200532446 九、發明說明： 【發明所屬之技術領域】 本發明係關於一種用於在一儲存裝置與一主機系統之間 即時處理資料流量的方法。 本發明亦關於一種用於在一儲存裝置與主機系統之間即 時處理資料流量的電路。 本發明係進一步關於一種用於處理已儲存之資料的裝 置。 、 、 本發明亦關於一種資料載體。 另外’本發明係關於一種程式化電腦。 【先前技術】 美國專利案第US 5,880,902號揭露一種用於適應校準硬 碟驅動器效能之方法。驅動器之效能參數係由硬碟驅動哭 電路所監視。當參數值改變’則硬碟驅器内的運作係隨著 參數之改變而予以執行。 【發明内容】 本發明之-個目的在於提供一種在_健存裝置與一主機 系統之間即時處理資料流量的方法，丨中操作該儲存裝置 之主機系統在儲存裝置之參數改變時會得到通知。 該目的係藉由提供依據本發明之方法而予以達成，盆步 驟包含，#由主機系統產生資料處理要求；藉由依據二绪 存裝置模型之儲存裝置安排該用於執行之#料處理要求， 該拉型係由主機系統予以儲存在 于隹芏祛糸統内並且包含該儲 存展置之至少一個效能參數·以 此^ ^，以及執行該資料處理要求； 98002.doc 200532446 該儲存裝置之模型係藉由至少執行後 量測健存裝置之至少-個效能參數；以及^予以設定： 量測到的效能參數值儲存到主機系統中:主機糸統將 置之一個模型。 &而形成儲存裝 本發明係基於以下認知··對於一完全主 制並操取來自-儲存裳置之資料的合適之即日二^中即時控 參數變化值之認知應與主機一致而非儲存以此，效能 主機知道是那-個即時要求在請求資料處理-㈣里由在於 ，而儲存裝置在大部分情況下僅是一個得儲存- 行之里各子。告主擔金处立 、；斗要求以執 罢/'Ί 意識到健存裝置效能參數值之偏 :θ考慮到對未來之資料處理要求安排 在-貝料必須根據即時要求予^理㈣财 ^ 些資料必須在最後限期之前予以傳送。^ 、亦即某 在根據本發明之方法之—個具體實施例中，儲存量 X能魏值從㈣成—更新後之儲存裝置模型 由後述步驟所取代：將該量測到的效能參數與 = 椒型所包含之效能參數作比較，其中該模型係儲存在= ^以及當㈣到的效能參數與模型所包含之效能 差一預定值時’以量測到的參數取代模型所包含之效妒參 數值。 ^ 此具體實施例之—項優點在於效能參數值不予以 要的更新。這降低了主機和儲存裝置之功率消耗並且降低200532446 IX. Description of the invention: [Technical field to which the invention belongs] The present invention relates to a method for processing data traffic in real time between a storage device and a host system. The invention also relates to a circuit for processing data traffic between a storage device and a host system in real time. The invention further relates to a device for processing stored data. The invention also relates to a data carrier. In addition, the present invention relates to a stylized computer. [Prior Art] US Patent No. 5,880,902 discloses a method for adapting and calibrating the performance of a hard disk drive. The performance parameters of the drive are monitored by the hard drive driver circuit. When the parameter value is changed ', the operation in the hard disk drive is executed as the parameter changes. [Summary of the Invention] One object of the present invention is to provide a method for real-time processing of data traffic between a health storage device and a host system. The host system operating the storage device will be notified when the storage device parameters are changed . The object is achieved by providing a method according to the present invention. The step includes: # generating a data processing request by a host system; and arranging the # material processing request for execution by a storage device according to a binary storage device model. The pull system is stored in the host system by the host system and contains at least one performance parameter of the storage deployment. Based on this, and the data processing requirements are executed; 98002.doc 200532446 The model system of the storage device By measuring at least one performance parameter of the storage device after execution, and ^ setting: the measured performance parameter value is stored in the host system: the host system will set a model. & Forming the storage device The present invention is based on the recognition that the knowledge of a real-time control parameter change value that is suitable for a complete mastering and manipulation of data from the storage equipment should be consistent with the host, not stored In this way, the performance host knows what the real-time request is in requesting data processing, and the storage device is, in most cases, just one storage-all. Inform the owner of the burden of execution; Requirement to be implemented / 'Ί Be aware of the deviation of the performance parameter values of the health storage device: θ Considering the requirements for future data processing, the materials must be arranged according to the immediate requirements. ^ Some information must be transmitted before the deadline. ^ In other words, in a specific embodiment of the method according to the present invention, the storage capacity X can be changed from the original value. The updated storage device model is replaced by the following steps: the measured performance parameter and = The performance parameters included in the pepper type are compared, where the model is stored at = ^ and when the obtained performance parameter is different from the performance included in the model by a predetermined value, 'the measured parameters are used to replace the effects included in the model. Envy parameter value. ^ One advantage of this embodiment is that the value of the performance parameter is not updated. This reduces the power consumption of the host and storage devices and reduces
