[go: up one dir, main page]

TWI242741B - Method for accessing frame data and data accessing device thereof - Google Patents

Method for accessing frame data and data accessing device thereof Download PDF

Info

Publication number
TWI242741B
TWI242741B TW93113043A TW93113043A TWI242741B TW I242741 B TWI242741 B TW I242741B TW 93113043 A TW93113043 A TW 93113043A TW 93113043 A TW93113043 A TW 93113043A TW I242741 B TWI242741 B TW I242741B
Authority
TW
Taiwan
Prior art keywords
memory
data
address
frame data
bank
Prior art date
Application number
TW93113043A
Other languages
Chinese (zh)
Other versions
TW200537369A (en
Inventor
Hua-Chang Chi
Original Assignee
Faraday Tech Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Faraday Tech Corp filed Critical Faraday Tech Corp
Priority to TW93113043A priority Critical patent/TWI242741B/en
Application granted granted Critical
Publication of TWI242741B publication Critical patent/TWI242741B/en
Publication of TW200537369A publication Critical patent/TW200537369A/en

Links

Landscapes

  • Image Input (AREA)

Abstract

A method for accessing frame data and data accessing device thereof are provided to access X-bit frame data. The method includes providing Y memories BANKi (0 < Y <= X), wherein BANKi represent ith memory bank (0 <= i < Y); arranging partial frame data WL,A (X/Y bits) in BANKj, wherein WL,A represent Lth line Ath frame data word and j=(L+A) mod Y; receiving and according to Y word addresses WAk to determine WL,A be located at which memory bank, wherein WAk represent address of kth frame data (0 <= k < Y); and obtaining the partial frame data (X/Y bits) from each BANKi, according to the determining result and combine them to the frame data (X bits).

Description

1242741 五、發明說明(1) 發明所屬之技術領^ 本發明是有關於一種資料存取的方法與裝置,且特別 是有關於一種圖框資料存取方法及其資料存取裝置。 先前技術 在動態補償視訊壓縮演算法(例如MPEG-1、MPEG-2、 MPEG-4等)中’需於圖框(frame)中依照移動向量(m〇ti〇rl vector)而去擷取參考方塊(reference bl〇ck)。基本方塊 (basic block)之尺寸通常為8*8或者16*16個像素 (pixel) ’由於移動向量於水平與垂直的擷取單位可能分 別比像素以及水平線多出半個像素大小,因此參考方塊之 擷取通常為9*9或者17*17個像素。圖1是顯示一般在搜尋 窗(search window) 1〇〇中擷取9*9參考方塊(例如圖中之 虛線框110)之範例。其中Pi」表示第i行第^個像素資料(8位 兀)。因為移動向量可能發生在搜尋窗中任何位置,因此 參考方塊110通常並不同於搜尋窗中之基本方塊範圍 (block boundary)(圖中粗線框120)。 假叹有6 4位元記憶體匯流排於每一時脈週期擷取基本 η圍12〇中一整行,亦即每次可以存取基本方塊範圍 120中8:固像素資料。參考方塊11〇中每一行涵蓋了二個基 塊她圍120,因此擷取9*9參考方塊u〇將需要9*2 = 18 由圖1中我們可以清楚看出所擷取的資料中 二疋不需要的。例如於第一行中共擷取了 料而·已,:,但是卻只需要使用P2,3、L、…、P21丨像素資 '已。擷取其他各行時亦有相同情形。因此,造成浪1242741 V. Description of the invention (1) The technical field to which the invention belongs ^ The present invention relates to a method and device for data access, and more particularly to a frame data access method and data access device. In the prior art, in dynamic compensation video compression algorithms (such as MPEG-1, MPEG-2, MPEG-4, etc.), 'the frame vector must be used to capture the reference according to the motion vector (m0ti〇rl vector). Reference block. The size of a basic block is usually 8 * 8 or 16 * 16 pixels (pixels). Because the capture unit of the motion vector in the horizontal and vertical directions may be half a pixel larger than the pixel and the horizontal line, so refer to the block. The capture is usually 9 * 9 or 17 * 17 pixels. Fig. 1 shows an example of extracting a 9 * 9 reference block (for example, a dashed box 110 in the figure) in a search window 100. Where "Pi" represents the ^ th pixel data of the i-th row (8 bits). Because the motion vector may occur anywhere in the search window, the reference block 110 is usually not the same as the basic block boundary in the search window (frame 120 in the figure). The fake sigh has 64-bit memory buses to capture a whole line of basic η around 120 in each clock cycle, that is, it can access 8: fixed pixel data in the basic box range 120 each time. Each line in the reference box 11〇 covers two basic blocks of her circumference 120, so to retrieve 9 * 9 reference box u〇 will require 9 * 2 = 18. From Figure 1, we can clearly see that the two data in the extracted data are two. Not required. For example, in the first line, a total of data is retrieved :, but only P2, 3, L, ..., P21 丨 pixel data is used. The same is true when fetching other rows. Therefore causing waves

1242741 五、發明說明(2) 記憶體匯流排頻寬之缺點。 發明内宏 本發明的目的4 S 士上日 節省記憶體存取Ϊί:,供一種圖框資料存取方法,以 本發明的Ϊ 增進整體系統效能。 諸目^ 目的是提供一種資料存取裝置,除前述 頻j因=了非必要之資料存取而可以操作於較 ^ ^ 功率消耗亦因此降低。 $ β ^提出一種圖框資料存取方法,以獲取具有X位 方二Ό圖框資料,其中Χ為正整數。此圖框資料存取 BANK·矣+楚、 首先提供Υ個記憶庫ΒΑΝΚί,其中 \ ^ \個記憶庫,Y為大於1並且小於等於X之整數, 八II 、吹4:^等於〇並且小於Y之整數。將具有χ/γ位元之部 存放於ΒΑ叫中,其&quot;U表示第L行第Α個部 Y”, : ^與^皆為大於等於〇之整數,卜(L + A) m〇d ’二為換數運算。然後接收並依據Y個字元位址WAk判斷 ^欲=取之^分圖框資料分別位於哪個記憶庫中,其中WA 不k個欲讀取部分圖框資料之位址,k為大於等於〇並 且小於Υ之整數。依據前述之判斷結果自記憶庫BANKi獲得 具有Χ/Υ位元之各部分圖框資料。各記憶庫BANKi所輸出之 部分圖框資料之組合即為欲讀取圖框資料。 2明另提出一種資料存取裝置,用以依據位址訊號 輸出具有X位元之一預儲存資料,其中乂為正整數。此資料 存取裝置包括記憶體控制器、γ個記憶庫以及多工組合電 路。記憶體控制器用以接收位址訊號並輸出γ個記憶庫位1242741 V. Description of the Invention (2) Disadvantages of memory bus bandwidth. Inner macro of the invention The purpose of the present invention is to save memory access time and provide a frame data access method to improve the overall system performance with the present invention. The purpose of each item ^ is to provide a data access device, in addition to the aforementioned frequency j, which can be operated at a lower power consumption because of unnecessary data access. $ β ^ Proposes a frame data access method to obtain frame data with X bits and squares, where X is a positive integer. This picture frame data access BANK · 矣 + Chu, first provide a memory bank ΑΝΚί, where \ ^ \ memory banks, Y is an integer greater than 1 and less than or equal to X, eight II, blowing 4: ^ equal to 0 and less than An integer of Y. Store the part with χ / γ bits in the ΒAA title, where "U represents the Lth row and the Ath part Y",: ^ and ^ are integers greater than or equal to 0, and (L + A) m〇 d 'two is a conversion operation. Then it receives and judges according to the Y character address WAk ^ desire = taken ^ which frame data is located in each memory bank, among which WA is not k to read some frame data Address, k is an integer greater than or equal to 0 and less than Υ. According to the foregoing judgment result, each part of frame data with X / Υ bits is obtained from memory bank BANK. A combination of part of frame data output by each bank BANKi That is to read the frame data. 2 Ming also proposed a data access device for outputting pre-stored data with one of X bits according to the address signal, where 乂 is a positive integer. This data access device includes memory Controller, γ memory bank and multiplexing combination circuit. The memory controller is used to receive address signals and output γ memory bank bits

