V c I should be recognized. ν Ρ I / V C I conversion can be accomplished using traditional two-bit search techniques. However, due to time constraints, the maximum acceptable search is a level of 512 binary search. On the other hand, the most important paper size of the functioning VP Ⅰ / VC I is applicable to China National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling this page) Φ Intellectual Property Printed by the Bureau's Consumer Cooperatives -M nimm K · — mmmmmmm i I n- mn -nn 1 «ϋ m -38-508504 A7 _— B7 V. Description of the invention (36) The large number should be greater than 5 1 2 for support A large number of users communicate at the same time. To meet this purpose, the V P I / V C I table is divided into 5 1 2 entry sections. Each entry represents a relationship between V P I / V C I, and an internal string ID is entered into a section as a distribution mechanism ID, and in each section, the entries are arranged. When an A T MU unit is received, once a correct segment has been found, a binary search can be performed on the segment to find the correct entry. Therefore, the distribution mechanism that allocates V Pl / V C I s must allow immediate indexing into a sector based on V P I / V C I. Moreover, in order to use the V Pl / V C I table efficiently, this mechanism must allow a wide distribution of V Pl / V C I s. In other words, this mechanism must make all entries as randomly distributed as possible across the entire V P I / V C I table. If a V Pl / VCI corresponds to a section of the V Pl / VCI table that is completely full, he must reject it, even though there may be room in other sections. A distribution mechanism that meets these requirements consists in simply using ν The lower X bit of c I (where X is an integer; for example, 3) 'is used as a hash key for indexing into the VP I / VCI table. When there is a large number of ν ρ / V C s, the lower bits will be the most random ′ in the field of ν Pl / V C I and it allows for uniform distribution. This is reasonable. The organization using this model 'can meet the requirements of quick reference and no illegal or unacceptable V Pl / ν c 1' The implementation of this organization is as follows: · Circle 14 display ^? 1 / ^ (: 1 converter 8 5 Block diagram. When a new virtual path / virtual channel 7 P / VC becomes effective, _ this paper size applies + National National Standard (CNS) A4 specification (MO X 297 public love)----- ---- t (Please read the precautions on the back before filling this page)! i I 1 line 丨 · Printed by the Employee Consumption Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs-39- Printed by the Employee Consumption Cooperative of the Intellectual Property Bureau of the Ministry of Economy 508504 A7 B7 V. Description of the invention (37) The new entry of the new vp / vc VP i / VC 1 and the internal string ID corresponding to the VPI / VCI is entered through a messy function 4 0 0 The segment of the lower three bits (ie, the 7th, 6'-5th bits of the 4th byte in Figure 13). More specifically, if the lower 3 bits of VC I are 〇〇〇 'then the entry is stored in the VPI / VCI table 402 in section 1. If the lower 3 bits of the VCI is 〇01, then the entry is stored in the VP I / VC I table In section 2 in 4 0 .. If the lower 3 bits of VC I are 0 1 0, the entry is stored in section 3 in VP I / VC I table 4 02. Similarly All new entries are stored in the appropriate sector according to the lower 3 bits of VC I. So 'for example VP I / VC I table with 4 0 9 6 entries 40 2 is divided into for example 5 1 of the 8 sections of the 2 entrances (Section 1 to Section 8) 'In each section, the entrances are rearranged in ascending or descending order in order to perform a binary search. At the reception of the AT MU unit, all The received V Pl / VCI of the AT Μ unit is provided to a search processor 4 2 0 and a scramble function 400. The scramble function 400 provides a sector index to the lower 3 bits of the VP I / VC I The search processor 4 2 0. After that, the search processor 4 2 0 performs a binary search on the section corresponding to the section index in order to find the correct entry. For example, if the VC of the received ATM unit is The lower three bits of I are 〇〇 〇, the chaotic function 4 00 provides 3 as the sector index for the search processor 4 2 0. After that, the search processor 4 2 0 for sector 3 Perform a two-digit search to find the correct entry 'and output the internal string ID of the found entry. If the paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (please read the back first) Please note this page before filling in this page) «· 1 n 11 l_i nni J , a an aKm lm & ii 1 1— e I nn--40-508504 Α7 Β7 V. Description of the invention (38) of the AT Μ unit received The lower 3 bits of the VCI are 1 1 1 and the scramble function 4 0 0 provides 8 as the sector index for the search processor 4 2 0. After that, the search processor 420 performs a two-bit search on section 8 in order to find the correct entry, and outputs the internal string 1 D of the found entry, and the output internal string I D