存有參數值之記憶體的損耗。後者在該等值存於非揮發性 洋動閘記憶體時尤具意義。 X 98002.doc 200532446 二據本發明之方法之一個具體實施例進一步包含執疒夕 f:::步驟’·從多重量測"導出至少-個代表效能參 來數1以及將該導出的數值視為在儲存量測到的效能 k而$成_更新之儲存裝置㈣（根 之方法）之步驟中量測到的值。 頁弟1項 4此具體實施例之—項優點在於#數值係以該種方法 時’突然發生之單一效能混乱得以平均掉。 在根據本發明之方、、表 2 ^ ^ ^ 月之方法之另一具體貫施例中，導出的值至 ;V、、、&下其中之一 ··該等量測值之中間值；該等量测值之 平均值"亥寺量測值之平均值或中間值中的任何一個數值 加上一預定實數乘以標準差的總合。 舉例採用量測到的值加上標準差乘以一預定實數值的一 項優點在於-特定之合理上限係提供予一參數。當談 舉料：儲存裝置要求_在_配置單元内之資料^理:需 之日守間日寸’僅採用平均值並假設所有要求皆採用該時間旦 ，例可成會在儲存裝置老化時造成混亂且資料也許會在全 ρ擷取之則必須予以讀取兩次。在這種狀況並且儲存裝置 要求尚未於預期時間内完成即遭中斷的情況下，許多資料 处里要求將不予以處理。這會讓使用者非常困擾，但對於 j 士 、’罔路％境中的一個重要系統則有致命影響。當採用 、、上=，其中（假設）99%的資料得以處理時，大部分的要 长可知到處理且適當的要求安排仍是有可能的。 壯在根據請求項1之方法之又一具體實施例中，保存儲存 衣置原始換型之備份，緊接著是經過更新的模型。 98002.doc 200532446 此具體實施例之-項優點在於當參數已量測錯誤而導致 2機及/或儲存裝置誤動作時，可重新設置原始參數以便 讓主機使賴存裝置之原始效能參數值。當該等值嚴重偏 差時，-具有主機和儲存裝置之系統可能無法完全正確地 工作，但至少將儘可能地能夠量測真實的效能參數值並予 以儲存以供進一步使用。 該處理單元經進一步設想以儲存量測到的&能參數值 根據本發明之電路包含—用於健存—儲存裝置模型之記 憶體’該模型至少包含—儲存裝置效能參數；—根據一儲 存裝置模型用於安排資料處理要求以供儲存裝置執行之處 理單元；以&一用於量測儲存装置效能參數之量測單元； 從 而形成一經過更新之儲存裝置模型。 根據本發明之裝置包含—用於儲存資料之儲存裝置；一 主機電路，該主機電路經設想係用於：產生資料處理要求 供儲存裝置執行；並且處理與資料處理要求有關之資料； 用於輸出處理過之資料的輸出；以及根據請求項i之電 路，其中該模型係一由該裝置所包含之儲存裝置模型。 根據本發明之資料載體承載電腦可讀取資料，包含能使 一電腦執行根據請求項丨之方法的電腦可執行指令。 根據本發明之程式化電腦係經程式化以執行根據請求項 1之方法。 【實施方式】 圖1表示一儲存系統軟體架構100,用於說明根據本發明 之方法之個具體貫施例。儲存系統軟體架構丨00包含一 98002.doc 200532446 作荼糸、、死i 1ϋ、一應用程式（appHcati〇n) 120、一即時槽案 系統130以及一 IDE驅動器14〇，其中該IDE驅動器140在較 佳具體貫施例中與一作為一儲存裝置之（實體）硬碟驅動器 15 0通^ 即日T指案糸統13 0包含一即時資料處理器13 2、 一最大效果資料處理器134、一檔案系統136和一排程器 (scheduler) 138 〇 即時資料和最大效果資料之間的差異在於即時資料必須 在個預疋日守框内予以處理並且該前提比受處理資料之完 1性還要重要。另一方面，不論處理資料所需的時間為 何，最大效果資料都必須儘可能好地予以處理（或不遭致 破壞）。 即時資料處理（讀取及/或寫入資料）具重要性之情況的一 個貫施例為處理視訊資料時。當MpEG壓縮視訊資料未在 錄放期間内（但及時）予以完全正確地傳送時，會使視訊部 分產生某些缺陷。另一方面，當為了正確取得資料而行使 多重嘗試（attempt)時，例如超過一秒鐘，會使資料顯示造 成遲滯（hick-ups)或甚至是黑掉（black-〇uts);不僅是對於 視訊本身，對於其它由硬碟驅動器15〇處理的串流⑽議） 也是一樣。 取大效果需求比即時需求更好的情況之一個實施例為應 用程式（像是電腦程式）資料。當—應用程式未予完整正= 地擷取時，會使一處理單執行到錯誤的指令。這會接著造 成應用程式之失序（c〇llapse)。對於較常被比pc使用者之 經驗還少之使用者使用之消費性電子裝置而言，此類故障 98002.doc -10- 200532446 是無法被接受的。取而代之的是，資料擷取採用較長時間 並擷取無誤被列為優先。 當應用程式120產生一或多個資料處理要求時，每一類 資料皆由其自身擁有的處理器所處理，對檔案系統US提 出要求。檔案系統136查詢硬碟驅動器150上的位置。使用 該資訊和各種要求之優先權（即時或最大效果），資料處理 要求係為了按照硬碟驅動器15〇之最有效處理而由排程哭 138予以排程。這舉例意指懸置（pending)資料處理要求係 經排程而使得硬碟驅動器15〇之一個驅動臂（arm)(未示）必 須僅在磁碟（未示）上掃描一次以處理（從而讀取及/或寫入） 所有資料而非連續兩次（a c〇uple 〇f times)由内掃描至外 緣。 資料處理要求中與分散在磁碟之資料有關的有效排程具 有重要性。當資料分散在磁碟上時，意味著切換過度 (switch overhead)，其包含搜尋時間和旋轉延遲。就一 2 肌之3.5 fHDD而5，完全旋轉（⑽r〇tati〇n)係比傳送⑽ 计位元組之時間還多。最大搜尋時間（介於最内執與最外 軌之間）通常是完全旋轉時間的三到四倍。該時間為傳送 五十萬位元組之時間。不適當的排程會在切換過度上輕易 地浪費90%的HDD時間。Loss of memory with parameter values. The latter is particularly significant when these values are stored in non-volatile ocean brake memory. X 98002.doc 200532446 A specific embodiment of the method according to the present invention further includes the following steps: f ::: step '. Derive at least one representative performance parameter from multi-weight measurement " 1 and the derived value. It is regarded as the value measured in the step of storing the measured performance k and updating the updated storage device (the root method). Page 1 item 4-one of the advantages of this embodiment is that the # value is based on this method and the single performance chaos that occurred suddenly is averaged out. In another specific embodiment of the method according to the method of the present invention, Table 2 ^ ^ ^ month, the derived value is to one of the following: V, ,, &... The average of these measurements " Any one of the average or intermediate values of the Haisi measurement plus a predetermined real number times the standard deviation. One advantage of using the measured value plus the standard deviation multiplied by a predetermined real value, for example, is that a specific reasonable upper limit is provided to a parameter. When talking about materials: storage device requirements _ information in the _ configuration unit ^ management: the day of the need to keep the day inch 'only use the average value and assume that all requirements are used that time, for example, the storage device will age Causes confusion and data may be retrieved at full ρ but must be read twice. In this situation and the storage device request was interrupted before it was completed within the expected time, many data service requests will not be processed. This can be very confusing for users, but it has a fatal effect on an important system in the context of j, ’Kushiro%. When using,,, and =, of which (assumed) 99% of the data is processed, most of the directors know that processing and appropriate request arrangements are still possible. In yet another specific embodiment of the method according to claim 1, a backup copy of the original replacement of the stored clothes is stored, followed by the updated model. 