1242741 五、發明說明(3) 址以及記憶庫判斷訊號,其中 、 整數。γ個記憶庫皆耦接至、為大於1並且小於等於X之 別接收對應之記憶庫位址其中,思體—控制器’任—記憶庫分 具有X/Y位元之部分預儲存、之一’並且分別輪出對應之 憶體控制器以及各記憶庫,、用、、。多工組合電路耦接至記 接收之具有X/Y位元之各部八以依據記憶庫判斷訊號將所 輸出為X位元之預儲存資料:3存貧^多I切換並組合 依據位址訊號判斷所欲讀取之預存控^器接收並 預儲存資料分別位於哪些記 存貝科所包3之各部分 為記憶庫判斷訊號。 思 並將其判斷結果輸出 广放ΐ ί =因將資料(例如為圖框資料、搜尋窗資料)分開 存放於不同之記恃、座中,Α 7士 —, 四只竹y刀開 ^ +4 Φ ^ ^ ^ ,ν ,7 而使母ζ人讀取資料時其所欲讀取 貝料中之各部分均可同日卑於士 ^ #丨7 a J J U呀從相對應之記憶庫中獲得,因此 之資料存取、節省記憶體存取頻寬進而增進 楣鱼、1此故而可以使記憶體之存取操作於較低之時 脈頻率’因此降低功率消耗。 為讓本發明之上述和其他目的、特徵和優點能更明顯 易懂,下文特舉較佳實施例,並配合所附圖式,作詳細說 明如下。 實施方式 圖2是依照本發明較佳實施例所繪示的一種圖框資料 存取方法之流程圖。請參照圖2,本實施例譬如用於視訊 處理(video process),尤其用於視訊處理中圖框參考方 塊(reference block of frame)之取得,以獲取具有X位1242741 V. Description of the invention (3) Address and memory judgment signal, where, are integers. The γ memory banks are all coupled to, and the corresponding memory bank addresses are greater than 1 and less than or equal to X. Among them, the mind-controller's-memory bank is partially pre-stored with X / Y bits. A ', and turn out the corresponding memory controller and each memory bank, use, and. The multiplexing combination circuit is coupled to each part with X / Y bits received in order to output the pre-stored data of X bits according to the judgement signal of the memory bank: 3 save the poor ^ more I switch and combine according to the address signal It is determined by which pre-storage controller which is to be read and which pre-stored data are respectively stored in each part of the package 3 which is a memory judgment signal. Thinking and outputting its judgment results widely. Ί = Because the data (such as frame data and search window data) are stored separately in different records and seats, Α 7 士 —, four bamboo knives open ^ + 4 Φ ^ ^ ^, ν, 7, so that when the mother ζ reads the data, all parts of the shell material he wants to read can be humbled by the same day ^ # 丨 7 a JJU 呀 obtained from the corresponding memory bank Therefore, data access, memory access bandwidth is saved, and the sturgeon is improved. Therefore, the memory access operation can be performed at a lower clock frequency, thereby reducing power consumption. In order to make the above and other objects, features, and advantages of the present invention more comprehensible, preferred embodiments are described below in detail with reference to the accompanying drawings, as follows. Embodiment FIG. 2 is a flowchart of a method for accessing frame data according to a preferred embodiment of the present invention. Please refer to FIG. 2. This embodiment is used, for example, for video processing, and particularly for obtaining a reference block of frame in video processing to obtain an X-bit

12427411242741

圖框資料。其中,X為正整數。該圖框資料 BANK 。'1匕下列各步-驟。步驟S2 1 〇,提供Y個記憶庫 等於、β: ’BANKi表示第H固記憶庫,γ為大於1並且小於 萼於X之正數,i為大於等於〇並且小於¥之整數。 S220,將圖框資料Wla存放於記憶庫βανκ〗中,並中^表示 =于第A個部分圖框資料(χ/γ位元),L射皆為大^於〇 正 ,J —(L + A) mod Y(其中mod為模數運算)。步驟 S230,接收並依據γ個字元位址W4判斷所欲讀取之部分圖 框資料分別在哪個記憶庫中。其中表示第}^個欲讀取 部分圖框資料之位址,k為大於等於〇並且小之整數。 步驟S240,依據步驟S23〇之判斷結果自各記憶庫^㈣獲得 欲讀取圖框資料。 1 上述步驟S240可參照下列各步驟實施之。步驟S241 , 依據步驟S230之判斷結果產生γ個記憶庫位址,其中 表示第i個記憶庫之存取位址。步驟S242,依據記憶庫位1 址BAi存取記憶庫BANKi。步驟S243 ,自各記憶庫BANKi獲得 對應之部分圖框資料。步驟S244,依據字元位址判斷各 記憶庫BANL所輸出部分圖框資料(χ/γ位元)之排列順序並 依此組合輸出為欲讀取圖框資料(X位元)。Frame information. Where X is a positive integer. The picture frame information BANK. '1 Dagger following each step-step. In step S2 1 0, Y banks are provided. Β: ′ BANKi represents the H-th memory bank, γ is a positive number greater than 1 and less than X, and i is an integer greater than or equal to 0 and less than ¥. S220, storing the frame data Wla in the memory bank βανκ〗, and the middle ^ indicates = in the A part of the frame data (χ / γ bit), the L shots are all larger than 0, J — (L + A) mod Y (where mod is a modulo operation). Step S230: Receive and judge in which memory part of the frame data to be read according to the γ character addresses W4. Among them, it represents the address of the frame data to be read, and k is an integer greater than or equal to 0 and smaller. In step S240, the frame data to be read is obtained from each memory bank according to the judgment result in step S23. 1 The above step S240 can be implemented with reference to the following steps. In step S241, according to the judgment result in step S230, γ memory bank addresses are generated, where the access addresses of the ith memory bank are indicated. In step S242, the bank BANKi is accessed according to the bank position 1 address BAi. Step S243: Obtain corresponding frame data from each memory bank BANKi. In step S244, the arrangement order of some frame data (χ / γ bit) output from each bank BANL is judged according to the character address, and the combination is output as the frame data (X bit) to be read.