is used by the filtering program. In the above embodiment, the lower three bits of the V P I / V C I column are simply used as the sector index. However, more complex messy functions can be used for the V P I / V C I columns to generate sector indexes. In the above embodiment, when a new V P / V C becomes effective, the new entry enters the appropriate section through the mess function 4 0 0. However, it is possible to create a new VP i / v CI table including a new entry in the host 28 having the same chaotic function as the mechanism of the chaotic function 4 0 0, and transfer the new V P I / VCI table to v The p I / VCI converter 85 is updated with the new VP I / VC I table VP I / VC I table 4 02. '[Pattern Detector] The host 2 8 knows what kind of data is coming in the specified V C' The host 2 8 indicates to each V C what type the pattern detector 9 2 will scan. The purpose of the pattern detector 92 is to detect a preset bit pattern in the input data string. Each time a match is detected, the pattern detector 92 notifies the host 2 8 and the data is detected. When the host 28 receives the detected information, he adds the address of its occurrence to a list in the main body 302. Since the detection itself is done automatically, the host 2 8 can apply the Chinese National Standard (CNS) A4 specification (210 X 297 mm) at this time. --------- 1 (Please read first Note on the back, please fill out this page again) Order --- LINE 丨 Φ Printed by the Consumers' Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs-41-508504 A7 B7 V. Description of Invention (39) Perform other work. If the preset bit pattern is detected, only the host 2 8 needs to be interrupted and actions can be taken. Figure 15 shows a block diagram of the pattern detector 92. Before the input data string is transmitted via the receiving path 80, the host 2 8 indicates to each V C the pattern detection controller 5 6 to scan the pattern. The pattern detection controller 506 can set a pre-programmed bit pattern of 4 32 bits wide in the register 504 for each string. The alignment circuit 5 0 0 performs byte alignment of the input data. The matching circuit 502 performs byte-alignment matching on each string with 4 pre-programmed bit patterns. Each time a match is detected, the matching circuit 50 2 notifies the controller 5 0 6 of the detection result. An example of the purpose of the pattern detector 92 is to find the position of the I-picture in the video bit string compressed according to the MPEG standard. In the MPEG bit string, the position immediately following the GOP header Images are always I-images. Therefore, it is possible to detect the group-start-code (32 bits ^) of the beginning of the G 0 P title and the image-start-code (32 bits) to confirm the beginning of the image title. Find I — the location of the image. For example, when the MP EG bit string of a movie is transmitted from another SMU 12 to copy the movie, the group-start-one code and the picture-start-code system are set in the register 504 as the pre- Set the bit pattern. The pattern detector 92 detects the group-start-code and the image-start-one-code in the received MPG bit string. Every time when the group-start-code is detected in the matching circuit 5 02, the image-start-code is detected immediately, and the pattern detection controller 5 0 6 notifies the master of the detection status. CPU30. The main CPU 30 will store the storage device 20 (I — the image storage rate is in accordance with the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (please read the precautions on the back before filling in this page) • t order ---- economy Printed by the Ministry of Intellectual Property Bureau's Consumer Cooperatives-42-508504 A7 B7 5. The address of the invention description (40) is added to the ~~ list of the main memory 32. Therefore, during the MPEE bit stream into the receiving path 80, a table showing the position of the I-picture is created. When the stored M PEG bit string is transmitted to S T B 1 8, this list is used for VCR operation 'If S T B 1 8 requests VCR operation (eg fast forward, fast rewind)' then host 2 8 Referring to this list, and instructing the storage device controller 22 to access and retrieve the image. With this feature, the data stored in the storage device 20 is, original, or is not specified as an application and is formatted as' This increases the `` application independence '' of the server system 10 (Figure 1) and Ability to interact. [RX address converter] The purpose of RX address converter 9 6 is to gather the same (non-connected) characters from a series of buffers 4 6 and to create a burst of data to PCI bus 2 4, which ~ Basically, it is the opposite function of the TX address converter 54 address conversion function. The difference is that in this case, the dynamic buffer structure must be considered. The burst data is transmitted to the storage device 20 through the PCI bus. Host 2 8. FIG. 16 shows an example of address conversion applied to the input data string to be stored in the storage device 20 of the disk 0, 1, 2, 3. In this example, the following is used for the string buffer 4 8 The sequence of the read address is generated by the RX address converter 96, in order to create a batch of data 6 00 178, 182, 13, 17, 1099, 1103. This paper standard applies Chinese National Standard (CNS) A4 size (210 X 297 mm) * -43- ---------- t (Please read the notes on the back before filling out this page) The Intellectual Property Bureau of the Ministry of Economic Affairs ’s Consumer Cooperative Cooperative Office d ^ -r ^ JI nn ι ϋ n I ϋ I dnn ϋ ninnn I ϋ nn I nn