98002.doc 200532446 One advantage of this specific embodiment is that when the parameters have been measured incorrectly and the machine and / or storage device malfunctions, the original parameters can be reset to allow the host to rely on the original performance parameter values of the device. When these values are severely deviated, a system with a host and a storage device may not work completely correctly, but at least it will be possible to measure the true performance parameter values and store them for further use. The processing unit is further conceived to store the measured & energy parameter values according to the circuit of the present invention including-memory for storage-storage device model ', the model includes at least-storage device performance parameters;-according to a storage The device model is used to arrange a data processing request for the processing unit for the storage device to execute; & a measurement unit for measuring the performance parameters of the storage device; thereby forming an updated storage device model. The device according to the invention comprises a storage device for storing data; a host circuit which is envisaged to: generate a data processing request for execution by the storage device; and process data related to the data processing request; for output Output of processed data; and a circuit according to claim i, wherein the model is a storage device model included by the device. The data carrier according to the present invention carries computer-readable data, and includes computer-executable instructions that enable a computer to execute the method according to the request. A stylized computer according to the present invention is programmed to perform the method according to claim 1. [Embodiment] Fig. 1 shows a storage system software architecture 100 for explaining a specific embodiment of the method according to the present invention. The storage system software architecture 00 includes a 98002.doc 200532446 program, a dead i 1 card, an application program (appHcati0n) 120, a real-time slot system 130, and an IDE driver 140. The IDE driver 140 is in In the preferred embodiment, a (physical) hard disk drive 15 0 as a storage device is used ^ Today T refers to the case system 13 0 includes a real-time data processor 13 2, a maximum effect data processor 134, a The file system 136 and a scheduler 138 〇 The difference between the real-time data and the maximum effect data is that the real-time data must be processed within a predetermined day frame and the premise is more than the completeness of the processed data important. On the other hand, regardless of the time required to process the data, the maximum effect data must be processed as well as possible (or not damaged). A consistent example of when real-time data processing (reading and / or writing data) is important is when processing video data. When MpEG compressed video data is not completely and correctly transmitted during the recording (but in time), it will cause some defects in the video part. On the other hand, when multiple attempts are made to obtain the data correctly, for example, more than one second, the data display may cause hick-ups or even black-outs; not only for The video itself is the same for other streams that are handled by the hard drive 15). One example of a situation where a large effect requirement is better than a real-time requirement is application data (such as a computer program). When-the application is not fully fetched, it will cause a processing order to execute the wrong instruction. This will then cause the application to be out of order (c0llapse). For consumer electronic devices that are more commonly used by users who have less experience than PC users, such failures 98002.doc -10- 200532446 are unacceptable. Instead, it takes longer to retrieve data and it is prioritized. When the application program 120 generates one or more data processing requests, each type of data is processed by its own processor, which makes a request to the file system US. The file system 136 queries the location on the hard disk drive 150. Using this information and the priority of various requests (immediate or maximum effect), data processing requests are scheduled by Scheduler 138 for the most efficient processing of the hard drive 150. This example means that the pending data processing request is scheduled so that one drive arm (not shown) of the hard disk drive 15 must be scanned only once on the disk (not shown) for processing (thus Read and / or write) All data is scanned from the inside to the outer edge instead of two consecutive times. Effective scheduling in data processing requests related to data scattered across disks is important. When data is spread across the disk, it means switch overhead, which includes seek time and spin latency. For a 2 muscle 3.5 fHDD and 5, a full rotation (⑽rotation) is more than the time to transmit the ⑽count. The maximum seek time (between the innermost and outermost tracks) is usually three to four times the full rotation time. This time is the time when 500,000 bytes are transmitted. Improper scheduling can easily waste 90% of HDD time on excessive switching.
即時檔案系統130與硬碟驅動器150之間的通訊係由IDE 驅動器140所執行。 貝。料處理要求中藉由排程器138所作的排程係以硬碟驅 動器15 0之一個模1 a λ & # u 、& 160為基礎。模型160至少包含硬碟驅 98002.doc 200532446 動器15〇之—個參數，如硬碟驅動器」 碟驅動器150傳送至一*德/ 月見貝科由硬 旦有一祛系統之速率-或搜尋時間-切換— 有一頃取頭而將資料由硬碟之一置 處理（讀取或寫入）資料 另位置以 間〇 更樂凝軺速度所需的（平均）時 硬碟驅動器1 5〇之效能尤、當^ # ,效此在運作期間經常改變。例如，已 知在熱環境中-70°C以上-银八#曰 已 π , φ5 上現今$見之系統之服務時 間大幅增加。這巖會汕旦/始 r 地衫響一使用硬碟驅動器15〇作為一 儲存裝置之系統的運作，在接 ”、、 ^在後面的說明中也可明顯看出。 因此’本發明建議在運作期間更新硬碟驅動器⑼之模型 160而在排程資料處理要求時考慮模型⑽之效能參數值中 的偏差，該偏差將不總是下降。為此，在儲存系統軟體架 構100中引用一量測應用程式17〇。 置測應用程式170從硬碟驅動器15〇導出效能參數。量測 應用程式…能夠處理一效能參數之一組多重量測值並且 至少由該組值導出-個數值。所導出的數值係視為量測值 且模型160中之參數值係以該值予以更新，所以該導出值 係用於更新模型160中之參數值。 在一具體實施例中，導出之值為該組量測值之中間值。 在另一具體貫施例中，導出瓷值為該組量測值之平均值。 在又一具體實施例中，導出之值為該組量測值之中間值再 加上標準差或標準差乘以一實數。排程器138在排程下一 個要求時使用更新過的模型丨6〇。在根據本發明之方法之 另一具體實施例中，平均值係得自一所謂的「移動視窗 98002.doc -12- 200532446 (moving wind〇w)」，為 ^ ^ m,. 所遍知。該具體實施例之 一個優點在於平均值以及因而模 使主機系統時間適應。 、之多數值逐漸變化， 圖2表示一消費性電子裝置㈣’作為根據本發明 之一個具體實施例。裝置細包含一錯 硬碟210、一用於卢裡：欠土、丨、 取子為 、处貝只斗之視訊提供電路（video rendering circuit) 220、一用於验者抑 +、， 用於將處理過之資料輸出至例如 一電視機（未示）之輸出230以及一作為根據本發明之電路之 一個具體貫施例用於即時資料處理之電路MO。 電路請包含-處理單元242(該處理單元242將予以說 明，係用於執行根據本發明之方法之具體實施例）、一 ROM 248(唯讀記憶體）（該職州將予以說明，係作為根 據本發明之資料載體之一個具體實施例用於儲存電腦可讀 取之資料’該電腦可讀取之資料包含用於程式化處理單元 242以執行根據本發明將予以說明之方法之具體實施⑷以 及-記憶體246，該記憶體246係用於儲存硬碟驅動器2ι〇 之-個模型。電路240最好亦包含一DMA(直接記憶體存 取）控制器244 ;該DMA控制器244對視訊提供電路22〇之系 統記憶體具有直接存取權且可用比處理單元242快很多的 速度執行硬碟驅動器210與視訊提供電路22〇之間的資料傳 送0 處理單元242最好在消費性電子裝置2〇〇啟動時從r〇m 248讀取電腦可讀取之資料，該資料包含電腦可執行之指 令，該電腦可執行之指令係用於程式化處理單元242以執 98002.doc •13· 200532446 行根據本發㈣μ說明之方法之具體實關。其次，硬 碟驅動器210之模型係讀自記憶體246。 電路240在此狀態已準傷好要處理視訊提供電路220所發 出之資料處理要求。在本發明之另—具體實施射，視訊 提供電路220可由-像是處理單元242之類的通用處理器予 以取代。根據本發明之裝置之其它具體實施例亦是有可能 的，如MP3播放器。 时在從視訊提供電路22G接收到諸處理要求之後，處理 單元42使用儲存在冗憶體246内之硬碟驅動器⑽之模型 對硬碟驅動器21G之最有效處理排程該等接收到的資料處 要长硬碟驅動杰21〇與視訊提供電路22〇之間的真實資 料料係經由DMA控制器244予以完成。該資料傳輸亦由 處理早70242予以監視。資料處理要求藉由硬碟驅動器㈣ 予以處理之速度尤其受到嚴密監視；這對硬碟驅動器210 之效能係一重要晉；p丨|。分，丨i、 ▲ 乂 资里成1依此方式，硬碟驅動器21〇之效能 蒼數係由處理輩亓940工、》胃 予以I測。在另一具體實施例中， 電路240包含一分離之量 里测早70 (未不），伹此工作同樣地恰 可藉由處理單元242予以執行。 效能參數中僅有一吾、、目及 里測係可行的，但在其它根據本發明 之方法之具體實施例中，如上所述，多重量㈣經完成且 一數值係從該組量測值導出。 在本發明之一個具體實施例中，處理單元242將量測到 勺(或導出k而視為置測到的)效能參數值與儲存在記憶體 246内之核型中所儲存之效能參數值作比較。當量測值以 98002.doc -14- 200532446 -預定值(絕對值或相對值)偏離儲存值時，儲存值則由量 測值予以取代。當不是這種狀況時，則像之前一樣維時原 值。在本發明之另-具體實施例中，儲存值總是由量測值 予以取代。 接著，儲存於記憶體246内之硬碟驅動器21〇之經過更新 之模型係用於排程資料處理要求。原始模型之備份最好係 保留在記憶體246内。原始模型最好為電路24〇、裝置_ 或儲存裝置2U)生錢_存之硬碟㈣&狀模型，或 者為一使用者於居家位置操作期間第一次量測之硬碟驅動 !§ 2 1 〇之模型。 在根，本發明之方法之另—具體實施例中，所有如上述 由處理早π 242執行之工作係由一個在視訊提供電路⑽上 執行之控制程式予以執行。本發明之意圖在於一處理單元 可予以用在執行根據本發明之方法。 _本發明之優點將藉由圖3A、圖3B、圖4入圖仙予以詳細 洗明。圖3A表示由一主機對一硬碟驅動器發出之大量資料 要求之服務時間之第-長條圖3〇〇。在水平軸上，服務時 間係以隨意之單位予以描繪；在垂直軸上標示要求之數 二。第-導條dead) 3K)包含所謂的快取命中㈣心仙）。 當主機要求之資料在硬碟驅動器的快取内找到，則主機可 非常快地得到資料。第—具总闽。λα ^ 貝寸十弟長條圖300之第一頂條（t〇p) 32〇 對第-命中標示中間服務時間；這些係所要求之資料可在 一個成功嘗試内予以導出之資料處理要求。 繞著頂條之數值變異對決定服務時間之方式係固有的， 98002.doc -15- 200532446 因為其包含硬碟之旋轉延遲。就此點之進一步資訊而言， 讀者得參考本說明底下提及的兩份文章。第一長條圖3〇〇 之第二頂條330包含在第二旋轉之後成功之嘗試。在這些 例子中，無法在第一次直接讀取資料。在此種狀況下，第 二嘗試係在硬碟之一個旋轉之後予以實施。第一尾條34〇 包含任何在兩次嘗試之後未成功之嘗試。 示於圖3B之第一長條圖350表示相同硬碟之相同資料， 但部疋在更嚴謹的環境下，例如與室溫度相比較高之溫 又已知在5亥等情況下，硬碟驅動器之服務時間增加，導 致效能降低。 &與第一導條310—樣大的第二導條36〇表示快取命中之效 月b未降低。代表第一次成功讀取嘗試之第三頂條確實 表^效能衰減，因為第三頂條37〇發生於一比第一長條圖 之第-頂條320還長之服務時間。服務時間已增加大約 2二至30%，一個相當大的量。同樣地，第二次成功讀取 y式用更長的時間發生’從第四頂條彻之位置可看出。 第二尾條390同樣地比第-尾條340長。總而言之，硬碟驅 動器之效能在新環境下已清楚地衰減。 、於一排㈣對硬碟闕器上之執行排程多重資料處理要 求之更高的H這會使服務之品f導致嚴重衰減。此係 猎由圖4A和圖4B予以表示。圖4a表示一個第一週期彻， 其包含第-執行之資料處理要求術、第二執行之資料處 理要求楊和第三執行之資料處理要求4G6。區塊之長度代 表執行該等資料處理要求之時間（durati〇n)。第一週期表示 _02.doc -16- 200532446 資料處理要求於標準環境之 之合適環;竟。 方'至皿/又有衝撞或震動 圖4B表示第二週期450 ’其代表在更惡劣之環境中對相 同之要求執行相同之資料處理要求。從-硬碟驅動哭至 資料傳輸速度已因提升之環境溫度和震動之;;的變 W兄荼數而降低3G%。原因有报多，例子有硬碟驅 與主機之間的連接頻寬減少以及硬碟驅動器因震動導致: =重產生之平均服務時間增加。處理速度降低 使執订_貝料處理要求所消耗之時間增加，如第四執 料處理要求412、第五執行之資料處理要求414和第六執行 之資料處理要求416的長度所標示。這三個區塊之長度係 增加30%’代表因較低3〇%之處理速度而多消耗观之時 間。 在第二週期中，第-指標422、第二指標似和第三指標 426代表-排程器使用一從製造商得來之硬碟驅動器新品 (fresh)之模型預期資料處理要求將要在—合適環境中結束 之時間點°當該排程器在—惡劣環境中使用該硬碟模型， 舉例第二要求僅有三分之—會予以處理，從圖街看出。 當硬碟驅動器如T13委員會對於設定硬碟ata標準所建議 -般地使用-可能性切斷一要求之執行時，這意指許多要 擷取之資料將不被擷取（在寫入的情況時也一樣）。 