綜上所述,在此假設系統記憶體匯流排(mem〇ry bus) 計有6 4位元’並且使用2個記憶庫以儲存圖框資料。換句 活說’即假a又X專於6 4 ’而Y為2。因此,每個記憶庫各自 輸出3 2位元之部分圖框資料。圖3是依照本發明較佳實施 例所繪示的一種搜尋窗(search wi ndow)中使用2個記憶庫To sum up, it is assumed here that the system memory bus (memory bus) counts 64-bits' and uses 2 banks to store frame data. In other words, let's say ‘that false a and X are specific to 6 4’ and Y is 2. Therefore, each memory bank outputs a part of 32-bit frame data. FIG. 3 is a diagram illustrating a search window (search window) using two memories according to a preferred embodiment of the present invention.

1242741 五、發明說明(5) BANKG與BANK!的資料結構之範例。社 尋窗300之大小為64*48個像辛,、圖3,在此假設搜 字元,其包含4個像素資:本列:,、·譬如為32位元 P2, 1、?2, 2、P2, 3等像素資料。因此1242741 V. Description of the invention (5) Examples of the data structure of BANKG and BANK! The size of the search window 300 is 64 * 48 images, as shown in Figure 3. Here, it is assumed that the search character contains 4 pixel data: this column: ,, for example, 32-bit P2, 1 ,? 2, 2, P2, 3 and other pixel data. therefore

貝才叶例如,%,〇即包含有如圖1中P 2, 參考方塊(reference bloc]〇 u「^在中所欲擷取之 方塊31〇中。 ⑻110即被包含於圖3中之虛線 ㈣本使用二個記憶庫,然而卻不應以此限制本 βανΓΛ中流排具有X位元’則可使用⑽記憶庫 元):而” Γα :整數(通常為2的幂次方,例如64位 :ί 並且小於等於Χ之整數(例如為2、4、8 於Υ之整I 1表示第1個記憶庫’丨為大於等於0並且小 同之ΪΓ座V相鄰之部分圖框資係存放於不 則W 例如’若^存放於第0個記憶細NK〇中, ^,tk 4BANKl &quot; ° ^ ^For example, %%, 〇 is included in P 2, reference block (reference bloc), as shown in FIG. 1, and ^ is in block 31, which is to be retrieved. ⑻110 is included in the dotted line in FIG. 3 This uses two memory banks, but this should not be used to limit this βανΓΛ to have X-bits. 'You can use ⑽ memory cells): and "Γα: integer (usually a power of two, for example, 64-bit: ί and an integer less than or equal to X (for example, 2, 4, 8 and the whole I 1 represents the first memory bank '丨 is greater than or equal to 0 and the part of the frame adjacent to the block 座 Γ of V is stored in Otherwise, for example, 'if ^ is stored in the 0th memory NK〇, ^, tk 4BANKl &quot; ° ^ ^

如,若w&quot;户1;斗i,j與%+1,〗亦須存放於不同之記憶庫中。例 中。Itr 記憶庫8通°中,則W3,。存放於記憶庫BAW i資將具有χ/γ位元(例如為32位元)之部分圖 之ΪΪ,Α目 記憶庫BAMj中,其中皆為大於等於〇 、、卜(L + A) mod Y。前式中mod為模數運算。 圖1之於I習#知技術中,若欲透過64位元記憶體匯流排擷取如 施例,\ ’需要9*2=18個時脈週期。若依本實 /、而掏取圖3之虛線方塊3 1 0即可。例如,於第J時 13203twf.ptd 第10頁 1242741 五、發明說明(6) &quot; -------- =週期令擷取Wu (從BANK〇取得)與(從βΑΝΚι取得)· 時脈週期中擷取町2 (從BANKq取得)與^。(從βΑΝΚι取' 传),於第3時脈週期中擷取WS1 (從BANKg取得)與^ BANK取得),以此類推。每一時脈週期中均可同時各= 憶庫BANKi中獲取欲取得之部分圖框資料。最後,於第w日士 脈週期中擷取W1G,G (從BANKQ取得)與wi(M (從bANKi取得);才 於第14時脈週期中擷取〜2 (從BANKq取得)。因此,本每 施例只須14個時脈週期即可完成參考方塊丨1()之擷取工只 =。所以,本實施例明顯改善了習知技術中浪費記憶體匯 沭排頻寬之缺點,進而加速參考方塊之讀取效率。 ^在此另舉一實施例以說明本發明。本實施例與前述實 施例相似,其不同之處在於使用4個記憶庫以輸出64位元 之部分圖框資料。換句話說,即假設乂等於64,而γ為4。 ^此,每個記憶庫各自輸出16位元之部分圖框資料。圖4 是依照本發明另一較佳實施例所繪示的一種搜尋窗 (search window)中使用4個記憶庫BANK0至3八皿3的資料結 構之範例。請參照圖4,在此亦假設搜尋窗4〇〇之大小為 6 4*48個像素資料。為與圖3之32位元部分圖框資料% j作區 別,圖4中以札,〗表示第i行第j個部分圖框資料(16位元)。 於本實施例中,譬如包含2個像素資料。例如,h2,q即包 含有如圖1中匕,Q、匕,1等像素資料。本實施例只須1 2個時脈 週期即:完成9*9參考方塊之擷取工作。本實施例之詳細 操作與前述實施例相似,凡熟習此藝者可由前述實施例中 類推獲知,故不在此贅述。For example, if w &quot; house 1; bucket i, j, and% + 1, it must also be stored in a different memory bank. Example. Itr memory 8-way, then W3 ,. Stored in the memory bank BAW i will have part of the map with χ / γ bits (for example, 32 bits), in the A-memory bank BAMj, all of which are greater than or equal to 0, and (L + A) mod Y . In the previous formula, mod is a modulo operation. In Figure 1 of the Ix # technique, if you want to capture through a 64-bit memory bus as in the embodiment, \ 'requires 9 * 2 = 18 clock cycles. According to the actual situation, the dotted box 3 1 0 in FIG. 3 can be obtained. For example, at the time of J 13203twf.ptd page 10 1242741 V. Description of the invention (6) &quot; -------- = Period order fetch Wu (obtained from BANK〇) and (obtained from βΑΝΚι) · Hours During the pulse cycle, machi 2 (taken from BANKq) and ^ are extracted. (Take the pass from βΑΝΚι), extract WS1 (obtained from BANKg) and ^ BANK in the third clock cycle, and so on. Each clock cycle can be simultaneously = Get some of the frame data you want to obtain in the memory bank BANKi. Finally, W1G, G (obtained from BANKQ) and wi (M (obtained from bANKi)) are extracted during the wirth cycle on the wth day; only ~ 2 (obtained from BANKq) are acquired during the 14th clock cycle. Therefore, In this embodiment, only 14 clock cycles are required to complete the fetching process of the reference block 1 (). Therefore, this embodiment significantly improves the disadvantage of wasting memory bandwidth in the conventional technology. This further accelerates the reading efficiency of the reference block. ^ Another embodiment is used to illustrate the present invention. This embodiment is similar to the previous embodiment, except that it uses 4 memory banks to output a portion of the frame of 64 bits In other words, it is assumed that 乂 is equal to 64 and γ is 4. ^ Each memory bank outputs a 16-bit portion of the frame data. Figure 4 is a drawing according to another preferred embodiment of the present invention. An example of a data structure using 4 memory banks BANK0 to 3 and 8 in a search window of the. Please refer to FIG. 4, and also assume that the size of the search window 400 is 6 4 * 48 pixel data To distinguish from the frame data% j of the 32-bit part of the frame in FIG. 3, in FIG. j partial frame data (16 bits). In this embodiment, for example, it contains 2 pixel data. For example, h2, q contains pixel data such as dagger, Q, dagger, 1 in this example. This embodiment It only needs 12 clock cycles to complete the extraction of 9 * 9 reference blocks. The detailed operation of this embodiment is similar to that of the previous embodiment. Those skilled in this art can be obtained by analogy in the previous embodiment, so it will not be repeated here. .