3 8 1-1 9 7 TX3 8 1-1 9 7 TX
οο ί—I 1 8 IXοο ί—I 1 8 IX
4 ο IX 1—I 碟 .磁 於 用4 ο IX 1—I plate. Magnetic for use
4 8 1—Iο 8 ,I 5 ί-14 8 1—Iο 8, I 5 ί-1
IXο IX IX 9 IXIXο IX IX 9 IX
5ο IX Γ—I 2 碟 磁 於 用 5 8 IX IX 8 6 3 碟 磁 於 用 (請先閱讀背面之注意事項再填寫本頁) f 經濟部智慧財產局員工消費合作社印製 I n I I n n n n I ί ϋ n n n an I m n n n n I— —a n ϋ n n a^i n n 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -44-5ο IX Γ—I 2 magnetic disks for use 5 8 IX IX 8 6 3 magnetic disks for use (please read the precautions on the back before filling out this page) f Printed by the Intellectual Property Bureau Staff Consumer Cooperative of the Ministry of Economic Affairs I n II nnnn I ί ϋ nnn an I mnnnn I— —an ϋ nna ^ inn The paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -44-
Claims (1)
Applications Claiming Priority (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP96203340A EP0847216A1 (en) | 1996-11-27 | 1996-11-27 | Apparatus, system and method for sending/receiving data to/from an ATM-network |
EP96203339A EP0847171A1 (en) | 1996-11-27 | 1996-11-27 | Method and device for streaming data, including provision of write addresses |
EP19960203334 EP0845905B1 (en) | 1996-11-27 | 1996-11-27 | Apparatus for delivering data, including pattern detection |
EP96203338A EP0847215A1 (en) | 1996-11-27 | 1996-11-27 | Method and device for delivering data and loading of a command |
EP96203341A EP0847217A1 (en) | 1996-11-27 | 1996-11-27 | Method and apparatus for translating VPI/VCI of an ATM cell into an internal ID |
Publications (1)
Publication Number | Publication Date |
---|---|
TW508504B true TW508504B (en) | 2002-11-01 |
Family
ID=27671121
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW86117151A TW508504B (en) | 1996-11-27 | 1997-11-17 | Method and apparatus for serving data |
Country Status (1)
Country | Link |
---|---|
TW (1) | TW508504B (en) |
-
1997
- 1997-11-17 TW TW86117151A patent/TW508504B/en not_active IP Right Cessation
Similar Documents
Publication | Publication Date | Title |
---|---|---|
USRE44702E1 (en) | Method and apparatus for serving data | |
TW387183B (en) | Extending asynchronous transfer mode (ATM) QOS across local area networks | |
USRE42081E1 (en) | Information transmission apparatus, traffic control apparatus, method of managing bandwidth resources using the same and method of admitting a call, using variable-rate encoding | |
US5966546A (en) | Method and apparatus for performing TX raw cell status report frequency and interrupt frequency mitigation in a network node | |
US8073006B2 (en) | ATM cell conversion circuit and ATM cell conversion method | |
TWI277324B (en) | Network packet storage method and network packet transmitting apparatus using the same | |
JPH1079738A (en) | Data transfer rate control method and data transfer rate controller using the same | |
US5941952A (en) | Apparatus and method for transferring data from a transmit buffer memory at a particular rate | |
EP1124357B1 (en) | Method and device for communicating between a first and a second network | |
US5822612A (en) | Apparatus and method for managing schedule table pointers | |
US6067563A (en) | Method and apparatus for avoiding control reads in a network node | |
US6618380B1 (en) | Priority queueing | |
US5960215A (en) | Transmit data FIFO for flow controlled data | |
US5862206A (en) | Method and apparatus for performing raw cell status report frequency mitigation on transmit in a network node | |
TW508504B (en) | Method and apparatus for serving data | |
EP1555620B1 (en) | Stream memory manager | |
US7290062B2 (en) | Method for mutual transfers of control parameters through a communications network | |
Baugher et al. | A multimedia client to the IBM LAN Server | |
EP0847166B1 (en) | Method for traffic shaping and traffic shaper | |
US7349406B1 (en) | Method and apparatus for virtual network connection merging | |
US6212567B1 (en) | Method and apparatus for performing raw cell status report frequency mitigation on receive in a network node | |
EP0845905B1 (en) | Apparatus for delivering data, including pattern detection | |
EP0847217A1 (en) | Method and apparatus for translating VPI/VCI of an ATM cell into an internal ID | |
EP0847216A1 (en) | Apparatus, system and method for sending/receiving data to/from an ATM-network | |
EP0847171A1 (en) | Method and device for streaming data, including provision of write addresses |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
GD4A | Issue of patent certificate for granted invention patent | ||
MM4A | Annulment or lapse of patent due to non-payment of fees |