當這發生在硬碟係用於一錄影機時，播回之視訊資料串 流遲滯並有可能黑掉或者在記錄程式時導致硬碟上的視訊 資料遺漏。這對視訊錄放器之使用者造成高度困擾。另一 98002.doc -17- 200532446 方面/若使用更新之硬碟模型，則該_將不會發生。 本行人士將明顯地知道對所述具體實施例作各種修改而 不脫離本發明之範嘴係有可能的。各種所述由一元 行之工作可劃分成多重處理區塊，反之，經說明由多重元 件所執行的工作可由一處理區塊予以執行。 另外，本發明亦可應用於其它種與硬碟驅動器不同之記 fe體，如固態記憶體或一使用雷射（紅内、組、藍、紫或 务、外光或者其它顏色）之光碟驅動器。 ' 3 在圖2所呈現之具體實施例中，主機係由視訊提供電路 220和電路24G予以形成。各種對此主機系統架構之改變但 不脫離本發明之範疇係有可能的’本行人士將很快地察 知。 本發明係關於一種用於即時資料處理之方法及電路。隨 著硬碟和其它儲存裝置對即時應用的實現越來越多，效能 參數因其數值需用於資料之即時處理而越來越重要。隨著 忒等數值在可攜式裝置中的使用越來越頻繁，效能參數因 例如震動或溫度變化而偏差。因此，一具有最新效能參數 值貢訊用於以一儲存裝置可配合之速率連續即時並正確地 排程資料處理要求之主機系統係重要的。 [參考文章] C. Reummler and J. Wilkes, An introduction to disk drive modeling, Hewlett-Packard Laboratories Palo Alto CA, USA，IEEE Computer，Vol. 27, pp· 17-28，1994 J. Korst，V·The communication between the real-time file system 130 and the hard disk drive 150 is performed by the IDE driver 140. shell. The schedule made by the scheduler 138 in the material processing request is based on a mode 1 a λ &# u, & 160 of the hard disk drive 150. The model 160 contains at least one parameter of the hard disk drive 98002.doc 200532446 actuator 15, such as the hard disk drive. The hard disk drive 150 is transmitted to a * German / Monkey Beco from the hard disk has a system speed-or search time -Switching—there is a pick-up and the data is processed (read or written) by one of the hard disks, and the other position is at a time (average) of the speed required by the hard disk drive. Especially, when ^ #, this often changes during operation. For example, it is known that in a hot environment, the temperature of -70 ° C or more-银 八 # is already π, φ5, and the service time of the system now greatly increased. The operation of the system that uses the hard disk drive 15 as a storage device will be apparent in Shandan / Shirung floor shirts, which can also be clearly seen in the following description. Therefore, the present invention suggests The model 160 of the hard disk drive 更新 is updated during operation and the deviation in the performance parameter values of the model 考虑 is taken into account when scheduling data processing requests. This deviation will not always decrease. For this reason, a quantity is cited in the storage system software architecture 100 The measurement application program 170. The measurement application program 170 derives performance parameters from the hard disk drive 150. The measurement application program can process a set of multi-weight measurement values of a performance parameter and derive at least one value from the set of values. The derived value is regarded as a measurement value and the parameter value in the model 160 is updated with the value, so the derived value is used to update the parameter value in the model 160. In a specific embodiment, the derived value is the The middle value of the group of measured values. In another specific embodiment, the derived value is the average value of the group of measured values. In yet another specific embodiment, the derived value is the middle value of the group of measured values Plus mark The standard deviation or standard deviation is multiplied by a real number. The scheduler 138 uses the updated model when scheduling the next request. 60. In another embodiment of the method according to the present invention, the average is derived from one The so-called "moving window 98002.doc -12- 200532446 (moving wind〇w)" is known ^ ^ m ,. One advantage of this particular embodiment is the average value and thus the adaptation of the host system time. Many values gradually change. Fig. 2 shows a consumer electronic device ㈣ 'as a specific embodiment according to the present invention. The device includes a wrong hard disk 210, one for Luli: owing soil, 丨, picking, video rendering circuit 220, one for testers, +, for The processed data is output to, for example, the output 230 of a television (not shown) and a circuit MO for real-time data processing as a specific embodiment of the circuit according to the present invention. The circuit should include a processing unit 242 (the processing unit 242 will be described, which is used to implement a specific embodiment of the method according to the present invention), a ROM 248 (read-only memory) (this state will be explained, and it will A specific embodiment of the data carrier according to the present invention is used to store computer-readable data. The computer-readable data includes a specific implementation for the programmatic processing unit 242 to perform the method to be described in accordance