1242741 五、發明說明(7) 綜合前述,茲以存取搜尋窗資料為例將本發明之資料 存取方法及資料結構與習知技術作一比較,比較結果如圖 5所示。圖5是本發明與習知技術之資料讀取效能比較表。 由圖5可知,越多記憶庫(越小資料寬度)將有越佳讀取效 能01242741 V. Description of the invention (7) In summary, the data access method and data structure of the present invention are compared with the conventional techniques by taking the search window data as an example. The comparison result is shown in FIG. 5. FIG. 5 is a comparison table of data reading performance of the present invention and the conventional technology. As can be seen from Figure 5, the more memory banks (the smaller the data width), the better the reading performance.

在此依照本發明再舉一實施例,如圖6所示。圖6是依 照本發明較佳實施例所繪示之一種資料存取裝置方塊圖。 此資料存取裝置用以依據位址訊號addr輸出具有X位元之 預儲存資料(例如是圖框資料或搜尋窗資料)rdata。記憶 體控制器61 0接收位址訊號addr、讀取要求req — r、寫入要 求req —w以及寫入資料data — w,並且輸出Y個記憶庫位址 b〇 — addr至bY-1—addr、記憶庫致能訊號CSO至CSY-1、讀寫 控制訊號r/w、寫入資料bO一data一w至bY-1 一data — w以及記 隐庫判斷说號BS。其中’ X與Y之定義與前述實施例相同。Here is another embodiment according to the present invention, as shown in FIG. 6. FIG. 6 is a block diagram of a data access device according to a preferred embodiment of the present invention. This data access device is used to output pre-stored data (such as frame data or search window data) rdata with X bits according to the address signal addr. The memory controller 61 receives the address signal addr, the read request req — r, the write request req — w, and the write data data — w, and outputs Y memory bank addresses b 0 — addr to bY-1 — addr, memory enable signals CSO to CSY-1, read-write control signals r / w, write data bO_data_w to bY-1_data_w, and memory bank judgement signal BS. Among them, the definitions of X and Y are the same as those in the foregoing embodiment.

記憶庫BANL SBANKw耦接至記憶體控制器。於本 實施例中,例如將搜尋窗資料依照前述實施例之資料結構 分開存放於記憶庫BANK。至ΒΑΝΚη中。每個記憶庫BANKq至 bank^分別接收對應之記憶庫位址、記憶庫致能訊號^S()至 CSY-1、讀寫控制訊號r/w以及寫入資料b〇一data — w至 bY-1—data —w,以儲存搜尋窗資料,或者各自輸—出對應之 部分預儲存資料b0 —data — r至bY-1 —data — r (X/Y位元)。 其中,記憶體控制器610係接收並依據位址訊號^計 =斷所欲讀取之預儲存資料rdata所包含之各部分預儲存 貝料分別位於記憶庫中之何者,並將其判斷結果輸出為記The memory bank BANL SBANKw is coupled to the memory controller. In this embodiment, for example, the search window data is stored separately in the memory bank BANK according to the data structure of the foregoing embodiment. Into BANKK. Each bank BANKq to bank ^ receives the corresponding bank address, bank enable signal ^ S () to CSY-1, read / write control signal r / w, and write data b〇 一 data — w to bY -1—data —w, to store the search window data, or each output—output the corresponding pre-stored data b0 —data — r to bY-1 —data — r (X / Y bits). Among them, the memory controller 610 receives and determines according to the address signal ^ = the pre-stored data rdata to be read, where each part of the pre-stored material is located in the memory, and outputs its judgment result. For the record

1242741 五、發明說明(8) 憶庫判斷訊號BS。多工組合電路62 0耦接至記憶體_ 610以及記憶庫BANK。至ΒΑΝΚη,用以依據記憶庫判&amp; = BS將所接收具有χ/γ位元之部分預儲存資料加以多工^ = 並組合輸出為X位元之預儲存資料rdata (本實施例中链、 是圖框資料或搜尋窗資料)。 &amp; ° 為能更清楚說明本發明,以下假設經由系統記情體匯 流排所讀取之預儲存資料rdata計有64位元,並且使用2個 記憶庫以儲存搜尋窗資料。換句話說,即於本實施例中假 設X等於64,而Y為2。因此,每個記憶庫各自輸出“位元又 之=分搜尋窗資料,如圖7A所示。圖7A是依照本發明較佳 實施例所繪示之一種使用二個記憶庫之資料存取裝置方塊 圖。 4參照圖7A,其中例如將搜尋窗資料依照前述實施例 中圖3之資料結構分開存放於記憶庫BANKG以及BANK!中。位 址產生器AG產生讀取要求req —r、讀取位址訊號addr —r〇以 及addr — rl以便操取對應之第一字元(w〇r(j 〇)與第二字 =二經由寫入要求req — w、寫入位址訊號addr — w以及寫入 資=d^ta — w而使外部電路更新記憶庫BANKq以及Μ·〗中之 搜尋_資料。於本實施例中,讀取位址訊號addr_r〇、 add^—rl以及寫入位址訊號addr — w例如皆為〗〇位元,而第 尸子元第一子元與寫入資料data 一w例如皆為32位元(若 每一像素資料為8位元,則其包含有4個像素資料)。 記憶體控制器710用以仲裁讀取要求與寫入要求,並 且分別產生記憶庫BANKq與ΒΑΝΚι所需之讀寫控制訊號r/w、 i^· 13203twf.ptd 第13頁 12427411242741 V. Description of the invention (8) The memory judges the signal BS. The multiplexing combination circuit 620 is coupled to the memory_610 and the memory bank BANK. To ΒΑΝΚη, used to judge &amp; = BS to multiplex the received part of pre-stored data with χ / γ bits ^ = and combine to output X-bit pre-stored data rdata (chain in this embodiment , Is frame data or search window data). In order to explain the present invention more clearly, the following assumes that the pre-stored data rdata read by the system memory bus has 64 bits, and uses 2 memories to store the search window data. In other words, it is assumed that X is equal to 64 and Y is 2 in this embodiment. Therefore, each memory bank outputs “bit-wise = minute search window data, as shown in FIG. 7A. FIG. 7A is a data access device using two banks according to a preferred embodiment of the present invention. Refer to FIG. 7A, for example, the search window data is stored separately in the memory banks BANKG and BANK! According to the data structure of FIG. 3 in the foregoing embodiment. The address generator AG generates a read request req —r, read Address signals addr — r0 and addr — rl in order to manipulate the corresponding first character (w〇r (j 〇) and the second word = two via write request req — w, write address signal addr — w And write data = d ^ ta — w to cause the external circuit to update the search_data in the memory banks BANKq and M ·. In this embodiment, the address signals addr_r0, add ^ _rl, and the write bit are read. The address signal addr — w is, for example, 0 bits, and the first child of the first child and the written data data are w, for example, 32 bits (if each pixel data is 8 bits, it contains 4 pixel data). The memory controller 710 is used to arbitrate read requests and write requests, and Do not generate memory read and write the desired BANKq and ΒΑΝΚι control signal r / w, i ^ · 13203twf.ptd Page 131242741