with the present invention. And-a memory 246, which is used to store a model of the hard disk drive 2m. The circuit 240 preferably also includes a DMA (direct memory access) controller 244; the DMA controller 244 for video The system memory providing the circuit 22 has direct access and can perform data transfer between the hard disk drive 210 and the video providing circuit 22o at a much faster speed than the processing unit 242. The processing unit 242 is preferably in a consumer electronic device 2 〇〇 Reads computer-readable data from r〇m 248 at startup, which contains computer-executable instructions that are used to program the processing unit 242 According to 98002.doc • 13 · 200532446, the detailed practical method based on the method described in this publication. Second, the model of the hard disk drive 210 is read from the memory 246. The circuit 240 has been damaged in this state, and it is necessary to process the video. The data processing requirements issued by the circuit 220. In another embodiment of the present invention, the video providing circuit 220 may be replaced by a general-purpose processor such as the processing unit 242. Other specific embodiments of the device according to the present invention also It is possible, such as an MP3 player. After receiving processing requests from the video providing circuit 22G, the processing unit 42 uses the model of the hard disk drive 储存 stored in the redundant memory 246 to be the most effective for the hard disk drive 21G. Processing schedules The received data should be processed by the hard disk drive J21 and the video-providing circuit 22. The actual data is completed by the DMA controller 244. The data transmission is also monitored by the processing 70242. The speed at which data processing requests are processed by the hard disk drive 尤其 is particularly closely monitored; this is an important improvement on the performance of the hard disk drive 210; p 丨 |. 分丨 i, ▲ In this way, the performance of the hard disk drive 21 is measured by the processor 940, and measured by the stomach. In another specific embodiment, the circuit 240 includes a separate The measurement is 70 as early as possible (not yet), so this work can also be performed by the processing unit 242. Among the performance parameters, only 吾, 目, and 测 are feasible, but in other methods according to the present invention, In a specific embodiment, as described above, multiple weights are completed and a value is derived from the set of measured values. In a specific embodiment of the present invention, the processing unit 242 measures to a spoon (or derives k depending on The value of the performance parameter measured is compared with the value of the performance parameter stored in the karyotype stored in the memory 246. When the measured value deviates from the stored value by 98002.doc -14- 200532446-the predetermined value (absolute or relative value), the stored value is replaced by the measured value. When this is not the case, the original value is maintained as before. In another embodiment of the invention, the stored value is always replaced by the measured value. Then, the updated model of the hard disk drive 21 stored in the memory 246 is used for scheduling data processing requests. A backup of the original model is preferably kept in memory 246. The original model is best for the circuit 24, the device _ or the storage device 2U) to generate money _ stored hard disk ㈣ & shape model, or for a user to measure the hard disk drive for the first time during operation at home! § 2 1 〇 model. At the root, in another embodiment of the method of the present invention, all the tasks performed by the processing early 242 as described above are performed by a control program executed on the video providing circuit ⑽. The intention of the invention is that a processing unit can be used to perform the method according to the invention. _ The advantages of the present invention will be explained in detail by referring to FIG. 3A, FIG. 3B, and FIG. FIG. 3A shows the first bar graph 300 of the service time of a large number of data requests issued by a host to a hard disk drive. On the horizontal axis, service time is depicted in arbitrary units; on the vertical axis, the required number is two. Article-guide dead) 3K) contains the so-called cache hit ㈣ 心 仙). When the data requested by the host is found in the cache of the hard drive, the host can get the data very quickly. No.-with total Fujian. λα ^ The first top bar (t0p) of the 10th bar graph 300 (t0p) 32o indicates the intermediate service time for the first-hit; these are data processing requirements that can be derived in a successful attempt. The numerical variation around the top bar is inherent to the way the service time is determined, 98002.doc -15- 200532446 because it includes the rotation delay of the hard disk. For further information on this point, the reader is referred to the two articles mentioned at the bottom of this note. The second top bar 330 of the first bar graph 300 contains a successful attempt after the second rotation. In these examples, the data cannot be read directly the first time. In this case, the second attempt was made after a spin of the hard disk. The first tail bar 34o contains any unsuccessful attempts after two attempts. The first bar graph 350 shown in FIG. 3B shows the same information of the same hard disk, but the ministry is in a more rigorous environment, such as a higher temperature compared to the room temperature. Drive service time increases, resulting in reduced performance. & A second guide bar 36 as large as the first guide bar 310 indicates that the effectiveness of the cache hit month b has not decreased. The third top bar, which represents the first successful read attempt, does indicate a decrease in performance, because the third top bar 370 occurred at a longer service time than the top bar 320 of the first bar graph. The service time has increased by approximately 22 to 30%, a considerable amount. Similarly, the second successful reading of the y-form takes longer time 'can be seen from the position of the fourth top bar. The second tail bar 390 is also longer than the first-tail bar 340. All in all, the performance of hard disk drives has clearly degraded in the new environment. The execution of multiple data processing schedules on the hard disk drive in a row requires a higher H, which will cause the service product f to be seriously attenuated. This system is shown in Figures 4A and 4B. Figure 4a shows a first cycle, which includes the first-execution data processing technique, the second-execution data processing request Yang, and the third-execution data processing request 4G6. The length of the block represents the time (duratiOn) to execute these data processing requests. The first cycle represents _02.doc -16- 200532446 data processing requirements in the appropriate environment of the standard environment; actually. Figure 4B shows the second cycle 450 ', which represents the same data processing requirements for the same requirements in a more severe environment. From the hard disk drive to the data transmission speed has been reduced by 3G% due to the increased ambient temperature and vibration. There are many reasons, for example, the connection bandwidth between the hard disk drive and the host is reduced and the hard disk drive is vibrated: = The average service time of re-generation is increased. The reduced processing speed increases the time consumed by the order processing request, as indicated by the lengths of the fourth order processing request 412, the fifth execution data processing request 414, and the sixth execution data processing request 416. The increase of the length of these three blocks by 30% 'means that it consumes more time due to the lower processing speed of 30%. In the second cycle, the -indicator 422, the second indicator and the third indicator 426 represent-the scheduler uses a model of a new hard disk drive (fresh) from the manufacturer. It is expected that the data processing requirements will be in-suitable The end time point in the environment ° When the scheduler uses the hard disk model in a harsh environment, for example, the second requirement is only one third-it will be processed, as seen from Tujie. When a hard disk drive is used-as recommended by the T13 committee for setting the hard disk ATA standard-the possibility of cutting off a request implementation means that much of the data to be retrieved will not be retrieved (when written Same time). When this happens when the hard disk is used in a video recorder, the video data stream that is played back is sluggish and may hack or cause the video data on the hard disk to be missed while recording the program. This causes great distress to users of video recorders. Another 98002.doc -17- 200532446 aspect / If using the updated hard disk model, the _ will not happen. Those skilled in the bank will obviously know that it is possible to make various modifications to the specific embodiments without