五、發明說明(9)V. Description of the invention (9)

ά己fe、庫致成5虎C S 0與C S1以及記憶庫位址b 〇 — a d d r與 b 1 一addr。記憶體控制器71 0亦產生記憶庫判斷訊號“以指 出第一字元係位於各記憶庫中之何者。例如,當BS = 〇即表 示第一字元係位於記憶庫BANKQ中,若BS=1則表示第一字元 位於記憶庫BANK〗中。由圖3所示之資料結構可以明顯看 出,第一字元與苐二字元之擷取必定是來自不同之記憶 庫。也就是說,當第一字元係位於記憶庫BANK。,則第二字 元位於記憶庫BANK!;反之,若當第一字元係位於記憶庫 BANK!,則第二字元位於記憶庫BANKQ。每一記憶庫之輸出 bO一data—r與bl—data — r(皆為32位元)將經由多工組合電路 7 2 0 (依照記憶庫判斷訊號b s )加以切換組合為欲讀取之搜 哥窗資料rdata( 64位元)。此搜尋窗資料rdata例如可以提 供視訊處理中之動態補償電路ME所使用。άfe, Ku Zhicheng 5 tigers C S 0 and C S1, and memory address b 〇 — a d d r and b 1 an addr. The memory controller 71 0 also generates a memory judgment signal "to indicate which of the first character system is located in each memory bank. For example, when BS = 0, it means that the first character system is located in bank BANKQ. If BS = 1 means that the first character is located in the memory bank BANK. From the data structure shown in Figure 3, it can be clearly seen that the first character and the second character must be extracted from different memories. That is to say When the first character line is located in the bank BANK., The second character is located in the bank BANK! Conversely, if the first character line is in the bank BANK !, the second character is located in the bank BANKQ. Every The output of a memory bank bO-data-r and bl-data-r (both are 32-bit) will be switched and combined into the search brother to be read via the multiplexing combination circuit 7 2 0 (based on the judgement signal bs of the memory bank). Window data rdata (64-bit). This search window data rdata can be used, for example, by a dynamic compensation circuit ME in video processing.

在此,上述之記憶體控制器71 0譬如可以參照圖7 B實 施之。圖7 B是依照本發明較佳實施例所繪示圖7 a中之一種 記憶體控制器710方塊圖。讀取位址訊號addr_rO、 ad dr — r 1以及寫入位址訊號add r—w經過多工器711與712(依 照讀取要求req一r及寫入要求req — w )切換以產生第一字元 位址w0 —addr與第二字元位址wl一addr。於本實施例中,例 如將第一字元位址w〇 — addr耦接至判斷電路71 3以產生記憶 庫判斷訊號bs。第一字元位址w0一addr與第二字元位址 wl—addr經由切換電路714依照記憶庫判斷訊號bs分別切換 輸出為記憶庫ΒΑΝΚ0與BANK1所需之記憶庫位址b0_addr與 bl—addr。例如,當bs = 0時,表示第一字元位於記憶庫Here, the above-mentioned memory controller 710 can be implemented by referring to FIG. 7B, for example. FIG. 7B is a block diagram of a memory controller 710 in FIG. 7a according to a preferred embodiment of the present invention. The read address signal addr_rO, ad dr — r 1 and the write address signal add r — w are switched by the multiplexers 711 and 712 (in accordance with the read request req-r and the write request req — w) to generate the first Character address w0 —addr and the second character address wl_addr. In this embodiment, for example, the first character address w0-addr is coupled to the judgment circuit 71 3 to generate a memory judgment signal bs. The first character address w0_addr and the second character address wl_addr are respectively switched and output to the memory addresses b0_addr and bl_addr required by the memory banks ΑΝΚ0 and BANK1 through the switching circuit 714 according to the memory judgment signal bs. . For example, when bs = 0, the first character is in the memory

13203twf.ptd 第14頁 1242741 五、發明說明(ίο) -- BANKG中,因此將第一字元位址W0一addr耦接輸出為記情庫 位址b0_addr,而將第二字元位址wl一addr轉接輸出為^己^ 庫位址bl—addr。反之,若bs = l則表示第一字元位於記情“ 庫BANI中,因此將第一字元位址w0一addr搞接輪出為—己忙 庫位址bl—addr,而將第二字元位址wl-addr||接輸出為士己 憶庫位址bO_addr。 切換電路714例如由多工器714a與714b所組成。其 中,多工器71 4a依據判斷訊號bs選擇第一字元位址八 w0 —addr以及第二字元位址wl一addr二者之一以輸出為記憶 庫位址b0 一addr。而多工器714b與多工器714a類似,其不 同在於若多工^|714a將第一字元位址w〇一addr輸出為記憒 庫位址1)0 — 8(1(11'時,則多工器7141)將第二字元位址“—&amp;(1(^ 輸出為記憶庫位址bl—addr ’以此類推。判斷訊號bs再經 由延遲電路71 5緩衝後輸出為判斷訊號bs。因為記憶庫執 行讀取指令時往往需要數個時脈週期(依照所採用之記憶 體形態之不同,其所需之時脈週期亦有所不同)才能輸出 所需資料’因此利用延遲電路71 5以配合於記憶庫之輸出 時序。 於本實施例中’判斷電路71 3例如可以參照圖7 c實施 之。圖7C是依照本發明較佳實施例所繪示圖7B中之一種判 斷電路713之電路圖。請同時參照圖3以及圖7C,由圖3可 以看出,字元位址w0一addr中之第〇位元(以w〇-addr[〇]表 不)與第4位元(以w0 —addr[4]表示)若同時為〇(或丨),則該 第一字元位址w0一addr所對應之參考窗資料(圖框資料)係13203twf.ptd Page 14 1242741 V. Description of the invention (ίο)-In BANKG, the first character address W0-addr is coupled and output as the memory bank address b0_addr, and the second character address wl An addr transfer output is ^ self ^ library address bl_addr. Conversely, if bs = l, it means that the first character is located in the memory "bank BANI", so the first character address w0-addr will be rotated out as-busy library address bl-addr, and the second character The character address wl-addr || is connected to the output of the memory address bO_addr. The switching circuit 714 is composed of, for example, multiplexers 714a and 714b. Among them, the multiplexer 71 4a selects the first character according to the determination signal bs Address eight w0 —addr and one of the second character addresses wl-addr are output as the memory address b0-addr. The multiplexer 714b is similar to the multiplexer 714a, except that the multiplexer ^ | 714a outputs the first character address w0_addr as the memory bank address 1) 0 — 8 (1 (11 ', then the multiplexer 7141) sets the second character address “— &amp; ( 1 (^ output is the memory address bl-addr 'and so on. The judgment signal bs is buffered by the delay circuit 71 5 and the output is judged signal bs. Because the memory bank often needs several clock cycles ( (Depending on the type of memory used, its required clock cycle is also different) in order to output the required data 'so the delay circuit 71 5 is used to match Combined with the output timing of the memory bank. In this embodiment, the 'judging circuit 713' can be implemented, for example, with reference to Fig. 7c. Fig. 7C is a circuit diagram of a judging circuit 713 shown in Fig. 7B according to a preferred embodiment of the present invention. Please refer to FIG. 3 and FIG. 7C at the same time. It can be seen from FIG. 3 that the 0th bit (denoted by w0-addr [〇]) and the 4th bit (denoted by w0 of the character address w0-addr). —Addr [4] indicates) If both are 0 (or 丨), the reference window data (frame data) corresponding to the first character address w0-addr is

13203twf.ptd 第15頁 1242741 五、發明說明(11) 存放於記憶庫BANK。。反之,若w〇 —addr[4]不 同時’則該第一字元位址wO一ad dr所對應之參考窗資料(圖 框資料)係存放於記憶庫BANK!。因此,判斷電路6 1 3即可以 簡單之互斥或閘X0R完成之。 於本實施例中,多工組合電路72〇例如可以參照圖7D 貫施之。圖7D是依照本發明較佳實施例所繪示圖7A中之一 種多工組合電路7 2 0之方塊圖。請參照圖7 D,其中13203twf.ptd Page 15 1242741 V. Description of the invention (11) Stored in memory bank BANK. . Conversely, if w0-addr [4] is different, the reference window data (frame data) corresponding to the first character address wO_ad dr is stored in the memory bank BANK !. Therefore, the judgment circuit 6 1 3 can be completed by simply mutating the OR gate X0R. In this embodiment, the multiplexing combination circuit 72 may be implemented by referring to FIG. 7D, for example. FIG. 7D is a block diagram illustrating a multiplexing combination circuit 7 2 0 in FIG. 7A according to a preferred embodiment of the present invention. Please refer to Figure 7D, where

r data [ 63: 32 ]表示搜尋窗資料r data中第32至63位元資 料,同理,rdata[31 : 0]表示搜尋窗資料rdata中第〇至31 位元資料。所以,獲得64位元之搜尋窗資料rdata以提供 下一級電路(例如動態補償電路)做進一步處理。於圖7 a中 之記憶庫BANK0與BANK所輸出資料b〇 —data_r與 bl—data一r (皆為32位元)將連接至多工器721與722。多工 器7 2 1依照記憶庫判斷訊號BS (由記憶體控制器7丨〇所產 生)選擇資料b0 一data一r與bl 一data —r二者中為第一字元者 輸出為搜尋窗資料rdata[63:32]。反之,多工器722依照 記憶庫判斷訊號BS選擇資料b0一data — r與bl—data — r二者中 為第二字元者輸出為搜尋窗資料rdata [31 : 0]。例如,當 BS = 0時,則多工器721選擇將資料b0 — data一r輸出為搜尋窗r data [63: 32] represents the 32th to 63th bits of data in the search window data r data. Similarly, rdata [31: 0] represents the 0th to 31st bits of data in the search window data rdata. Therefore, the 64-bit search window data rdata is obtained to provide the next-level circuit (such as a dynamic compensation circuit) for further processing. The data b0 —data_r and bl —data—r (both 32 bits) output by the memory banks BANK0 and BANK in FIG. 7a will be connected to the multiplexers 721 and 722. The multiplexer 7 2 1 selects data b0-data-r and bl-data-r according to the memory judgment signal BS (produced by the memory controller 7 丨 〇) and outputs it as the search window. Data rdata [63:32]. Conversely, the multiplexer 722 selects the data b0_data_r and bl_data_r as the second character according to the memory judgment signal BS and outputs it as the search window data rdata [31: 0]. For example, when BS = 0, the multiplexer 721 chooses to output data b0 — data_r as a search window.

資料rdata[ 63:32 ],並且多工器72 2選擇將資料bl—data_r 輸出為搜尋窗資料rdata [31 : 0];反之,若BS = 1,則多工 器721選擇將資料bl_data一r輸出為搜尋窗資料 rdata[ 63:32 ],並且多工器722選擇將資料b0 一 data_r輸出 為搜尋窗資料rdata[31:0]。Data rdata [63:32], and multiplexer 72 2 chooses to output data bl_data_r as search window data rdata [31: 0]; otherwise, if BS = 1, multiplexer 721 chooses to output data bl_data_r The output is search window data rdata [63:32], and multiplexer 722 chooses to output data b0-data_r as search window data rdata [31: 0].

1242741 五、發明說明(12) 雖然本發明已以較佳實施例揭露如上,然其並非用以 限定本發明,任何熟習此技藝者,在不脫離本發明之精神 和範圍内,當可作些許之更動與潤飾,因此本發明之保護 範圍當視後附之申請專利範圍所界定者為準。 ❿1242741 V. Description of the Invention (12) Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make some changes without departing from the spirit and scope of the present invention. Changes and retouching, therefore, the protection scope of the present invention shall be determined by the scope of the appended patent application. ❿

13203twf.ptd 第17頁 1242741 圖式簡單說明 圖1是顯示一般在搜尋窗(search window)中擷取9*9 參考方塊(圖中之虛線框)之範例。 圖2是依照本發明較佳實施例所繪示的一種圖框資料 存取方法之流程圖。 圖3是依照本發明較佳實施例所繪示的一種搜尋窗中 使用二個記憶庫的資料結構之範例。 圖4是依照本發明另一較佳實施例所緣示的一種搜尋 窗(search window)中使用4個記憶庫BANKq SBANK3的資料 結構之範例。 圖5是本發明與習知技術之資料讀取效能比較表。 圖6是依照本發明較佳實施例所繪示之一種資料存取 裝置方塊圖。 圖7 A是依照本發明較佳實施例所繪示之一種使用二個 吕己憶庫之資料存取裝置方塊圖。 圖7 B是依照本發明較佳實施例所繪示圖7 A中之一種—己 憶體控制器方塊圖。 ° 圖7C是依照本發明較佳實施例所繪示圖7B中之一種 斷電路之電路圖。 圖7 D是依照本發明較佳實施例所繪示圖7 A中之—種多 工組合電路之方塊圖。 夕 【圖式標不說明】 100 :習知搜尋窗(search window)資料之結構 11 0 : 9*9參考方塊 1 2 0 :擷取基本方塊範圍13203twf.ptd Page 17 1242741 Brief description of the figure Figure 1 shows an example of a 9 * 9 reference box (the dashed box in the figure) that is generally taken in the search window. FIG. 2 is a flowchart of a frame data access method according to a preferred embodiment of the present invention. FIG. 3 is an example of a data structure using two memories in a search window according to a preferred embodiment of the present invention. FIG. 4 is an example of a data structure of four memory banks BANKq SBANK3 in a search window according to another preferred embodiment of the present invention. FIG. 5 is a comparison table of data reading performance of the present invention and the conventional technology. FIG. 6 is a block diagram of a data access device according to a preferred embodiment of the present invention. FIG. 7A is a block diagram of a data access device using two Lu Jiyi banks according to a preferred embodiment of the present invention. FIG. 7B is a block diagram of a memory controller in FIG. 7A according to a preferred embodiment of the present invention. ° Fig. 7C is a circuit diagram showing a kind of interrupt circuit in Fig. 7B according to a preferred embodiment of the present invention. FIG. 7D is a block diagram of a multiplexed combination circuit in FIG. 7A according to a preferred embodiment of the present invention. Xi [The icon is not explained] 100: Know the structure of the search window data 11 0: 9 * 9 Reference box 1 2 0: Extract the basic box range