departing from the scope of the present invention. The various tasks described by a single line can be divided into multiple processing blocks. Conversely, the work performed by multiple components has been described as being performed by a single processing block. In addition, the present invention can also be applied to other types of hard disk drives, such as solid-state memory or a disc drive using laser (red inner, group, blue, purple, or external light or other colors). . '3 In the specific embodiment shown in FIG. 2, the host is formed by the video providing circuit 220 and the circuit 24G. Various changes to this mainframe system architecture that are possible without departing from the scope of the invention will be readily apparent to those in the bank. The invention relates to a method and a circuit for real-time data processing. As hard disks and other storage devices implement more and more real-time applications, performance parameters become more important because their values need to be used for real-time processing of data. As 忒 and other values are used more and more frequently in portable devices, performance parameters deviate due to, for example, vibration or temperature changes. Therefore, a host system with the latest performance parameter values is used to continuously and correctly schedule data processing requests in real time at a rate that can be matched by a storage device. [Reference Article] C. Reummler and J. Wilkes, An introduction to disk drive modeling, Hewlett-Packard Laboratories Palo Alto CA, USA, IEEE Computer, Vol. 27, pp. 17-28, 1994 J. Korst, V.
Pionk and P. Counians, Disk scheduling for variable-rate 98002.doc -18- 200532446 data streams, Philips Research Laboratories Eindhoven, The Netherlands，Proc. European Workshop on Interactive Distributed Multimedia Systems and Telecommunication Services，IDMSf97，1997. 【圖式簡單說明】 本發明已藉由圖示予以說明，其中： 圖1表示一用於即時資料處理之儲存系統軟體架構； 圖2表示-消費性電子裝置，作為根據本發明之裝置之 一個具體實施例； 圖3A和圖姆示兩個長條圖，用於描述根據本發明之 方法之一個具體實施例；以及 圖4A和圖4B表示兩個經過安排之資料處理要求 用於描述根據本發明之方法之一個具體實施例。 【主要元件符號說明】 100 儲存系統軟體架構 110 作業系統 120 應用程式 130 即時稽案系統 132 即時資料處理器 134 隶大效果資料處理 136 檔案系統 138 排程器 140 IDE驅動器 150 (貫體）硬碟驅動器 98002.doc -19- 200532446 160 170 200 210 220 230 240 242 244 246 248 300 310 320 330 340 350 360 370 380 390 400 402 404 模型 量測應用程式 消費性電子裝置 硬碟 視訊提供電路 輸出 電路 處理單元 直接記憶體存取控制器 記憶體 唯讀記憶體 第一長條圖 第一導條 第一頂條 第二頂條 第一尾條 第二長條圖 第二導條 第三頂條 第四頂條 第二尾條 第一週期 第一執行之資料處理要求 第二執行之貧料處理要求 98002.doc -20- 200532446 406 第三執行之資料處理要求 412 第四執行之資料處理要求 414 第五執行之資料處理要求 416 第六執行之資料處理要求 422 第一指標 424 第二指標 426 第三指標 450 第二週期 98002.doc -21 -Pionk and P. Counians, Disk scheduling for variable-rate 98002.doc -18- 200532446 data streams, Philips Research Laboratories Eindhoven, The Netherlands, Proc. European Workshop on Interactive Distributed Multimedia Systems and Telecommunication Services, IDMSf97, 1997. [Schematic Brief description] The present invention has been illustrated by diagrams, in which: FIG. 1 shows a software architecture of a storage system for real-time data processing; FIG. 2 shows a consumer electronic device as a specific embodiment of the device according to the present invention Figures 3A and 3B show two bar graphs for describing a specific embodiment of the method according to the present invention; and Figures 4A and 4B show two arranged data processing requirements for describing the method according to the present invention A specific embodiment. [Description of main component symbols] 100 Storage system software architecture 110 Operating system 120 Application program 130 Real-time auditing system 132 Real-time data processor 134 Large-scale effect data processing 136 File system 138 Scheduler 140 IDE driver 150 (Continuous) Hard disk Driver 98002.doc -19- 200532446 160 170 200 210 220 230 240 240 242 244 246 248 300 310 320 330 340 350 360 370 380 390 400 402 404 Model Measurement Application Consumer Electronics Hard Disk Video Provide Circuit Output Circuit Processing Unit Direct memory access controller memory read-only memory first bar chart first guide bar first top bar second top bar first tail bar second bar bar second guide bar third top bar fourth bar Article 2 Tail Article Data Processing Request for the First Execution in the First Cycle 98002.doc -20- 200532446 406 Data Processing Request for the Third Execution 412 Data Processing Request for the Fourth Execution 414 Fifth Execution Data processing request 416 Sixth implementation of data processing request 422 First indicator 424 Second indicator 426 Third indicator 450 Second cycle 98002.doc -21-