1242741 圖式簡單說明 300 310 610 620 711 713 715 4 0 0 :依照本發明較佳實施例之搜尋窗資料結構 欲讀取參考方塊之所在範圍 7 1 0 :記憶體控制器 720 :多工組合電路 712 、 714a 、 714b 、 721 、 722 :多工器 判斷電路 延遲電路 S2 10〜S244 :依照本發明較佳實施例所述的一種圖框 資料存取方法之各步驟1242741 Schematic illustration 300 310 610 620 711 713 715 4 0 0: Search window data structure according to the preferred embodiment of the present invention. The range of the reference block to be read. 7 1 0: Memory controller 720: Multiplex combination circuit 712, 714a, 714b, 721, 722: Multiplexer judgment circuit delay circuits S2 10 to S244: Steps of a frame data access method according to a preferred embodiment of the present invention

13203twf.ptd 第19頁13203twf.ptd Page 19

Claims (1)

1242741 六、申請專利範圍 1 · 一種圖框資料存取方法,以獲取具有X位元之一欲 讀取圖框資料,其中X為正整數,該圖框資料存取方法包 括下列步驟: a·提供Y個記憶庫BANKi,其中BANKi表示第i個記憶 庫’Y為大於1並且小於等於X之整數,i為大於等於〇並且 小於Y之整數; b·將圖框資料WL A存放於BANK】中,其中WL,A表示具有 X/Y位元之第L行第A個部分圖框資料,L與A皆為大於等於〇 之整數,卜(L + A) mod Y,m〇d為模數運算; c ·接收並依據γ個字元位址ψ Ak判斷所欲讀取之部分圖 框資料分別位於該些記憶庫中之何者,其中表示第k個 欲讀取部分圖框資料之位址,k為大於等於〇並且小於γ之 整數;以及 d·依據步驟c之判斷結果自記憶庫BANKi獲得具有X/Y位 元之該些部分圖框資料,該些記憶庫BANKi所輸出之該些部 分圖框資料之組合即為該欲讀取圖框資料。 2 ·如申請專利範圍第1項所述之圖框資料存取方法, 其中步驟d包括: 依據步驟c之判斷結果產生γ個記憶庫位址,其中 BAi表示第i個記憶庫之存取位址; 依據記憶庫位址BAi存取記憶庫BANKi ; 自§己憶庫BANKi獲得對應之該部分圖框資料;以及 依據字元位址WAk判斷各記憶庫BANK所輸出具有χ/γ位 元之該些部分圖框資料之排列順序並依此組合輸出為具有1242741 VI. Scope of patent application 1 · A frame data access method to obtain frame data with one of X bits to read, where X is a positive integer, the frame data access method includes the following steps: a · Provide Y memory banks BANKi, where BANKi represents the i-th memory bank 'Y is an integer greater than 1 and less than or equal to X, i is an integer greater than or equal to 0 and less than Y; b. Store the frame information WL A in BANK] Among them, WL, A represents the frame data of the Ath part of the Lth row with X / Y bits, L and A are integers greater than or equal to 0, and (L + A) mod Y, and m〇d is the modulus Number operation; c · Receive and judge which part of the frame data to be read is located in the memory according to γ character addresses ψ Ak, which indicates the position of the kth part of the frame data to be read Address, k is an integer greater than or equal to 0 and less than γ; and d. According to the judgment result of step c, the partial frame data with X / Y bits are obtained from the memory bank BANKi, and the memory bank BANKi outputs the The combination of some frame data is the frame data to be read. 2 · The frame data access method described in item 1 of the scope of patent application, wherein step d includes: generating γ memory bank addresses according to the judgment result of step c, where BAi represents the access bit of the i-th memory bank Access to the memory bank BANKi based on the memory bank address BAi; obtain the corresponding frame data from § self-memory bank BANKi; and determine the output of each memory bank BANK with the χ / γ bit according to the character address WAk The arrangement order of the partial frame data is output as 第20頁 1242741 六、申請專利範圍 X位元之該欲讀取圖框資料。 3 ·如申請專利範圍第彳 其係用於-視訊處理(=〇員所返之圖框資料存取方法, 、Vldeo process)。 4 ·如申請專利範圚笛q 5 JL # π 11 A圍第3項所述之圖框資料存取方法, 其係用於ά亥視成處理φ 一闽 of frame)之取得。 圖框彡考方塊(feference b 1 ock 5· —種資料存取裝番 γ . ^ ^ f k凌置用以依據一位址訊號輸出具有 X位兀之一預儲存資料,盆由 • 、 八中X為正整數,該資料存取裝置 包括· 'S己憶體控制器,用拉Ur 憶庫位址以及-記情庫該位址訊號並輸出γ個記 等於X之整數;隐庫判斷訊被’其中Υ為大於1並且小於 Υ個記憶庫,耦接至該記憶 庫分別接收對應之該此記情座π g # ^涊一圯隐 —口己隱庫位址其中之一並且分別輸出 對應j有X/Y位元之一部分預儲存資料;以及 ,障康,:2二:!路’耦接至該記憶體控制器以及該些記 ^ ^ 康“記憶庫判斷訊號將所接收之具有X/Y位 =存資料多工切換並組合輸出為具有X位 讀取並依據該位址訊號判斷所欲 於该此記/t +、4斤l 3之該些部分預儲存資料分別位 於 °己隐庫中之何者,並將其判斷έ士旲於屮A兮々降法 判斷訊號。 竹丹y斷、^果輸出為該記憶庫 6·如申請專利範圍第5項所述之資料存取裝置,其中Y 1242741Page 20 1242741 Sixth, the scope of patent application X-bit should read the frame information. 3 · If the scope of patent application is the first one, it is used for video processing (= frame data access method returned by members, Vldeo process). 4 · The method for accessing frame data as described in the patent application Fan Qidi J Q 5 JL # π 11 A, which is used to obtain the frame processing. Picture frame test box (feature b 1 ock 5 · — a kind of data access equipment γ. ^ ^ Fk is used to output pre-stored data with one of the X bits according to a single address signal. X is a positive integer. The data access device includes a 'S memory controller, which uses Ur memory to recall the address of the memory and-to remember the address signal of the memory and outputs γ integers equal to X; It is' where Υ is greater than 1 and less than Υ memory banks, which are respectively coupled to the memory bank and receive the corresponding memorizer π g # ^ 涊 一 — 隐 —one of the mouth and cryptic bank addresses and output them separately Corresponding j has a part of the X / Y bits of pre-stored data; and, the barriers are: 22 :! Road is coupled to the memory controller and the records ^ ^ "Kang" judgement signal will be received With X / Y bits = multiplexed switching of stored data and combined output as X-bit read and judged according to the address signal to the desired / t +, 4 kg of these pre-stored data are located in ° Which one is in the hidden library, and judge it by έA 屮 々 々, to determine the signal. 竹 丹 y , Memory 6. The information of the application of paragraph 5 ^ patentable scope of the access device for fruit output, wherein Y 1242741 六、申請專利範圍 等於2。 7·如申請專利範圍第6項所述 該位址訊號包含—筮 ^ ^ 〈貝枓存取裝置,其中 ^ 弟一子兀位址以及一笛-今一 記憶體控制器包括: 弟一子兀位址,該 一判斷電路,用以接收該第— 第一字元位址所對庵夕”部八猫^予凡位址並據以判斷該 作庫盆中t 對應之δ亥部分預儲存資料係存在於該此圮 G犀兵中之一,以依 =γ 4二。己 以及 御出5亥圮憶庫判斷訊號; 刀換電路,用以依照該記憶 電路輸入盥給目 早力斯Α號決定該切換 一 W出間之一耦接狀態,該 之該第一丰分A t Λ祸接狀態係將所接收 ~ , 止一該第二字元位址分別耦接輸出為一第 口匕1S庫位址盥一筮一 q 币 —^ 7Γ +L Λ 第一 5己丨思庫位址,以及將所接收之該第 于疋位址与r贷—A— x t 庫位址盥嗲^二第一子兀位址^刀別耦接輸出為該第二記憶 位址盥;;一 I憶庫位址二者之一,其中該第-記憶庫 一。^ ^乐一汜憶庫位址分別為該些記憶庫位址其中之 該判斷電I =專利範圍第7項所述之資料存取裝置,其中 位址之邻^匕括一互斥或閘,該互斥或閘接收該第一字元 斷訊节分位元資料並進行互斤或運算後輸出該記憶庫判6. The scope of patent application is equal to 2. 7. As described in item 6 of the scope of the patent application, the address signal contains-筮 ^ ^ "Beijing access device, where ^ Diyiwu address and Yidi-jinyi memory controller include: Diyizi This address is a judgment circuit that is used to receive the first-byte character address to the "Xi Xi" part of the eight cats ^ Yufan address and use this to determine the corresponding delta portion of t in the library basin. The stored data exists in one of the 圮 G rhinoceros, so as to be equal to γ 4 2. Ji and Yude 5 Hai 圮 memory library to determine the signal; the knife change circuit, used to input to the eyes early force according to the memory circuit No. A decides to switch one of the coupling states between the W output, the first abundance A t Λ coupling state is to receive the received ~, and only the second character address is coupled to output a The address of the first library 1S is q coin — ^ 7Γ + L Λ The first 5 is the address of the library, and the received address is loaned with r —A — xt library address The second memory address is the second memory address and the output is the second memory address; one of the two memory addresses, where the first memory bank ^ ^ The memory address of Leyi is the judgment address of these memory addresses respectively I = the data access device described in item 7 of the patent scope, in which the neighbors of the addresses are mutually exclusive or Brake, the mutex or brake receives the first character break segment quantile data and performs mutual OR operation to output the memory judgment 該多工^申晴專利範圍第6項所述之資料存取裝置,其中 儲^ ^ ^合電路係依據該記憶庫判斷訊號決定該些部分預 ^貝^料之排列順序並合併輸出為該預儲存資料。 •如申請專利範圍第5項所述之資料存取裝置,其係The data access device described in item 6 of the multiplexing application patent scope, wherein the storage circuit determines the arrangement order of the parts of the pre-prepared materials according to the judgement signal of the memory bank, and combines and outputs the data. Pre-stored data. • The data access device described in item 5 of the scope of patent application, which is 1242741 六、申請專利範圍 用於一視訊處理(video process ) 〇 1 1.如申請專利範圍第1 0項所述之資料存取裝置,其 係用於存取一圖框(frame)資料。 1 2.如申請專利範圍第1 1項所述之資料存取裝置,其 係用於該圖框資料中之一參考方塊(reference block)之 取得。1242741 6. Scope of patent application For a video process 〇 1 1. The data access device as described in item 10 of the scope of patent application, which is used to access a frame of data. 1 2. The data access device as described in item 11 of the scope of patent application, which is used to obtain a reference block in the frame data. 11 13203twf.ptd 第23頁11 13203twf.ptd Page 23
TW93113043A 2004-05-10 2004-05-10 Method for accessing frame data and data accessing device thereof TWI242741B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW93113043A TWI242741B (en) 2004-05-10 2004-05-10 Method for accessing frame data and data accessing device thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW93113043A TWI242741B (en) 2004-05-10 2004-05-10 Method for accessing frame data and data accessing device thereof

Publications (2)

Publication Number Publication Date
TWI242741B true TWI242741B (en) 2005-11-01
TW200537369A TW200537369A (en) 2005-11-16

Family

ID=37022586

Family Applications (1)

Application Number Title Priority Date Filing Date
TW93113043A TWI242741B (en) 2004-05-10 2004-05-10 Method for accessing frame data and data accessing device thereof

Country Status (1)

Country Link
TW (1) TWI242741B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI391912B (en) * 2008-11-14 2013-04-01 Orise Technology Co Ltd Frame memory access method and display driver using the same

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI391912B (en) * 2008-11-14 2013-04-01 Orise Technology Co Ltd Frame memory access method and display driver using the same

Also Published As

Publication number Publication date
TW200537369A (en) 2005-11-16

Similar Documents

Publication Publication Date Title
JP5663009B2 (en) Apparatus for displaying original image by warping or rotating, method thereof, computer program thereof, and recording medium recording the same
US5752260A (en) High-speed, multiple-port, interleaved cache with arbitration of multiple access addresses
US7209405B2 (en) Memory device and method having multiple internal data buses and memory bank interleaving
US9641464B2 (en) FIFO buffer system providing same clock cycle response to pop commands
KR100977339B1 (en) Semiconductor devices
EP1474747A1 (en) Address space, bus system, memory controller and device system
US6922770B2 (en) Memory controller providing dynamic arbitration of memory commands
CN101645305B (en) Static random access memory (SRAM) for automatically tracking data
US6853588B2 (en) First-in first-out memory circuit and method for executing same
US8688891B2 (en) Memory controller, method of controlling unaligned memory access, and computing apparatus incorporating memory controller
US7742063B2 (en) Efficient and high speed 2D data transpose engine for SOC application
TWI242741B (en) Method for accessing frame data and data accessing device thereof
US6545936B1 (en) Pipeline structure of memory for high-fast row-cycle
US20100232238A1 (en) Dual port memory device, memory device and method of operating the dual port memory device
JP2009110600A (en) MEMORY ACCESS METHOD AND MEMORY CONTROL DEVICE
US7672177B2 (en) Memory device and method thereof
JP2000195262A (en) Sdram and method for accessing sdram data
JP3276822B2 (en) Video signal processing circuit
JPH09106669A (en) Synchronous DRAM and semiconductor memory device
JP2003151273A (en) Storage device, storage device internal control method, system, and storage device control method in system
JP2002112163A (en) Image data storage device and image data storage method
US20060007235A1 (en) Method of accessing frame data and data accessing device thereof
JP2000057777A (en) Synchronous burst semiconductor memory device
JP4549001B2 (en) Information processing apparatus and semiconductor integrated circuit
JP2002184179A (en) Random access device for sdram

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees