TWI807668B - Memory device with high content density - Google Patents

Memory device with high content density Download PDF

Info

Publication number
TWI807668B
TWI807668B TW111108079A TW111108079A TWI807668B TW I807668 B TWI807668 B TW I807668B TW 111108079 A TW111108079 A TW 111108079A TW 111108079 A TW111108079 A TW 111108079A TW I807668 B TWI807668 B TW I807668B
Authority
TW
Taiwan
Prior art keywords
voltage level
bit
search
memory
data
Prior art date
Application number
TW111108079A
Other languages
Chinese (zh)
Other versions
TW202336763A (en
Inventor
曾柏皓
Original Assignee
旺宏電子股份有限公司
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 旺宏電子股份有限公司 filed Critical 旺宏電子股份有限公司
Priority to TW111108079A priority Critical patent/TWI807668B/en
Application granted granted Critical
Publication of TWI807668B publication Critical patent/TWI807668B/en
Publication of TW202336763A publication Critical patent/TW202336763A/en

Links

Landscapes

  • Static Random-Access Memory (AREA)
  • Pharmaceuticals Containing Other Organic And Inorganic Compounds (AREA)

Abstract

A memory device is disclosed, which includes a first driving circuit, a second driving circuit, a sensing circuit and an in-memory search (IMS) array. A plurality of memory units of the in-memory search array are arranged as a plurality of horizontal rows and a plurality of vertical columns. A respective control terminal of each the memory units in the same vertical column is coupled to the first driving circuit through a corresponding word line. The memory units of the same vertical column are connected in series and coupled to the second driving circuit through the bit line, and coupled to the sensing circuit through the source line. Every 2N adjacent memory units in the same vertical column are arranged as a memory cell, and this memory cell is used to store an encoded data of 2N bits corresponding to an original data of M bits, where N and M are positive integers, and N is greater than or equal to two.

Description

高內容密度之記憶體裝置 High content density memory device

本揭示係關於一種半導體裝置,特別有關於一種可儲存高內容密度資料的記憶體裝置。 The present disclosure relates to a semiconductor device, and more particularly to a memory device capable of storing high content density data.

在人工智慧演算法中,需要執行大量的資料比較與資料搜尋。為了滿足大量的資料比較/搜尋的運算需求,習知技術常使用三態內容可定址記憶體(ternary content addressable memory,TCAM)以執行高度平行搜尋。 In artificial intelligence algorithms, a large number of data comparisons and data searches need to be performed. In order to meet the computational requirements of a large amount of data comparison/search, conventional technologies often use ternary content addressable memory (TCAM) to perform highly parallel search.

習知的TCAM記憶體採用的記憶體型式例如為:靜態隨機存取記憶體(SRAM)、2T2R架構的電阻式記憶體(RRAM)及鐵電式隨機存取記憶體(FeRAM),等。然而,上述型式的TCAM記憶體應用於平行搜尋的運算時,在耗電量及開啟/關閉比例(on/off ratio)的表現上仍不理想,以至於進行資料比較時不易區分完全匹配(all-match)及一位元不匹配(1-bit-mismatch)的狀況。此外,習知的TCAM記憶體的儲存內容的密度較低,對於大量內容的資料搜尋而言略顯不足。 The types of memory used in the conventional TCAM memory are, for example, Static Random Access Memory (SRAM), 2T2R Resistive Memory (RRAM), Ferroelectric Random Access Memory (FeRAM), etc. However, when the above-mentioned type of TCAM memory is applied to parallel search operations, its performance in power consumption and on/off ratio is still not ideal, so that it is difficult to distinguish between an all-match and a 1-bit-mismatch when comparing data. In addition, the conventional TCAM memory has a low density of storage content, which is not enough for searching data of a large amount of content.

因此,本技術領域之相關產業之技術人員係致力於改良TCAM記憶體的資料配置,對於TCAM記憶體的原始資料進行編碼,期能有效提高TCAM記憶體的儲存內容的密度。 Therefore, technicians in related industries in this technical field are devoting themselves to improving the data configuration of the TCAM memory, encoding the original data of the TCAM memory, and hoping to effectively increase the density of the storage content of the TCAM memory.

根據本揭示之一方面,係提供一種記憶體裝置,包括第一驅動電路、第二驅動電路、感測電路以及記憶體內搜尋(IMS)陣列。記憶體內搜尋陣列的多個記憶單元被配置為多個橫向列及多個縱向行,同一縱向行的記憶單元各自的控制端經由對應的字元線耦接至第一驅動電路,同一縱向行的記憶單元串聯連接並經由位元線耦接至第二驅動電路且經由源極線耦接至感測電路。同一縱向行的記憶單元中每2N個相鄰的記憶單元被配置為一個記憶胞,此記憶胞用以儲存2N個位元的編碼資料而對應至M個位元的原始資料,N、M為正整數且N大於或等於二。 According to one aspect of the present disclosure, a memory device is provided, including a first driving circuit, a second driving circuit, a sensing circuit and an In-Memory Search (IMS) array. A plurality of memory cells of the internal memory search array are configured into a plurality of horizontal columns and a plurality of vertical rows, the respective control terminals of the memory cells in the same vertical row are coupled to the first drive circuit via corresponding word lines, and the memory cells in the same vertical row are connected in series and coupled to the second drive circuit via the bit lines and coupled to the sensing circuit via the source lines. Every 2N adjacent memory cells in the memory cells of the same vertical row are configured as a memory cell, and the memory cell is used to store 2N bits of encoded data corresponding to M bits of original data, where N and M are positive integers and N is greater than or equal to two.

根據本揭示之另一方面,係提供一種記憶體裝置,包括第一驅動電路、感測電路以及記憶體內搜尋(IMS)陣列。記憶體內搜尋陣列的多個記憶單元被配置為多個橫向列及多個縱向行,同一橫向列的記憶單元各自的控制端經由對應的字元線耦接至第一驅動電路,同一橫向列的記憶單元經由匹配訊號線耦接至感測電路。同一橫向列的記憶單元中每2N個相鄰的記憶單元被配置為一記憶胞,此記憶胞用以儲存2N個位元的編碼資料而對應至M個位元的原始資料,N、M為正整數且N大於或等於二。 According to another aspect of the present disclosure, a memory device is provided, including a first driving circuit, a sensing circuit, and an In-Memory Search (IMS) array. A plurality of memory cells of the internal memory search array are arranged in a plurality of horizontal columns and a plurality of vertical rows, the respective control terminals of the memory cells in the same horizontal column are coupled to the first driving circuit through corresponding word lines, and the memory cells in the same horizontal column are coupled to the sensing circuit through matching signal lines. Every 2N adjacent memory cells in the same horizontal column are configured as a memory cell, and the memory cell is used to store 2N bits of encoded data corresponding to M bits of original data, where N and M are positive integers and N is greater than or equal to two.

藉由上述技術方案,對於TCAM記憶體的原始資料進行編碼以提高TCAM記憶體的儲存內容的密度。可應用於NAND型式或NOR型式的記憶體,並配合對應的編碼資料配置及搜尋資料配置以有效的執行平行資料比較及資料搜尋。 By means of the above technical solution, the raw data of the TCAM memory is encoded to increase the density of the storage content of the TCAM memory. It can be applied to NAND type or NOR type memory, and cooperates with the corresponding encoding data configuration and search data configuration to effectively perform parallel data comparison and data search.

透過閱讀以下圖式、詳細說明以及申請專利範圍,可見本揭示之其他方面以及優點。 Other aspects and advantages of this disclosure can be seen by reading the following drawings, detailed description and claims.

1000,1000b,2000,3000,3000b:記憶體裝置 1000, 1000b, 2000, 3000, 3000b: memory device

100:第一驅動電路 100: the first drive circuit

200:第二驅動電路 200: the second drive circuit

300:感測電路 300: sensing circuit

400,400b:記憶體內搜尋(IMS)陣列 400, 400b: In-memory search (IMS) array

WL1~WL12,WL2N,WL(2N+1),WL(2N+2):字元線 WL1~WL12, WL2N, WL(2N+1), WL(2N+2): word line

WL4N,WLm,WL13,WL14,WL47:字元線 WL4N, WLm, WL13, WL14, WL47: word line

WL48,WL95,WL96:字元線 WL48,WL95,WL96: word line

BL1~BL12,BLn:位元線 BL1~BL12, BLn: bit line

SL1~SL12,SLn:源極線 SL1~SL12, SLn: source line

ML1~ML4,MLn:匹配訊號線 ML1~ML4, MLn: matching signal line

MP-1,MP-n:預充電電晶體 MP-1, MP-n: pre-charged transistor

V-ML1~V-ML4,V-MLn:電壓位準 V-ML1~V-ML4, V-MLn: voltage level

VM1:第一匹配電壓位準 VM1: The first matching voltage level

VM2:第二匹配電壓位準 VM2: Second matching voltage level

St:控制訊號 St: control signal

Vref:參考電壓 Vref: reference voltage

VH:第一電壓位準 VH: the first voltage level

VL:第二電壓位準 VL: the second voltage level

VH2:第三電壓位準 VH2: the third voltage level

VH1:第四電壓位準 VH1: the fourth voltage level

VD1:第一汲極電壓位準 VD1: the first drain voltage level

VD2:第二汲極電壓位準 VD2: The second drain voltage level

VS1:第一閘極電壓位準 VS1: The first gate voltage level

VS2:第二閘極電壓位準 VS2: The second gate voltage level

H-Vt:第一臨界電壓 H-Vt: the first critical voltage

L-Vt:第二臨界電壓 L-Vt: the second critical voltage

1/0,1,0:邏輯值 1/0,1,0: logic value

Is:源極電流 Is: source current

Is-H:第一源極電流 Is-H: first source current

Is-L:第二源極電流 Is-L: second source current

C11~C23,C11~C41:記憶胞 C11~C23, C11~C41: memory cells

C11b,C11c,C11d,Cn1,Cn2:記憶胞 C11b, C11c, C11d, Cn1, Cn2: memory cells

M(1,1)a:第一端 M(1,1)a: first end

M(6,1)b:第二端 M(6,1)b: second end

M(1,1)~M(6,1),M(1,1)~M(1,12):記憶單元 M(1,1)~M(6,1), M(1,1)~M(1,12): memory unit

M(1,3)~M(6,3),M(13,1),M(14,1),M(47,1):記憶單元 M(1,3)~M(6,3), M(13,1), M(14,1), M(47,1): memory unit

M(48,1),M(9,2),M(12,2),M(95,1),M(96,1):記憶單元 M(48,1), M(9,2), M(12,2), M(95,1), M(96,1): memory unit

M(2,1)~M(2,12),M(2N,1),M(2N+1,1):記憶單元 M(2,1)~M(2,12), M(2N,1), M(2N+1,1): memory unit

M(2N+2,1),M(n,1)~M(n,8):記憶單元 M(2N+2,1), M(n,1)~M(n,8): memory unit

M(2N+1,2),M(2N+2,2):記憶單元 M(2N+1,2), M(2N+2,2): memory unit

M(2N,2),M(4N,1),M(4N,2):記憶單元 M(2N,2), M(4N,1), M(4N,2): memory unit

第1圖為本揭示一實施例的利用字元線搜尋(WL-wise search)之NAND型的記憶體裝置之電路圖。 FIG. 1 is a circuit diagram of a NAND memory device utilizing word line search (WL-wise search) according to an embodiment of the present disclosure.

第2A~2C圖為第1圖的記憶體裝置的其中一個記憶胞經由字元線進行資料搜尋的示意圖。 FIGS. 2A-2C are schematic diagrams of one of the memory cells of the memory device in FIG. 1 searching for data via word lines.

第3圖為第1圖的記憶體裝置經由字元線輸入的搜尋偏壓與記憶單元的臨界電壓的關係圖。 FIG. 3 is a relationship diagram between the search bias inputted through the word line and the threshold voltage of the memory cell of the memory device in FIG. 1 .

第4圖為第1圖的記憶體裝置經由字元線進行資料搜尋的示意圖。 FIG. 4 is a schematic diagram of the memory device in FIG. 1 performing data search through word lines.

第5圖為第1圖的記憶體裝置之另一實施例中,字元線輸入的搜尋偏壓與記憶單元的臨界電壓的關係圖。 FIG. 5 is a graph showing the relationship between the search bias voltage input by the word line and the threshold voltage of the memory cell in another embodiment of the memory device of FIG. 1 .

第6圖為第1圖的記憶體裝置之另一實施例中,經由字元線輸入第5圖的搜尋偏壓進行資料搜尋的示意圖。 FIG. 6 is a schematic diagram of inputting the search bias voltage in FIG. 5 via word lines in another embodiment of the memory device in FIG. 1 to perform data search.

第7A~7C圖分別為NAND型的記憶體裝置之不同層數的記憶胞的示意圖。 FIGS. 7A-7C are schematic diagrams of memory cells of different layers in a NAND memory device.

第8圖為本揭示一實施例的利用位元線搜尋(BL-wise search)之NAND型的記憶體裝置之電路圖。 FIG. 8 is a circuit diagram of a NAND memory device utilizing bit line search (BL-wise search) according to an embodiment of the present disclosure.

第9圖為本揭示一實施例之NOR型的記憶體裝置之電路圖。 FIG. 9 is a circuit diagram of a NOR memory device according to an embodiment of the present disclosure.

第10圖為第9圖的記憶體裝置之匹配訊號線的電壓位準之時序圖。 FIG. 10 is a timing diagram of voltage levels of matching signal lines of the memory device in FIG. 9 .

第11圖為本揭示另一實施例之NOR型的記憶體裝置之電路圖。 FIG. 11 is a circuit diagram of a NOR memory device according to another embodiment of the present disclosure.

第12圖為第9圖的記憶體裝置之另一實施例中,匹配訊號線的電壓位準之時序圖。 FIG. 12 is a timing diagram of the voltage levels of the matching signal lines in another embodiment of the memory device of FIG. 9 .

本說明書的技術用語係參照本技術領域之習慣用語,如本說明書對部分用語有加以說明或定義,該部分用語之解釋係以本說明書之說明或定義為準。本揭露之各個實施例分別具有一或多個技術特徵。在可能實施的前提下,本技術領域具有通常知識者可選擇性地實施任一實施例中部分或全部的技術特徵,或者選擇性地將這些實施例中部分或全部的技術特徵加以組合。 The technical terms in this specification refer to the customary terms in this technical field. If some terms are explained or defined in this specification, the explanations or definitions of these terms shall prevail. Each embodiment of the disclosure has one or more technical features. On the premise of possible implementation, those skilled in the art may selectively implement some or all of the technical features in any embodiment, or selectively combine some or all of the technical features in these embodiments.

第1圖為本揭示一實施例的利用字元線搜尋(WL-wise search)之NAND型的記憶體裝置1000之電路圖。。請參見第1圖,記憶體裝置1000為NAND型記憶體,記憶體裝置1000包括第一驅動電路100、第二驅動電路200、感測電路300、記憶體內搜尋(in-memory searching,IMS)陣列400、複數條字元線(word line)WL1~WLm、複數條位元線(bit line)BL1~BLn及複 數條源極線(source line)SL1~SLn。 FIG. 1 is a circuit diagram of a NAND type memory device 1000 utilizing word line search (WL-wise search) according to an embodiment of the present disclosure. . Please refer to FIG. 1, the memory device 1000 is a NAND memory, and the memory device 1000 includes a first driving circuit 100, a second driving circuit 200, a sensing circuit 300, an in-memory searching (IMS) array 400, a plurality of word lines (word lines) WL1-WLm, a plurality of bit lines (bit lines) BL1-BLn and complex Several source lines (source lines) SL1˜SLn.

IMS陣列400的記憶體類型可例如為浮動閘極(floating gate)記憶體、浮點(floating dot)記憶體、相變記憶體(phase-change memory,PCM)、鐵電式隨機存取記憶體(FeRAM)或可變電阻式隨機存取記憶體(ReRAM),等。IMS陣列400可包括配置為NAND型式的複數個記憶單元M(i,j),此些記憶單元M(i,j)可被配置為m個橫向列(row)及n個縱向行(column)。第i列的記憶單元M(i,j)的控制端(例如為閘極端)可經由字元線WLi耦接至第一驅動電路100。第j行的記憶單元M(i,j)串聯連接,並經由位元線BLj耦接至第二驅動電路200,且經由源極線SLj耦接至感測電路300。 The memory type of the IMS array 400 may be, for example, a floating gate memory, a floating dot memory, a phase-change memory (PCM), a ferroelectric random access memory (FeRAM), or a variable resistance random access memory (ReRAM), and the like. The IMS array 400 may include a plurality of memory cells M(i,j) configured in NAND type, and these memory cells M(i,j) may be configured into m horizontal columns (rows) and n vertical rows (columns). The control terminals (eg, gate terminals) of the memory cells M(i,j) in the i-th column can be coupled to the first driving circuit 100 via the word line WLi. The memory cells M(i,j) in the jth row are connected in series, and are coupled to the second driving circuit 200 via the bit line BLj, and coupled to the sensing circuit 300 via the source line SLj.

IMS陣列400具有參數N,N為正整數且N大於或等於二。IMS陣列400中同一行的2N個相鄰的記憶單元可配置為一個記憶胞(memory cell)。例如,第1行中的第1列至第2N列的記憶單元M(1,1)~M(2N,1)配置為記憶胞C11,第1行中的第(2N+1)列至第(4N)列的記憶單元M((2N+1),1)~M(4N,1)配置為記憶胞C21,第2行的第1列至第2N列的記憶單元M(1,2)~M(2N,2)配置為記憶胞C12,依此類推。每個記憶胞Cij的2N個記憶單元可儲存2N個位元的編碼資料。其中,第1~N位元的編碼資料可組成第一編碼區域,第(N+1)~2N位元的編碼資料可組成第二編碼區域,第一編碼區域與第二編碼區域的編碼組合數量為

Figure 111108079-A0305-02-0008-1
。對應的,可經由字元線WL1~WL2N輸入2N個搜尋偏壓以表示搜尋資料,此 些搜尋資料的組合數量亦為
Figure 111108079-A0305-02-0009-2
。可對於M個位元的原始資料執行編碼機制而得到上述2N個位元的編碼資料,M為正整數。2N個位元的編碼資料的編碼組合數量
Figure 111108079-A0305-02-0009-3
大於或等於「二的M次方」, 例如,N=3、M=4時,編碼組合數量
Figure 111108079-A0305-02-0009-4
大於或等於「二的4次方」。 The IMS array 400 has a parameter N, which is a positive integer and N is greater than or equal to two. 2N adjacent memory cells in the same row in the IMS array 400 can be configured as a memory cell. For example, memory cells M(1,1)~M(2N,1) from column 1 to column 2N in row 1 are configured as memory cell C11, memory cells M((2N+1),1)~M(4N,1) in row 1 from column (2N+1) to column (4N) are configured as memory cell C21, and memory cells M(1,2)~M(2N,2) in column 1 to column 2N in row 2 are configured as memory cell C11. cell C12, and so on. The 2N memory units of each memory cell Cij can store 2N bits of encoded data. Among them, the coded data of the 1st~N bits can form the first coding area, and the coded data of the (N+1)~2N bits can form the second coded area, and the number of coding combinations between the first coded area and the second coded area is
Figure 111108079-A0305-02-0008-1
. Correspondingly, 2N search bias voltages can be input through the word lines WL1~WL2N to represent the search data, and the number of combinations of these search data is also
Figure 111108079-A0305-02-0009-2
. The above-mentioned 2N-bit coded data can be obtained by performing a coding mechanism on the M-bit raw data, where M is a positive integer. Number of encoding combinations for 2N bits of encoded data
Figure 111108079-A0305-02-0009-3
Greater than or equal to "two to the M power", for example, when N=3, M=4, the number of coding combinations
Figure 111108079-A0305-02-0009-4
Greater than or equal to "two to the 4th power".

第一驅動電路100可經由字元線WL1~WLm對於記憶單元M(i,j)施加編程電壓以改變記憶單元M(i,j)的臨界電壓(threshold volage,Vt),使記憶單元M(i,j)儲存邏輯值「1」或邏輯值「0」。可經由字元線WL1~WLm施加搜尋偏壓(search bias)以搜尋記憶胞Cij儲存的編碼資料。當搜尋偏壓匹配(match)於記憶胞Cij儲存的編碼資料時,對應的源極線SLj可輸出源極電流Is至感測電路300,表示搜尋成功。 The first driving circuit 100 can apply a programming voltage to the memory cell M(i,j) through the word lines WL1˜WLm to change the threshold voltage (threshold voltage, Vt) of the memory cell M(i,j), so that the memory cell M(i,j) stores a logic value “1” or a logic value “0”. A search bias can be applied through the word lines WL1 ˜WLm to search the encoded data stored in the memory cells Cij. When the search bias voltage matches the encoded data stored in the memory cell Cij, the corresponding source line SLj can output a source current Is to the sensing circuit 300, indicating that the search is successful.

第2A~2C圖為第1圖的記憶體裝置1000的其中一個記憶胞C11經由字元線WL1~WL6進行資料搜尋的示意圖。請先參見第2A圖,以參數N=3為例,記憶胞C11為六層(6-layer)記憶胞,其包括六個記憶單元M(1,1)~M(6,1)。記憶單元M(1,1)~M(6,1)的控制端(例如為閘極端)經由字元線WL1~WL6直接或間接耦接至第一驅動電路100。記憶單元M(1,1)的第一端(例如為汲極端)M(1,1)a經由位元線BL1直接或間接耦接至第二驅動電路200。記憶單元M(6,1)的第二端(例如為源極端)M(6,1)b經由源極線SL1直接或間接耦接至感測電路300。 FIGS. 2A-2C are schematic diagrams of one memory cell C11 of the memory device 1000 in FIG. 1 performing data search through word lines WL1-WL6. Please refer to FIG. 2A first. Taking the parameter N=3 as an example, the memory cell C11 is a 6-layer memory cell, which includes six memory units M(1,1)˜M(6,1). The control terminals (eg, gate terminals) of the memory cells M(1,1)-M(6,1) are directly or indirectly coupled to the first driving circuit 100 through the word lines WL1-WL6. The first end (for example, the drain end) M(1,1)a of the memory cell M(1,1) is directly or indirectly coupled to the second driving circuit 200 via the bit line BL1. The second terminal (for example, the source terminal) M( 6 , 1 )b of the memory unit M( 6 , 1 ) is directly or indirectly coupled to the sensing circuit 300 via the source line SL1 .

記憶胞C11可儲存四個位元的原始資料,並經由編碼機制將原始資料編碼成為六個位元的編碼資料,並且編碼資料儲存於記憶單元M(1,1)~M(6,1)。記憶胞C11儲存的編碼資料的編碼組合數量及字元線WL1~WL6輸入的搜尋資料的組合數量皆為

Figure 111108079-A0305-02-0010-5
。表一所示為記憶胞C11儲存的原始資料、編碼資料及字元線WL1~WL6輸入的搜尋資料的對應關係:
Figure 111108079-A0305-02-0010-16
The memory cell C11 can store 4-bit original data, and encode the original data into 6-bit coded data through the coding mechanism, and the coded data is stored in the memory cells M(1,1)~M(6,1). The code combination quantity of the code data stored in the memory cell C11 and the combination quantity of the search data input by the word lines WL1~WL6 are both
Figure 111108079-A0305-02-0010-5
. Table 1 shows the correspondence between the original data stored in the memory cell C11, the coded data, and the search data input by word lines WL1~WL6:
Figure 111108079-A0305-02-0010-16

請參見表一,本實施例的編碼機制中,可將二進位數值「0000」至「1111」之四個位元的原始資料(對應至十進位數值「0」至「15」)編碼成為六個位元的編碼資料「000111」至「101100」。並且,編碼資料「111111」表示「隨意(don’t care)」以對應至原始資料的二進位數值「0」至「19」的任一者。再者,編碼資料「000000」表示「無效資料(invalid data)」。 Please refer to Table 1. In the encoding mechanism of this embodiment, the four-bit raw data of the binary value "0000" to "1111" (corresponding to the decimal value "0" to "15") can be encoded into six-bit encoded data "000111" to "101100". Furthermore, the encoded data "111111" represents "don't care" and corresponds to any one of the binary values "0" to "19" in the original data. Furthermore, the encoded data "000000" represents "invalid data".

請參見第2A圖,第一驅動電路100可經由字元線WL1~WL6將記憶胞C11的記憶單元M(1,1)~M(6,1)編程為具有「第一臨界電壓H-Vt」或「第二臨界電壓L-Vt」。第一臨界電壓H-Vt具有較高的電壓位準而對應至邏輯值「0」的儲存資料,第二臨界電壓L-Vt具有較低的電壓位準而對應至邏輯值「1」的儲存資料,並且,第一臨界電壓H-Vt大於第二臨界電壓L-Vt。可將記憶單元M(1,1)~M(6,1)的臨界電壓編程為「L-Vt、H-Vt、L-Vt、L-Vt、H-Vt、H-Vt」,以使記憶單元M(1,1)~M(6,1)儲存六個位元的編碼資料「101100」,其對應至四個位元的原始資料「1111」(即,十進位數值「15」)。 Please refer to FIG. 2A , the first driving circuit 100 can program the memory cells M(1,1)-M(6,1) of the memory cell C11 to have "first critical voltage H-Vt" or "second critical voltage L-Vt" via word lines WL1-WL6. The first threshold voltage H-Vt has a higher voltage level and corresponds to stored data of a logic value “0”, the second threshold voltage L-Vt has a lower voltage level and corresponds to stored data of a logic value “1”, and the first threshold voltage H-Vt is greater than the second threshold voltage L-Vt. The threshold voltages of the memory cells M(1,1)~M(6,1) can be programmed as "L-Vt, H-Vt, L-Vt, L-Vt, H-Vt, H-Vt", so that the memory cells M(1,1)~M(6,1) store the six-bit encoded data "101100", which corresponds to the four-bit original data "1111" (ie, the decimal value "15").

並且,第一驅動電路100可經由字元線WL1~WL6輸入搜尋偏壓以搜尋記憶單元M(1,1)~M(6,1)儲存的編碼資料。請同時參見第3圖,其繪示為第1圖的記憶體裝置1000經由字元線WLi輸入的搜尋偏壓與記憶單元M(i,j)的臨界電壓的關係圖。在本實施例中,輸入的搜尋偏壓可具有「第一電壓位準VH」或「第二電壓位準VL」。搜尋偏壓的第一電壓位準VH具有較高的電壓 位準,其表示搜尋資料「H」而對應至記憶單元M(i,j)儲存的邏輯值「0」的位元。搜尋偏壓的第二電壓位準VL具有較低的電壓位準,其表示搜尋資料「L」而對應至記憶單元M(i,j)儲存的邏輯值「1」的位元。搜尋偏壓的第一電壓位準VH大於第二電壓位準VL,搜尋偏壓的第一電壓位準VH大於記憶單元M(i,j)的第一臨界電壓H-Vt,且搜尋偏壓的第二電壓位準VL小於記憶單元M(i,j)的第一臨界電壓H-Vt且大於第二臨界電壓L-Vt。第一電壓位準VH或第二電壓位準VL的搜尋偏壓經由字元線WLi輸入記憶單元M(i,j)的閘極。若輸入第一電壓位準VH的搜尋偏壓,不論記憶單元M(i,j)儲存邏輯值「1」或邏輯值「0」,第一電壓位準VH的搜尋偏壓皆可導通記憶單元M(i,j)而產生源極電流Is。若輸入第二電壓位準VL的搜尋偏壓,僅當記憶單元M(i,j)儲存邏輯值「1」時,第二電壓位準VL的搜尋偏壓可導通記憶單元M(i,j)而產生源極電流Is。 Moreover, the first driving circuit 100 can input a search bias voltage through the word lines WL1 ˜ WL6 to search the encoded data stored in the memory cells M( 1 , 1 ) ˜ M( 6 , 1 ). Please refer to FIG. 3 at the same time, which shows the relationship between the search bias inputted through the word line WLi and the threshold voltage of the memory cell M(i,j) of the memory device 1000 in FIG. 1 . In this embodiment, the input search bias may have a "first voltage level VH" or a "second voltage level VL". The first voltage level VH of the search bias has a higher voltage A level, which represents the bit corresponding to the logical value "0" stored in the memory unit M(i,j) for the search data "H". The second voltage level VL of the search bias voltage has a lower voltage level, which represents the bit of the search data "L" corresponding to the logical value "1" stored in the memory unit M(i,j). The first voltage level VH of the search bias is greater than the second voltage level VL, the first voltage level VH of the search bias is greater than the first threshold voltage H-Vt of the memory cell M(i,j), and the second voltage level VL of the search bias is smaller than the first threshold voltage H-Vt of the memory cell M(i,j) and greater than the second threshold voltage L-Vt. The search bias of the first voltage level VH or the second voltage level VL is input to the gate of the memory cell M(i,j) through the word line WLi. If the search bias of the first voltage level VH is input, no matter the memory cell M(i,j) stores logic value “1” or logic value “0”, the search bias of the first voltage level VH can turn on the memory cell M(i,j) to generate the source current Is. If the search bias of the second voltage level VL is input, the search bias of the second voltage level VL can turn on the memory cell M(i,j) to generate the source current Is only when the memory cell M(i,j) stores logic value “1”.

請再參見第2A圖,經由字元線WL1~WL6輸入電壓位準為「VL、VH、VL、VL、VH、VH」的搜尋偏壓以表示搜尋資料「LHLLHH」。記憶胞C11的記憶單元M(1,1)~M(6,1)儲存編碼資料「101100」,記憶單元M(1,1)~M(6,1)的臨界電壓為「L-Vt、H-Vt、L-Vt、L-Vt、H-Vt、H-Vt」。因此,電壓位準為「VL、VH、VL、VL、VH、VH」的搜尋偏壓可導通全部的記憶單元M(1,1)~M(6,1),使得記憶單元M(1,1)~M(6,1)的每一者皆產生源極電流Is。據此,感測電路300可感測到源極線SL1 輸出的源極電流Is,此源極電流Is可作為「匹配電流(match current)」。當感測電路300感測到匹配電流時,可判斷字元線WL1~WL6輸入的搜尋資料「LHLLHH」匹配於記憶胞C11儲存的編碼資料「101100」,表示搜尋成功。 Please refer to FIG. 2A again, the search bias voltage of “VL, VH, VL, VL, VH, VH” is input through the word lines WL1˜WL6 to represent the search data “LHLLHH”. The memory cells M(1,1)~M(6,1) of the memory cell C11 store the encoded data "101100", and the threshold voltages of the memory cells M(1,1)~M(6,1) are "L-Vt, H-Vt, L-Vt, L-Vt, H-Vt, H-Vt". Therefore, the search bias whose voltage level is “VL, VH, VL, VL, VH, VH” can turn on all the memory cells M(1,1)˜M(6,1), so that each of the memory cells M(1,1)˜M(6,1) generates a source current Is. Accordingly, the sensing circuit 300 can sense the source line SL1 The output source current Is can be used as a "match current". When the sensing circuit 300 senses the matching current, it can be judged that the search data “LHLLHH” input by the word lines WL1 ˜ WL6 matches the code data “101100” stored in the memory cell C11 , indicating that the search is successful.

接著參見第2B圖及表一,當字元線WL1~WL6的每一者皆輸入第一電壓位準VH的搜尋偏壓時,字元線WL1~WL6輸入的搜尋資料「HHHHHH」可作為「通配符(wildcard)」。此時,不論記憶單元M(1,1)~M(6,1)儲存邏輯值「0」而具有第一臨界電壓H-Vt或儲存邏輯值「1」而具有第二臨界電壓L-Vt,第一電壓位準VH的搜尋偏壓皆可導通記憶單元M(1,1)~M(6,1)而產生源極電流Is。感測電路300皆可感測到匹配電流而判斷搜尋成功。換言之,作為通配符的搜尋資料「HHHHHH」可無條件匹配於邏輯值「0」或「1」的各種組合的編碼資料。 Referring to FIG. 2B and Table 1, when each of the word lines WL1˜WL6 is input with a search bias of the first voltage level VH, the search data “HHHHHH” input by the word lines WL1˜WL6 can be used as a “wildcard”. At this time, regardless of whether the memory cells M(1,1)~M(6,1) store a logic value "0" and have the first threshold voltage H-Vt or store a logic value "1" and have the second threshold voltage L-Vt, the search bias of the first voltage level VH can turn on the memory cells M(1,1)~M(6,1) to generate the source current Is. The sensing circuit 300 can sense the matching current to determine that the search is successful. In other words, the search data "HHHHHH" as a wildcard can unconditionally match coded data of various combinations of logical values "0" or "1".

相對地,參見第2C圖及表一,當字元線WL1~WL6的每一者皆輸入第二電壓位準VL的搜尋偏壓時,字元線WL1~WL6輸入的搜尋資料「LLLLLL」可作為「無效搜尋(invalid-search)」。此時,若記憶單元M(1,1)~M(6,1)儲存邏輯值「0」而具有第一臨界電壓H-Vt,第二電壓位準VL的搜尋偏壓無法導通記憶單元M(1,1)~M(6,1),因而不產生源極電流Is。感測電路300無法感測到匹配電流而判斷搜尋失敗。 In contrast, referring to FIG. 2C and Table 1, when each of the word lines WL1˜WL6 is input with the search bias of the second voltage level VL, the search data “LLLLLL” input by the word lines WL1˜WL6 can be regarded as “invalid-search”. At this time, if the memory cells M(1,1)~M(6,1) store the logic value "0" and have the first critical voltage H-Vt, the search bias of the second voltage level VL cannot turn on the memory cells M(1,1)~M(6,1), thus no source current Is is generated. The sensing circuit 300 cannot sense the matching current and judges that the search fails.

第4圖為第1圖的記憶體裝置1000經由字元線 WL1~WL12進行資料搜尋的示意圖。記憶體裝置1000的IMS陣列400例如包括記憶胞C11、C12、C13、C21、C22、C23。記憶胞C11、C21耦接至位元線BL1,記憶胞C11儲存編碼資料「101100」,記憶胞C21儲存編碼資料「101001」。當經由字元線WL1~WL6輸入電壓位準為「VL、VH、VL、VL、VH、VH」的搜尋偏壓以表示搜尋資料「LHLLHH」時,儲存編碼資料「101100」的記憶胞C11可被導通而產生源極電流Is。並且,當經由字元線WL7~WL12輸入電壓位準為「VL、VH、VL、VH、VH、VL」的搜尋偏壓以表示搜尋資料「LHLHHL」時,儲存編碼資料「101001」的記憶胞C21可被導通而產生源極電流Is。據此,感測電路300可由源極線SL接收到源極電流Is(即,匹配電流),可判斷記憶胞C11儲存的編碼資料「101100」匹配於搜尋資料「LHLLHH」並且記憶胞C21儲存的編碼資料「101001」匹配於搜尋資料「LHLHHL」,表示搜尋成功。 Figure 4 shows the memory device 1000 in Figure 1 via word lines Schematic diagram of WL1~WL12 performing data search. The IMS array 400 of the memory device 1000 includes, for example, memory cells C11, C12, C13, C21, C22, and C23. The memory cells C11 and C21 are coupled to the bit line BL1, the memory cell C11 stores the encoded data “101100”, and the memory cell C21 stores the encoded data “101001”. When a search bias with a voltage level of “VL, VH, VL, VL, VH, VH” is input through the word lines WL1˜WL6 to represent the search data “LHLLHH”, the memory cell C11 storing the encoded data “101100” can be turned on to generate a source current Is. Moreover, when a search bias voltage with a voltage level of “VL, VH, VL, VH, VH, VL” is input through the word lines WL7˜WL12 to represent the search data “LHLHHL”, the memory cell C21 storing the encoded data “101001” can be turned on to generate a source current Is. Accordingly, the sensing circuit 300 can receive the source current Is (that is, the matching current) through the source line SL, and can judge that the code data “101100” stored in the memory cell C11 matches the search data “LHLLHH” and the code data “101001” stored in the memory cell C21 matches the search data “LHLHHL”, indicating that the search is successful.

另一方面,記憶胞C12與記憶胞C22耦接至位元線BL2,記憶胞C12儲存編碼資料「011010」,記憶胞C22儲存編碼資料「100110」。當字元線WL1~WL6輸入電壓位準為「VL、VH、VL、VL、VH、VH」的搜尋偏壓時,記憶胞C12的記憶單元M(1,2)、M(4,2)無法被導通,因此記憶胞C12無法提供源極電流Is。當字元線WL7~WL12輸入電壓位準為「VL、VH、VL、VH、VH、VL」的搜尋偏壓時,記憶胞C22的記憶單元M(9,2)、M(12,2)無法被導通,因此記憶胞C22亦無法提供源極電流Is。 感測電路300無法感測到源極線SL2的源極電流Is,可判斷記憶胞C12、C22儲存的編碼資料不匹配(mismatch)於搜尋資料,表示搜尋失敗。 On the other hand, the memory cell C12 and the memory cell C22 are coupled to the bit line BL2, the memory cell C12 stores the encoded data “011010”, and the memory cell C22 stores the encoded data “100110”. When the input voltage levels of the word lines WL1-WL6 are the search bias voltages of “VL, VH, VL, VL, VH, VH”, the memory cells M(1,2) and M(4,2) of the memory cell C12 cannot be turned on, so the memory cell C12 cannot provide the source current Is. When the input voltage levels of the word lines WL7-WL12 are the search bias voltages of “VL, VH, VL, VH, VH, VL”, the memory cells M(9,2) and M(12,2) of the memory cell C22 cannot be turned on, so the memory cell C22 cannot provide the source current Is. The sensing circuit 300 cannot sense the source current Is of the source line SL2, and it can be judged that the encoded data stored in the memory cells C12 and C22 do not match the search data, indicating that the search fails.

再者,記憶胞C13與記憶胞C23耦接至位元線BL3。記憶胞C13儲存的編碼資料「111111」表示「隨意」而對應至原始資料的二進位數值「0」至「19」的任一者。不論字元線WL1~WL6輸入第一電壓位準VH或第二電壓位準VL的搜尋偏壓,記憶單元M(1,3)~M(6,3)的每一者都可被導通而產生源極電流Is,因此記憶胞C13儲存的編碼資料「111111」皆可匹配於第一電壓位準VH或第二電壓位準VL的搜尋偏壓。並且,記憶胞C23儲存的編碼資料「101001」相同於記憶胞C21而匹配於字元線WL7~WL12輸入的搜尋資料「LHLHHL」。 Furthermore, the memory cell C13 and the memory cell C23 are coupled to the bit line BL3. The encoded data "111111" stored in the memory cell C13 represents "random" and corresponds to any one of the binary values "0" to "19" of the original data. No matter the word lines WL1-WL6 input the search bias of the first voltage level VH or the second voltage level VL, each of the memory cells M(1,3)-M(6,3) can be turned on to generate the source current Is, so the encoded data “111111” stored in the memory cell C13 can match the search bias of the first voltage level VH or the second voltage level VL. Moreover, the encoded data "101001" stored in the memory cell C23 is the same as the memory cell C21 and matches the search data "LHLHHL" input by the word lines WL7-WL12.

第5圖為第1圖的記憶體裝置1000之另一實施例中,字元線WLi輸入的搜尋偏壓與記憶單元M(i,j)的臨界電壓的關係圖。在第5圖的實施例中,搜尋偏壓可具有「第三電壓位準VH2」或「第四電壓位準VH1」。搜尋偏壓的第三電壓位準VH2表示搜尋資料「H」而對應至記憶單元M(i,j)儲存的邏輯值「0」的位元。搜尋偏壓的第四電壓位準VH1表示搜尋資料「L」而對應至記憶單元M(i,j)儲存的邏輯值「1」的位元。搜尋偏壓的第三電壓位準VH2大於第四電壓位準VH1,且第四電壓位準VH1大於記憶單元M(i,j)的第一臨界電壓H-Vt及第二臨界電壓L-Vt。 FIG. 5 is a graph showing the relationship between the search bias input from the word line WLi and the threshold voltage of the memory cell M(i,j) in another embodiment of the memory device 1000 in FIG. 1 . In the embodiment of FIG. 5 , the search bias may have a "third voltage level VH2" or a "fourth voltage level VH1". The third voltage level VH2 of the search bias represents the bit of the search data “H” corresponding to the logical value “0” stored in the memory unit M(i,j). The fourth voltage level VH1 of the search bias represents the bit of the search data “L” corresponding to the logical value “1” stored in the memory unit M(i,j). The third voltage level VH2 of the search bias is greater than the fourth voltage level VH1, and the fourth voltage level VH1 is greater than the first threshold voltage H-Vt and the second threshold voltage L-Vt of the memory cell M(i,j).

當第三電壓位準VH2的搜尋偏壓匹配於記憶單元M(i,j)儲存的邏輯值「0」的位元時,且邏輯值「0」對應於第一臨界電壓H-Vt,則第三電壓位準VH2的搜尋偏壓與第一臨界電壓H-Vt之間的電壓差較大;因此記憶單元M(i,j)可產生較大的源極電流Is(即,匹配電流)。當第四電壓位準VH1的搜尋偏壓匹配於記憶單元M(i,j)儲存的邏輯值「1」的位元時,且邏輯值「1」對應於第二臨界電壓L-Vt,則第四電壓位準VH1的搜尋偏壓與第二臨界電壓L-Vt之間的電壓差較大;因此記憶單元M(i,j)亦可產生較大的源極電流Is。另一方面,當第四電壓位準VH1的搜尋偏壓不匹配於記憶單元M(i,j)儲存的邏輯值「0」的位元時,此時第四電壓位準VH1的搜尋偏壓與邏輯值「0」對應的第一臨界電壓H-Vt之間的電壓差較小,因此記憶單元M(i,j)的源極電流Is較小。由上,當感測電路300感測到較大的匹配電流時,可判斷搜尋資料匹配於儲存資料的該位元。相反的,若匹配電流較小,則判斷搜尋資料不匹配於儲存資料的該位元。 When the search bias of the third voltage level VH2 matches the bit of logic value “0” stored in the memory cell M(i,j), and the logic value “0” corresponds to the first threshold voltage H-Vt, the voltage difference between the search bias of the third voltage level VH2 and the first threshold voltage H-Vt is relatively large; therefore, the memory cell M(i,j) can generate a relatively large source current Is (ie, matching current). When the search bias of the fourth voltage level VH1 matches the bit of the logic value “1” stored in the memory unit M(i,j), and the logic value “1” corresponds to the second threshold voltage L-Vt, the voltage difference between the search bias of the fourth voltage level VH1 and the second threshold voltage L-Vt is relatively large; therefore, the memory unit M(i,j) can also generate a relatively large source current Is. On the other hand, when the search bias of the fourth voltage level VH1 does not match the bit of the logic value “0” stored in the memory cell M(i,j), the voltage difference between the search bias of the fourth voltage level VH1 and the first threshold voltage H-Vt corresponding to the logic value “0” is relatively small, so the source current Is of the memory cell M(i,j) is relatively small. From the above, when the sensing circuit 300 senses a large matching current, it can be determined that the search data matches the bit of the stored data. On the contrary, if the matching current is small, it is determined that the search data does not match the bit of the storage data.

第6圖為第1圖的記憶體裝置1000之另一實施例中,經由字元線WL1~WL12輸入第5圖的搜尋偏壓(具有第三電壓位準VH1或第四電壓位準VH2)進行資料搜尋的示意圖。第6圖的記憶胞C11~C23儲存的編碼資料相同於第4圖的實施例。字元線WL1~WL6輸入電壓位準為「VH1、VH2、VH1、VH1、VH2、VH2」的搜尋偏壓以表示搜尋資料「LHLLHH」,字元線WL7~WL12輸入電壓位準為「VH1、VH2、VH1、VH2、VH2、 VH1」的搜尋偏壓以表示搜尋資料「LHLHHL」,此時,記憶胞C11、C21匹配於搜尋資料而產生電流值較大的第一源極電流Is-H。 FIG. 6 is a schematic diagram of inputting the search bias (with the third voltage level VH1 or the fourth voltage level VH2 ) in FIG. 5 through the word lines WL1-WL12 in another embodiment of the memory device 1000 in FIG. 1 for data search. The encoded data stored in the memory cells C11-C23 in FIG. 6 is the same as the embodiment in FIG. 4. The input voltage levels of word lines WL1~WL6 are “VH1, VH2, VH1, VH1, VH2, VH2” to represent the search data “LHLLHH”, and the input voltage levels of word lines WL7~WL12 are “VH1, VH2, VH1, VH2, VH2, The search bias voltage VH1" represents the search data "LHLHHL". At this time, the memory cells C11 and C21 match the search data to generate the first source current Is-H with a larger current value.

另一方面,記憶胞C12的記憶單元M(1,2)、M(4,2)儲存的邏輯值「0」的位元不匹配於第四電壓位準VH1的搜尋偏壓,記憶胞C22的記憶單元M(9,2)、M(12,2)儲存的邏輯值「0」的位元不匹配於第四電壓位準VH1的搜尋偏壓,因此源極線SL2提供第二源極電流Is-L,第二源極電流Is-L的電流值小於第一源極電流Is-H。 On the other hand, the bits of the logic value “0” stored in the memory cells M(1,2) and M(4,2) of the memory cell C12 do not match the search bias of the fourth voltage level VH1, and the bits of the logic value “0” stored in the memory cells M(9,2) and M(12,2) of the memory cell C22 do not match the search bias of the fourth voltage level VH1, so the source line SL2 provides the second source current Is-L, the current value of the second source current Is-L less than the first source current Is-H.

再者,記憶胞C13的記憶單元M(1,3)~M(6,3)的每一者皆儲存邏輯值「1」而對應至隨意「111111」,記憶胞C13匹配搜尋資料而產生電流值較大的第一源極電流Is-H。 Furthermore, each of the memory cells M(1,3)˜M(6,3) of the memory cell C13 stores a logic value “1” and corresponds to a random “111111”, and the memory cell C13 matches the search data to generate the first source current Is-H with a relatively large current value.

第7A~7C圖分別為NAND型的記憶體裝置之不同層數的記憶胞C11b、C11c及C11d的示意圖。第7A圖的記憶胞C11b為十四層(14-layer)記憶胞,由配置於同一行的14個相鄰的記憶單元M(1,1)~M(14,1)組成。可經由14條字元線WL1~WL14進行資料搜尋。十四層記憶胞C11b的參數N=7,其編碼資料及搜尋資料的組合數量皆為

Figure 111108079-A0305-02-0017-7
。由於9個位元的二進位數值的組合數量為512,10個位元的二進位數值的組合數量為1024,因此組合數量為924的十四層記憶胞C11b可儲存約9.5個位元的等效位元數。 FIGS. 7A-7C are schematic diagrams of memory cells C11b, C11c, and C11d in different layers of a NAND memory device, respectively. The memory cell C11b in FIG. 7A is a 14-layer memory cell, which is composed of 14 adjacent memory cells M(1,1)˜M(14,1) arranged in the same row. The data can be searched through 14 word lines WL1~WL14. The parameter N=7 of the 14-layer memory cell C11b, the number of combinations of its encoding data and search data is
Figure 111108079-A0305-02-0017-7
. Since the combination number of 9-bit binary values is 512 and the combination number of 10-bit binary values is 1024, the fourteen-layer memory cell C11b with a combination number of 924 can store an equivalent bit number of about 9.5 bits.

第7B圖的記憶胞C11c為四十八層(48-layer)記憶胞,由48個記憶單元M(1,1)~M(48,1)組成。可經由48條字元線WL1~WL48進行資料搜尋。四十八層記憶胞C11c的參數N=24,其編碼資料的組合數量及搜尋資料的組合數量皆為

Figure 111108079-A0305-02-0018-8
而相當於45個位元的等效位元數。 The memory cell C11c in FIG. 7B is a 48-layer memory cell, which is composed of 48 memory units M(1,1)~M(48,1). The data can be searched through 48 word lines WL1~WL48. The parameter N=24 of the forty-eight-layer memory cell C11c, the number of combinations of its coding data and the number of combinations of search data are both
Figure 111108079-A0305-02-0018-8
And the equivalent number of bits is equivalent to 45 bits.

第7C圖的記憶胞C11d為九十六層(96-layer)記憶胞,由96個記憶單元M(1,1)~M(96,1)組成。可經由96條字元線WL1~WL96進行資料搜尋。九十六層記憶胞C11d的參數N=48,其編碼資料的組合數量及搜尋資料的組合數量皆為

Figure 111108079-A0305-02-0018-9
而相當於92個位元的等效位元數。 The memory cell C11d in FIG. 7C is a 96-layer memory cell, which is composed of 96 memory units M(1,1)˜M(96,1). The data can be searched through 96 word lines WL1~WL96. The parameter N=48 of the ninety-six-layer memory cell C11d, the number of combinations of the encoded data and the number of combinations of the search data are both
Figure 111108079-A0305-02-0018-9
And the number of bits equivalent to 92 bits.

第8圖為本揭示一實施例的利用位元線搜尋(BL-wise search)之NAND型的記憶體裝置2000之電路圖。記憶體裝置2000亦為NAND型記憶體,可經由位元線BL1~BL12輸入搜尋資料。記憶體裝置2000的IMS陣列400b中的同一列的2N個相鄰的記憶單元可配置為一個記憶胞。以參數N=3的六層記憶胞為例,第1列第1~6行的記憶單元M(1,1)~M(1,6)配置為記憶胞C11,第1列第7~12行的記憶單元M(1,7)~M(1,12)配置為記憶胞C12,依此類推。 FIG. 8 is a circuit diagram of a NAND type memory device 2000 utilizing bit line search (BL-wise search) according to an embodiment of the present disclosure. The memory device 2000 is also a NAND type memory, and the search data can be input through the bit lines BL1-BL12. 2N adjacent memory cells in the same column in the IMS array 400b of the memory device 2000 can be configured as a memory cell. Taking a six-layer memory cell with parameter N=3 as an example, the memory cells M(1,1)~M(1,6) in the first column and row 1~6 are configured as memory cell C11, and the memory cells M(1,7)~M(1,12) in the first column and row 7~12 are configured as memory cell C12, and so on.

在本實施例中,記憶體裝置2000經由位元線輸入搜尋偏壓。搜尋偏壓可具有「第一汲極電壓位準VD1」或「第二汲極電壓位準VD2」。搜尋偏壓的第一汲極電壓位準VD1表示搜尋資料「H」而對應至記憶單元M(i,j)儲存的邏輯值「0」的位元。 搜尋偏壓的第二汲極電壓位準VD2表示搜尋資料「L」而對應至記憶單元M(i,j)儲存的邏輯值「1」的位元。搜尋偏壓的第一汲極電壓位準VD1大於第二汲極電壓位準VD2,且第二汲極電壓位準VD2大致為零伏特(第二汲極電壓位準VD2亦可表示為「0v」)。以記憶單元C11為例,經由位元線BL1~BL6輸入電壓位準為「VD2、VD1、VD2、VD2、VD1、VD1」(即,「0v、VD1、0v、0v、VD1、VD1」)的搜尋偏壓而對應至搜尋資料「LHLLHH」,且經由字元線WL1輸入選擇電壓Vsel而施加於記憶單元M(1,1)~M(1,6)的閘極。記憶單元M(1,2)、M(1,5)、M(1,6)被編程為具有第一臨界電壓H-Vt,因此記憶單元M(1,2)、M(1,5)、M(1,6)無法被選擇電壓Vsel導通而無法提供源極電流Is。另一方面,記憶單元M(1,1)、M(1,3)、M(1,4)被編程為具有第二臨界電壓L-Vt因此可被選擇電壓Vsel導通。然而記憶單元M(1,1)、M(1,3)、M(1,4)的汲極經由位元線BL1、BL3、BL4接收第二汲極電壓位準VD2的搜尋偏壓且第二汲極電壓位準VD2大致為零伏特,因此記憶單元M(1,1)、M(1,3)、M(1,4)亦無法提供源極電流Is。由上,當位元線BL1~BL6輸入的搜尋資料「LHLLHH」匹配於記憶胞C11儲存的編碼資料「101100」時,記憶胞C11的記憶單元M(1,1)~M(1,6)無法產生源極電流Is,感測電路300的感應放大器(sensing amplifier,SA)310無法感測到源極電流Is。 In this embodiment, the memory device 2000 inputs the search bias voltage through the bit lines. The search bias can have a "first drain voltage level VD1" or a "second drain voltage level VD2". The first drain voltage level VD1 of the search bias represents the bit of the search data “H” corresponding to the logical value “0” stored in the memory unit M(i,j). The second drain voltage level VD2 of the search bias represents the bit of the search data “L” corresponding to the logical value “1” stored in the memory unit M(i,j). The first drain voltage level VD1 of the search bias is greater than the second drain voltage level VD2, and the second drain voltage level VD2 is approximately zero volts (the second drain voltage level VD2 can also be expressed as “0v”). Taking the memory cell C11 as an example, the input voltage level is “VD2, VD1, VD2, VD2, VD1, VD1” (that is, “0v, VD1, 0v, 0v, VD1, VD1”) through the bit lines BL1~BL6 and the search bias corresponds to the search data “LHLLHH”, and the selection voltage Vsel is input through the word line WL1 to be applied to the memory cells M(1,1)~M(1,6 ) gate. The memory cells M(1,2), M(1,5), M(1,6) are programmed to have the first threshold voltage H-Vt, so the memory cells M(1,2), M(1,5), M(1,6) cannot be turned on by the selection voltage Vsel and cannot provide the source current Is. On the other hand, the memory cells M(1,1), M(1,3), M(1,4) are programmed to have the second threshold voltage L-Vt and thus can be turned on by the selection voltage Vsel. However, the drains of the memory cells M(1,1), M(1,3), and M(1,4) receive the search bias of the second drain voltage level VD2 through the bit lines BL1, BL3, and BL4, and the second drain voltage level VD2 is approximately zero volts, so the memory cells M(1,1), M(1,3), and M(1,4) cannot provide the source current Is. From the above, when the search data “LHLLHH” input by the bit lines BL1-BL6 matches the encoded data “101100” stored in the memory cell C11, the memory cells M(1,1)-M(1,6) of the memory cell C11 cannot generate the source current Is, and the sensing amplifier (sensing amplifier, SA) 310 of the sensing circuit 300 cannot sense the source current Is.

接著,字元線WL2輸入選擇電壓Vsel至記憶胞C21的記憶單元M(2,1)~M(2,6)的閘極。記憶單元M(2,1)、M(2,4)、M(2,6)被編程為具有第一臨界電壓H-Vt因此無法被選擇電壓Vsel導通,無法提供源極電流Is。另一方面,記憶單元M(2,2)、M(2,3)、M(2,5)被編程為具有第二臨界電壓L-Vt而能夠被選擇電壓Vsel導通,其中記憶單元M(2,2)、M(2,5)的汲極從位元線BL2、BL5接收第一汲極電壓位準VD1的搜尋偏壓,因此記憶單元M(2,2)、M(2,5)可產生源極電流Is。由上,當位元線BL1~BL6輸入的搜尋資料「LHLLHH」不匹配於記憶胞C21儲存的編碼資料「「011010」時,感應放大器310可感測到記憶胞C21其中部分的記憶單元(即,記憶單元M(2,2)、M(2,5))的源極電流Is。 Next, the word line WL2 inputs the selection voltage Vsel to the gates of the memory cells M(2,1)˜M(2,6) of the memory cell C21. The memory cells M(2,1), M(2,4), and M(2,6) are programmed to have the first threshold voltage H-Vt, so they cannot be turned on by the selection voltage Vsel, and cannot provide the source current Is. On the other hand, the memory cells M(2,2), M(2,3), M(2,5) are programmed to have the second threshold voltage L-Vt and can be turned on by the selection voltage Vsel, wherein the drains of the memory cells M(2,2), M(2,5) receive the search bias of the first drain voltage level VD1 from the bit line BL2, BL5, so the memory cells M(2,2), M(2,5) can generate the source current Is. From the above, when the search data “LHLLHH” input by the bit lines BL1-BL6 does not match the code data “011010” stored in the memory cell C21, the sense amplifier 310 can sense the source current Is of some of the memory cells in the memory cell C21 (ie, the memory cells M(2,2), M(2,5)).

在第8圖的實施例中,感應放大器310經由源極線SL1~SL12感測源極電流Is以判斷搜尋資料是否匹配於儲存的編碼資料。在其他實施例中,亦可對於感應放大器310的輸出進行計數(count)、累加及加權處理,並分析搜尋資料與儲存的編碼資料的匹配程度。 In the embodiment of FIG. 8 , the sense amplifier 310 senses the source current Is through the source lines SL1 ˜ SL12 to determine whether the search data matches the stored code data. In other embodiments, the output of the sense amplifier 310 can also be counted, accumulated and weighted, and the matching degree between the search data and the stored code data can be analyzed.

第9圖為本揭示一實施例之NOR型的記憶體裝置3000之電路圖。以參數N=2的四層記憶胞為例,記憶胞C11包括以NOR型式而配置的四個記憶單元M(1,1)~M(1,4)。記憶單元M(1,1)~M(1,4)的控制端(例如為閘極端)分別耦接至字元線WL1~WL4,記憶單元M(1,1)~M(1,4)的第一端(例如為汲極端)共同耦接至匹配訊號線(match signal line)ML1。並且,匹配訊 號線ML1耦接至預充電(pre-charge)電晶體MP-1的源極及感測電路300。 FIG. 9 is a circuit diagram of a NOR memory device 3000 according to an embodiment of the present disclosure. Taking a four-layer memory cell with parameter N=2 as an example, the memory cell C11 includes four memory cells M(1,1)˜M(1,4) configured in a NOR type. The control terminals (for example, gate terminals) of the memory cells M(1,1)-M(1,4) are respectively coupled to the word lines WL1-WL4, and the first terminals (for example, drain terminals) of the memory cells M(1,1)-M(1,4) are commonly coupled to a match signal line (match signal line) ML1. And, match the message The signal line ML1 is coupled to the source of the pre-charge transistor MP- 1 and the sense circuit 300 .

預充電電晶體MP-1的汲極耦接於直流電壓源或接收直流電壓訊號(圖中未顯示),並且直流電壓源或直流電壓訊號具有「第一匹配電壓位準VM1」,因此預充電電晶體MP-1的汲極亦具有第一匹配電壓位準VM1。並且,預充電電晶體MP-1的閘極接收控制訊號St。在輸入搜尋偏壓以進行資料搜尋前,控制訊號St可導通預充電電晶體MP-1,以將預充電電晶體MP-1的源極上拉(pull-up)至第一匹配電壓位準VM1,使得匹配訊號線ML1亦被上拉至第一匹配電壓位準VM1。在輸入搜尋偏壓後,根據搜尋偏壓與記憶單元的儲存資料的匹配結果,匹配訊號線ML1可維持於第一匹配電壓位準VM1或被下拉(pulled-down)至「第二匹配電壓位準VM2」。其中,第一匹配電壓位準VM1大於第二匹配電壓位準VM2。 The drain of the pre-charging transistor MP-1 is coupled to a DC voltage source or receives a DC voltage signal (not shown in the figure), and the DC voltage source or the DC voltage signal has a “first matching voltage level VM1”, so the drain of the pre-charging transistor MP-1 also has a first matching voltage level VM1. Moreover, the gate of the precharge transistor MP- 1 receives the control signal St. Before inputting the search bias for data search, the control signal St can turn on the precharge transistor MP-1 to pull up the source of the precharge transistor MP-1 to the first matching voltage level VM1, so that the matching signal line ML1 is also pulled up to the first matching voltage level VM1. After the search bias is input, the matching signal line ML1 can be maintained at the first matching voltage level VM1 or pulled-down to the “second matching voltage level VM2” according to the matching result of the search bias voltage and the stored data of the memory unit. Wherein, the first matching voltage level VM1 is greater than the second matching voltage level VM2.

在本實施例中,可經由字元線WL1~WL4輸入搜尋偏壓。搜尋偏壓可具有「第一閘極電壓位準VS1」或「第二閘極電壓位準VS2」。搜尋偏壓的第一閘極電壓位準VS1表示搜尋資料「H」而對應至記憶單元M(i,j)儲存的邏輯值「0」的位元。搜尋偏壓的第二閘極電壓位準VS2表示搜尋資料「L」而對應至記憶單元M(i,j)儲存的邏輯值「1」的位元。搜尋偏壓的第一閘極電壓位準VS1大於第二閘極電壓位準VS2,第一閘極電壓位準VS1大於該第一臨界電壓H-Vt,且第二閘極電壓位準VS2大致為零伏 特(第二閘極電壓位準VS2亦可表示為「0v」)。表二所示為記憶體裝置3000的記憶胞儲存的原始資料、編碼資料及字元線輸入的搜尋偏壓、搜尋資料的對應關係:

Figure 111108079-A0305-02-0022-10
In this embodiment, the search bias can be input through the word lines WL1 - WL4 . The search bias can have "first gate voltage level VS1" or "second gate voltage level VS2". The first gate voltage level VS1 of the search bias represents the search data "H" corresponding to the bit of logic value "0" stored in the memory unit M(i,j). The second gate voltage level VS2 of the search bias represents the bit of the search data “L” corresponding to the logical value “1” stored in the memory unit M(i,j). The first gate voltage level VS1 of the search bias is greater than the second gate voltage level VS2, the first gate voltage level VS1 is greater than the first threshold voltage H-Vt, and the second gate voltage level VS2 is approximately zero volts (the second gate voltage level VS2 can also be expressed as “0v”). Table 2 shows the corresponding relationship between the raw data stored in the memory cell of the memory device 3000, the coded data, the search bias voltage input by the word line, and the search data:
Figure 111108079-A0305-02-0022-10

第10圖為第9圖的記憶體裝置3000之匹配訊號線ML1、MLn的電壓位準V-ML1、V-MLn之時序變化圖。同時參見第9、10圖及表二,經由字元線WL1~WL4輸入電壓位準為「VS1、VS1、VS2、VS2」(即,「VS1、VS1、0v、0v」)的搜尋偏壓以表示搜尋資料「HHLL」,其匹配於記憶胞C11的記憶單元M(1,1)~M(1,4)所儲存的編碼資料「0011」(對應至兩個位元的原始資料「00」)。記憶單元M(1,1)~M(1,4)的臨界電壓編程為「H-Vt、H-Vt、L-Vt、L-Vt」,記憶單元M(1,1)~M(1,4)不會被電壓位準為「VS1、VS1、0v、0v」的搜尋偏壓導通,因此匹 配訊號線ML1的電壓位準V-ML1不會被下拉至電壓位準較低的第二匹配電壓位準VM2,感測電路300可感測到匹配訊號線ML1的電壓位準V-ML1仍維持於電壓位準較高的第一匹配電壓位準VM1。據此,可判斷記憶胞C11儲存的編碼資料「0011」匹配於搜尋資料「HHLL」。類似的,記憶胞C12的記憶單元M(1,5)~M(1,8)不會被電壓位準為「VS1、VS1、0v、0v」的搜尋偏壓導通,因此匹配訊號線ML1的電壓位準V-ML1仍維持於第一匹配電壓位準VM1。 FIG. 10 is a timing change diagram of the voltage levels V-ML1 and V-MLn of the matching signal lines ML1 and MLn of the memory device 3000 in FIG. 9 . At the same time, see Figure 9, 10, and Table 2. The voltage level is entered through the WL1 ~ WL4 of the word line. The information "0011" ("00" corresponding to the two -bit price). The threshold voltages of the memory cells M(1,1)~M(1,4) are programmed as "H-Vt, H-Vt, L-Vt, L-Vt", and the memory cells M(1,1)~M(1,4) will not be turned on by the search bias whose voltage levels are "VS1, VS1, 0v, 0v". The voltage level V-ML1 of the matching signal line ML1 will not be pulled down to the second matching voltage level VM2 which is a lower voltage level, and the sensing circuit 300 can sense that the voltage level V-ML1 of the matching signal line ML1 is still maintained at the first matching voltage level VM1 which is a higher voltage level. Accordingly, it can be determined that the encoded data “0011” stored in the memory cell C11 matches the search data “HHLL”. Similarly, the memory cells M(1,5)-M(1,8) of the memory cell C12 will not be turned on by the search bias with the voltage levels “VS1, VS1, 0v, 0v”, so the voltage level V-ML1 of the matching signal line ML1 is still maintained at the first matching voltage level VM1.

另一方面,配置於第n列的記憶胞Cn1的四個記憶單元M(n,1)~M(n,4)儲存了編碼資料「0101」,其不匹配於字元線WL1~WL4輸入的搜尋資料「HHLL」。其中,記憶單元M(n,2)編程為具有第二臨界電壓L-Vt而能夠被字元線WL2輸入的第一閘極電壓位準VS1的搜尋偏壓導通,因此匹配訊號線MLn的電壓位準VM-n可經由導通的記憶單元M(n,2)被下拉至第二匹配電壓位準。感測電路300感測到電壓位準VM-n小於參考電壓Vref時,可判斷記憶胞Cn1的儲存資料不匹配於搜尋資料。類似的,記憶胞Cn2的記憶單元M(n,6)能夠被導通,電壓位準VM-n可經由記憶單元M(n,6)被下拉至第二匹配電壓位準VM2。 On the other hand, the four memory cells M(n,1)~M(n,4) of the memory cell Cn1 in the nth column store the encoded data "0101", which does not match the search data "HHLL" input by the word lines WL1~WL4. Wherein, the memory cell M(n,2) is programmed to have the second threshold voltage L-Vt and can be turned on by the search bias of the first gate voltage level VS1 input from the word line WL2, so the voltage level VM-n of the matching signal line MLn can be pulled down to the second matching voltage level through the turned-on memory cell M(n,2). When the sensing circuit 300 senses that the voltage level VM-n is lower than the reference voltage Vref, it can determine that the storage data of the memory cell Cn1 does not match the search data. Similarly, the memory unit M(n,6) of the memory cell Cn2 can be turned on, and the voltage level VM-n can be pulled down to the second matching voltage level VM2 via the memory unit M(n,6).

第11圖為本揭示另一實施例之NOR型的記憶體裝置3000b之電路圖。記憶體裝置3000b的記憶胞C11、C21、C31及C41分別耦接於匹配訊號線ML1、ML2、ML3及ML4,並且記憶胞C11、C21、C31及C41的每一者都經由字元線WL1~WL4 接收搜尋偏壓。記憶體裝置3000b的字元線WL1~WL4的搜尋偏壓可進行反向(inverse)編碼。表三所示為記憶體裝置3000b的記憶胞儲存的原始資料、編碼資料及字元線輸入的搜尋偏壓的對應關係,其中,表二之搜尋偏壓的第一閘極電壓位準VS1反向為表三之搜尋偏壓的第二閘極電壓位準VS2(第二閘極電壓位準VS2可被表示為「0v」),反之亦然。 FIG. 11 is a circuit diagram of a NOR memory device 3000b according to another embodiment of the present disclosure. The memory cells C11, C21, C31 and C41 of the memory device 3000b are respectively coupled to the matching signal lines ML1, ML2, ML3 and ML4, and each of the memory cells C11, C21, C31 and C41 is connected via the word lines WL1˜WL4 Receive seek bias. The search biases of the word lines WL1 - WL4 of the memory device 3000b can be encoded inversely. Table 3 shows the corresponding relationship between the original data stored in the memory cell of the memory device 3000b, the coded data, and the search bias voltage input by the word line, wherein the first gate voltage level VS1 of the search bias voltage in Table 2 is reversed to the second gate voltage level VS2 of the search bias voltage in Table 3 (the second gate voltage level VS2 can be expressed as "0v"), and vice versa.

Figure 111108079-A0305-02-0024-11
Figure 111108079-A0305-02-0024-11

第12圖為第9圖的記憶體裝置3000b之另一實施例中,匹配訊號線ML1~ML4的電壓位準V-ML1~V-ML4之時序變化圖。請同時參見第11、12圖及表三,記憶胞C11儲存的編碼資料「1001」匹配於電壓位準為「VS1、0v、0v、VS1」的搜尋偏壓,記憶胞C11的記憶單元M(1,1)、M(1,4)具有第二臨界電壓L-Vt而能夠被第一閘極電壓位準VS1的搜尋偏壓導通。因此,匹配訊號線ML1的電壓位準V-ML1可經由兩個導通的記憶單元 M(1,1)、M(1,4)以第一放電速度而較快速的被下拉至第二匹配電壓位準VM2。感測電路300可感測到電壓位準V-ML1較快速的下降而小於參考電壓Vref。 FIG. 12 is a timing change diagram of the voltage levels V-ML1-V-ML4 of the matching signal lines ML1-ML4 in another embodiment of the memory device 3000b shown in FIG. Please refer to Figures 11 and 12 and Table 3 at the same time. The encoded data "1001" stored in the memory cell C11 matches the search bias voltage level "VS1, 0v, 0v, VS1". The memory cells M(1,1) and M(1,4) of the memory cell C11 have the second critical voltage L-Vt and can be turned on by the search bias voltage of the first gate voltage level VS1. Therefore, the voltage level V-ML1 of the matching signal line ML1 can pass through two memory cells that are turned on. M(1,1), M(1,4) are quickly pulled down to the second matching voltage level VM2 at the first discharge rate. The sensing circuit 300 can sense a relatively rapid drop of the voltage level V-ML1 which is smaller than the reference voltage Vref.

此外,記憶胞C21儲存的編碼資料「0011」部分不匹配於電壓位準為「VS1、0v、0v、VS1」的搜尋偏壓,記憶胞C21的記憶單元M(2,4)具有第二臨界電壓L-Vt而能夠被第一閘極電壓位準VS1的搜尋偏壓導通。因此,匹配訊號線ML2的電壓位準V-ML2僅經由一個記憶單元M(2,4)以第二放電速度而較慢速的被下拉至第二匹配電壓位準,第二放電速度小於第一放電速度。類似的,記憶胞C31儲存的編碼資料「0101」部分不匹配於電壓位準為「VS1、0v、0v、VS1」的搜尋偏壓,記憶胞C31的記憶單元M(3,4)具有第二臨界電壓L-Vt而能夠被第一閘極電壓位準VS1的搜尋偏壓導通。因此,匹配訊號線ML3的電壓位準V-ML3僅經由一個記憶單元M(3,4)以第二放電速度而較慢速的被下拉至第二匹配電壓位準VM2。 In addition, part of the encoded data “0011” stored in the memory cell C21 does not match the search bias voltage level of “VS1, 0v, 0v, VS1”. The memory cell M(2, 4) of the memory cell C21 has the second critical voltage L-Vt and can be turned on by the search bias voltage of the first gate voltage level VS1. Therefore, the voltage level V-ML2 of the matching signal line ML2 is pulled down to the second matching voltage level at a slower rate through only one memory cell M(2,4) at a second discharge rate, which is smaller than the first discharge rate. Similarly, part of the encoded data “0101” stored in the memory cell C31 does not match the search bias voltage level of “VS1, 0v, 0v, VS1”, and the memory cell M(3,4) of the memory cell C31 has the second threshold voltage L-Vt and can be turned on by the search bias voltage of the first gate voltage level VS1. Therefore, the voltage level V-ML3 of the matching signal line ML3 is pulled down to the second matching voltage level VM2 at a slower rate through only one memory unit M(3,4) at the second discharge rate.

另一方面,記憶胞C41儲存的編碼資料「0110」完全不匹配於電壓位準為「VS1、0v、0v、VS1」的搜尋偏壓,即,編碼資料「0110」的每一位元都不匹配於電壓位準為「VS1、0v、0v、VS1」的搜尋偏壓。記憶胞C41的記憶單元M(4,1)~M(4,4)都不被導通,匹配訊號線ML4的電壓位準V-ML4仍維持為電壓位準較高的第一匹配電壓位準VM1。 On the other hand, the encoded data “0110” stored in the memory cell C41 does not match the search bias at voltage levels “VS1, 0v, 0v, VS1”, that is, each bit of the encoded data “0110” does not match the search bias at voltage levels “VS1, 0v, 0v, VS1”. The memory cells M(4,1)-M(4,4) of the memory cell C41 are not turned on, and the voltage level V-ML4 of the matching signal line ML4 remains at the first matching voltage level VM1 which is a higher voltage level.

由上,記憶體裝置3000b根據匹配訊號線ML1~ML4的電壓位準V-ML1~V-ML4被下拉至第二匹配電壓位準的時間(即,放電時間)來判斷記憶胞C11~C41的儲存資料與搜尋資料的匹配程度。 From the above, the memory device 3000b judges the matching degree of the storage data and the search data of the memory cells C11-C41 according to the time (that is, the discharge time) when the voltage levels V-ML1-V-ML4 of the matching signal lines ML1-ML4 are pulled down to the second matching voltage level.

並且,在第9圖的記憶體裝置3000的另一實施例(未圖示)中,可將記憶單元的儲存資料進行反向編碼,如表四之示例。表四之編碼資料與表二之編碼資料的邏輯值為反向。 Moreover, in another embodiment (not shown) of the memory device 3000 in FIG. 9 , the storage data of the memory unit can be reverse-coded, as shown in Table 4. The logical value of the coded data in Table 4 and the coded data in Table 2 is reversed.

Figure 111108079-A0305-02-0026-13
Figure 111108079-A0305-02-0026-13

根據以上各實施例的記憶體裝置,利用編碼機制將M個位元的原始資料編碼成為2N個位元的編碼資料,並儲存於一個記憶胞的2N個記憶單元中,以提高儲存內容的密度。可經由字元線或位元線輸入搜尋偏壓而對應至搜尋資料,以對於NAND型或NOR型的記憶體裝置儲存的編碼資料進行搜尋。並且,可判斷編碼資料與搜尋資料的匹配程度。 According to the memory devices of the above embodiments, the original data of M bits is encoded into 2N bits of encoded data by using the encoding mechanism, and stored in 2N memory units of one memory cell, so as to increase the density of storage content. The search bias can be input through the word line or the bit line to correspond to the search data, so as to search the encoded data stored in the NAND type or NOR type memory device. Moreover, the degree of matching between the coded data and the searched data can be judged.

雖然本發明已以較佳實施例及範例詳細揭露如上,可理解的是,此些範例意指說明而非限制之意義。可預期的是,所屬技術領域中具有通常知識者可想到多種修改及組合,其多種修改及組合落在本發明之精神以及後附之申請專利範圍之範圍內。 Although the present invention has been disclosed above in detail with preferred embodiments and examples, it should be understood that these examples are meant to be illustrative rather than limiting. It is expected that those skilled in the art can conceive various modifications and combinations, and the various modifications and combinations fall within the spirit of the present invention and the scope of the appended patent application.

1000:記憶體裝置 1000: memory device

100:第一驅動電路 100: the first drive circuit

200:第二驅動電路 200: the second drive circuit

300:感測電路 300: sensing circuit

400:記憶體內搜尋(IMS)陣列 400: In-memory search (IMS) array

WL1,WL2,WL2N,WL(2N+1),WL(2N+2):字元線 WL1, WL2, WL2N, WL(2N+1), WL(2N+2): word line

WL4N,WLm:字元線 WL4N,WLm: word line

BL1,BL2,BLn:位元線 BL1, BL2, BLn: bit lines

SL1,SL2,SLn:源極線 SL1, SL2, SLn: source lines

1/0:邏輯值 1/0: logical value

Is:源極電流 Is: source current

C11,C12,C21,C22:記憶胞 C11, C12, C21, C22: memory cells

M(1,1),M(2,1),M(2N,1),M(2N+1,1):記憶單元 M(1,1), M(2,1), M(2N,1), M(2N+1,1): memory unit

M(2N+2,1),M(4N,1),M(1,2),M(2,2),M(2N,2):記憶單元 M(2N+2,1), M(4N,1), M(1,2), M(2,2), M(2N,2): memory unit

M(2N+1,2),M(2N+2,2),M(4N,2):記憶單元 M(2N+1,2), M(2N+2,2), M(4N,2): memory unit

Claims (18)

一種記憶體裝置,包括:一第一驅動電路;一第二驅動電路;一感測電路;以及一記憶體內搜尋(IMS)陣列,包括複數個記憶單元、複數條字元線及複數條位元線,該些記憶單元被配置為複數橫向列及複數縱向行,同一縱向行的該些記憶單元各自的一控制端經由該些字元線之對應一者耦接至該第一驅動電路,同一縱向行的該些記憶單元串聯連接並經由該些位元線分別耦接至該第二驅動電路、且經由一源極線共同耦接至該感測電路,其中,同一縱向行的每2N個相鄰的記憶單元被配置為一記憶胞,該記憶胞用以儲存2N個位元的編碼資料以表示M個位元的原始資料,N與M皆為大於或等於2的正整數,且N與M的關係式為N=(M/2)+1,該編碼資料具有一第一編碼區域與一第二編碼區域,該第一編碼區域包括該編碼資料的第1個位元至第N個位元,該第二編碼區域包括該編碼資料的第(N+1)個位元至第2N個位元,該編碼資料具有一編碼組合數量,該編碼組合數量相等於
Figure 111108079-A0305-02-0030-14
A memory device, comprising: a first drive circuit; a second drive circuit; a sensing circuit; and an internal memory search (IMS) array, including a plurality of memory cells, a plurality of word lines and a plurality of bit lines, the memory cells are configured as a plurality of horizontal columns and a plurality of vertical rows, a control terminal of each of the memory cells in the same vertical row is coupled to the first drive circuit through a corresponding one of the word lines, and the memory cells in the same vertical row are connected in series and respectively coupled through the bit lines Connected to the second driving circuit and coupled to the sensing circuit through a source line, wherein, every 2N adjacent memory cells in the same longitudinal row are configured as a memory cell, and the memory cell is used to store 2N bits of coded data to represent M bits of original data, N and M are both positive integers greater than or equal to 2, and the relationship between N and M is N=(M/2)+1, the coded data has a first coded area and a second coded area, and the first coded area includes the first coded data of the coded data bit to Nth bit, the second coding area includes the (N+1)th bit to the 2Nth bit of the coded data, the coded data has a number of coded combinations, the number of coded combinations is equal to
Figure 111108079-A0305-02-0030-14
.
如請求項1所述之記憶體裝置,其中該編碼資料的該編碼組合數量大於或等於「二的M次方」。 The memory device according to claim 1, wherein the number of encoding combinations of the encoding data is greater than or equal to "two to the M power". 如請求項1所述之記憶體裝置,其中各該記憶單元被編程為具有一第一臨界電壓以儲存該編碼資料中的邏輯值「0」的位元,或被編程為具有一第二臨界電壓以儲存該編碼資料中的邏輯值「1」的位元,該第一臨界電壓大於該第二臨界電壓。 The memory device as described in claim 1, wherein each of the memory cells is programmed to have a first threshold voltage to store a bit of logic value "0" in the coded data, or is programmed to have a second threshold voltage to store a bit of logic value "1" in the coded data, and the first threshold voltage is greater than the second threshold voltage. 如請求項3所述之記憶體裝置,其中該第一驅動電路經由該些字元線之一者施加一搜尋偏壓至對應的該記憶單元,該搜尋偏壓具有一第一電壓位準而匹配於該編碼資料中的邏輯值「0」的該位元,或該搜尋偏壓具有一第二電壓位準而匹配於該編碼資料中的邏輯值「1」的該位元,其中該搜尋偏壓之該第一電壓位準大於該第一臨界電壓,該搜尋偏壓之該第二電壓位準小於該第一臨界電壓且大於該第二臨界電壓。 The memory device as described in claim 3, wherein the first drive circuit applies a search bias to the corresponding memory cell via one of the word lines, the search bias has a first voltage level and matches the bit of the logic value "0" in the encoded data, or the search bias has a second voltage level and matches the bit of the logic value "1" in the encoded data, wherein the first voltage level of the search bias is greater than the first threshold voltage, and the second voltage level of the search bias is lower than the first The critical voltage is greater than the second critical voltage. 如請求項4所述之記憶體裝置,其中當該搜尋偏壓匹配於該編碼資料中的該位元時,儲存該位元的該記憶單元產生一源極電流,該感測電路經由對應的該源極線感測該源極電流。 The memory device according to claim 4, wherein when the search bias matches the bit in the encoded data, the memory cell storing the bit generates a source current, and the sensing circuit senses the source current through the corresponding source line. 如請求項4所述之記憶體裝置,其中當該編碼資料為2N個邏輯值「1」的位元時,經由該些字元線施加的該些搜尋偏壓匹配於該編碼資料。 The memory device according to claim 4, wherein when the coded data is 2N bits of logical value "1", the search biases applied through the word lines match the coded data. 如請求項4所述之記憶體裝置,其中當經由該些字元線施加的該些搜尋偏壓皆具有該第一電壓位準時,該些搜尋偏壓匹配於該編碼資料。 The memory device as claimed in claim 4, wherein when the search bias voltages applied through the word lines all have the first voltage level, the search bias voltages match the encoded data. 如請求項3所述之記憶體裝置,其中該第一驅動電路經由該些字元線之一者施加一搜尋偏壓至對應的該記憶單元,該搜尋偏壓具有一第三電壓位準而匹配於該編碼資料中的邏輯值「0」的該位元,或該搜尋偏壓具有一第四電壓位準而匹配於該編碼資料中的邏輯值「1」的該位元,其中該搜尋偏壓之該第三電壓位準大於該第四電壓位準,且該搜尋偏壓之該第四電壓位準大於該第一臨界電壓。 The memory device as described in claim 3, wherein the first driving circuit applies a search bias to the corresponding memory cell via one of the word lines, the search bias has a third voltage level and matches the bit of the logic value "0" in the code data, or the search bias has a fourth voltage level and matches the bit of the logic value "1" in the code data, wherein the third voltage level of the search bias is greater than the fourth voltage level, and the fourth voltage level of the search bias is greater than the first threshold voltage. 如請求項8所述之記憶體裝置,其中當該搜尋偏壓匹配於該編碼資料中的該位元時,儲存該位元的該記憶單元產生一第一源極電流,當該搜尋偏壓不匹配於該位元時,該記憶單元產生一第二源極電流,該第一源極電流的電流值大於該第二源極電流,該感測電路經由對應的該源極線感測該第一源極電流及該第二源極電流。 The memory device as described in claim 8, wherein when the search bias matches the bit in the encoded data, the memory cell storing the bit generates a first source current; when the search bias does not match the bit, the memory cell generates a second source current, the current value of the first source current is greater than the second source current, and the sensing circuit senses the first source current and the second source current through the corresponding source line. 如請求項3所述之記憶體裝置,其中該第一驅動電路經由施加一選擇電壓至以選擇該些橫向列的其中一者,且該第二驅動電路經由該些位元線之一者施加一搜尋偏壓至被選擇的該橫向列的對應的該記憶單元,該搜尋偏壓具有一第一汲極電壓位準而匹配於該編碼資料中的邏輯值「0」的該位元,或該搜尋偏壓具有一第二 汲極電壓位準而匹配於該編碼資料中的邏輯值「1」的該位元,該搜尋偏壓之該第一汲極電壓位準大於該第二汲極電壓位準,且該第二汲極電壓位準大致相等於零伏特(0v)。 The memory device as described in claim 3, wherein the first drive circuit selects one of the horizontal columns by applying a selection voltage, and the second drive circuit applies a search bias to the corresponding memory cell of the selected horizontal column through one of the bit lines, the search bias has a first drain voltage level matching the bit of the logic value "0" in the encoded data, or the search bias has a second The drain voltage level matches the bit of logic value "1" in the encoded data, the first drain voltage level of the search bias is greater than the second drain voltage level, and the second drain voltage level is substantially equal to zero volts (0v). 如請求項10所述之記憶體裝置,其中當該搜尋偏壓匹配於該編碼資料中的該位元時,儲存該位元的該記憶單元產生一源極電流,該感測電路經由對應的該源極線感測該源極電流。 The memory device according to claim 10, wherein when the search bias voltage matches the bit in the encoded data, the memory cell storing the bit generates a source current, and the sensing circuit senses the source current through the corresponding source line. 一種記憶體裝置,包括:一第一驅動電路;一感測電路;以及一記憶體內搜尋(IMS)陣列,包括複數個記憶單元、複數條字元線及複數條位元線,該些記憶單元被配置為複數橫向列及複數縱向行,同一橫向列的該些記憶單元各自的一控制端經由該些字元線之對應一者耦接至該第一驅動電路,同一橫向列的該些記憶單元經由一匹配訊號線共同耦接至該感測電路,其中,同一橫向列的每2N個相鄰的記憶單元被配置為一記憶胞,該記憶胞用以儲存2N個位元的編碼資料以表示M個位元的原始資料,N與M皆為大於或等於2的正整數,且N與M的關係式為N=(M/2)+1,該編碼資料具有一第一編碼區域與一第二編碼區域,該第一編碼區域包括該編碼資料的第1個位元至第N個位元,該第二編碼區域包括該編碼資料的第(N+1)個位元至第 2N個位元,該編碼資料具有一編碼組合數量,該編碼組合數量相等於
Figure 111108079-A0305-02-0034-15
A memory device, comprising: a first driving circuit; a sensing circuit; and an internal memory search (IMS) array, including a plurality of memory cells, a plurality of word lines and a plurality of bit lines, the memory cells are configured as a plurality of horizontal rows and a plurality of vertical rows, a control terminal of each of the memory cells of the same horizontal row is coupled to the first driving circuit via a corresponding one of the word lines, and the memory cells of the same horizontal row are commonly coupled to the sensing circuit via a matching signal line, wherein Each 2N adjacent memory cells in the same horizontal row are configured as a memory cell, which is used to store 2N bits of coded data to represent M bits of original data, N and M are both positive integers greater than or equal to 2, and the relationship between N and M is N=(M/2)+1, the coded data has a first coded area and a second coded area, the first coded area includes the first bit to the Nth bit of the coded data, and the second coded area includes the (N+th) of the coded data 1) bit to the 2Nth bit, the encoded data has a number of encoding combinations, the number of encoding combinations is equal to
Figure 111108079-A0305-02-0034-15
.
如請求項12所述之記憶體裝置,其中該編碼資料的該編碼組合數量大於或等於「二的M次方」。 The memory device according to claim 12, wherein the number of encoding combinations of the encoded data is greater than or equal to "two to the M power". 如請求項12所述之記憶體裝置,其中各該記憶單元被編程為具有一第一臨界電壓以儲存該編碼資料中的邏輯值「0」的位元,或被編程為具有一第二臨界電壓以儲存該編碼資料中的邏輯值「1」的位元,該第一臨界電壓大於該第二臨界電壓。 The memory device as described in claim 12, wherein each of the memory cells is programmed to have a first threshold voltage to store a bit of logic value "0" in the coded data, or is programmed to have a second threshold voltage to store a bit of logic value "1" in the coded data, and the first threshold voltage is greater than the second threshold voltage. 如請求項14所述之記憶體裝置,其中該第一驅動電路經由該些字元線之一者施加一搜尋偏壓至對應的該記憶單元,該搜尋偏壓具有一第一閘極電壓位準而匹配於該編碼資料中的邏輯值「0」的該位元,或該搜尋偏壓具有一第二閘極電壓位準而匹配於該編碼資料中的邏輯值「1」的該位元,該搜尋偏壓之該第一閘極電壓位準大於該第二閘極電壓位準,該第一閘極電壓位準大於該第一臨界電壓,且該第二閘極電壓位準大致相等於零伏特(0v)。 The memory device as described in claim 14, wherein the first driving circuit applies a search bias to the corresponding memory cell via one of the word lines, the search bias has a first gate voltage level and matches the bit of the logic value "0" in the encoded data, or the search bias has a second gate voltage level and matches the bit of the logic value "1" in the encoded data, the first gate voltage level of the search bias is greater than the second gate voltage level, the first gate The voltage level is greater than the first threshold voltage, and the second gate voltage level is approximately equal to zero volts (0v). 如請求項15所述之記憶體裝置,其中當該搜尋偏壓匹配於該編碼資料中的該位元時,對應之該匹配訊號線維持於一第一匹配電壓位準,當該搜尋偏壓不匹配於該位元時,該匹配訊號 線被下拉至一第二匹配電壓位準,該第一匹配電壓位準大於該第二匹配電壓位準。 The memory device as described in claim 15, wherein when the search bias voltage matches the bit in the encoded data, the corresponding matching signal line is maintained at a first matching voltage level, and when the search bias voltage does not match the bit, the matching signal line The line is pulled down to a second matching voltage level, the first matching voltage level is greater than the second matching voltage level. 如請求項14所述之記憶體裝置,其中該第一驅動電路經由該些字元線之一者施加一搜尋偏壓至對應的該記憶單元,該搜尋偏壓具有一第一閘極電壓位準而匹配於該編碼資料中的邏輯值「1」的該位元,或該搜尋偏壓具有一第二閘極電壓位準而匹配於該編碼資料中的邏輯值「0」的該位元,該搜尋偏壓之該第一閘極電壓位準大於該第二閘極電壓位準,該第一閘極電壓位準大於該第一臨界電壓,且該第二閘極電壓位準大致相等於零伏特(0v)。 The memory device as described in claim 14, wherein the first drive circuit applies a search bias to the corresponding memory cell via one of the word lines, the search bias has a first gate voltage level matching the bit of the logic value "1" in the encoded data, or the search bias has a second gate voltage level and matches the bit of the logic value "0" in the encoded data, the first gate voltage level of the search bias is greater than the second gate voltage level, the first gate The voltage level is greater than the first threshold voltage, and the second gate voltage level is approximately equal to zero volts (0v). 如請求項17所述之記憶體裝置,其中當該搜尋偏壓匹配於該編碼資料中的該位元時,對應之該匹配訊號線以一第一放電速度被下拉至一第二匹配電壓位準,當該搜尋偏壓不匹配於該位元時,該匹配訊號線維持於一第一匹配電壓位準或以一第二放電速度被下拉至該第二匹配電壓位準,其中該第一匹配電壓位準大於該第二匹配電壓位準,且該第一放電速度大於該第二放電速度。 The memory device according to claim 17, wherein when the search bias voltage matches the bit in the encoded data, the corresponding matching signal line is pulled down to a second matching voltage level at a first discharge rate, and when the search bias voltage does not match the bit, the matching signal line is maintained at a first matching voltage level or is pulled down to the second matching voltage level at a second discharge rate, wherein the first matching voltage level is greater than the second matching voltage level, and the first discharge rate is greater than the second discharge rate.
TW111108079A 2022-03-04 2022-03-04 Memory device with high content density TWI807668B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW111108079A TWI807668B (en) 2022-03-04 2022-03-04 Memory device with high content density

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW111108079A TWI807668B (en) 2022-03-04 2022-03-04 Memory device with high content density

Publications (2)

Publication Number Publication Date
TWI807668B true TWI807668B (en) 2023-07-01
TW202336763A TW202336763A (en) 2023-09-16

Family

ID=88149151

Family Applications (1)

Application Number Title Priority Date Filing Date
TW111108079A TWI807668B (en) 2022-03-04 2022-03-04 Memory device with high content density

Country Status (1)

Country Link
TW (1) TWI807668B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140347933A1 (en) * 2012-08-31 2014-11-27 Aplus Flash Technology, Inc. Nor-based bcam/tcam cell and array with nand scalability
US20150003164A1 (en) * 2008-01-25 2015-01-01 Micron Technology, Inc. Content addressable memory
US20200211640A1 (en) * 2018-12-26 2020-07-02 Western Digital Technologies, Inc. On-chip non-volatile memory (nvm) search
US20210280250A1 (en) * 2020-03-06 2021-09-09 Micron Technology, Inc. Architecture for ternary content-addressable memory search

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150003164A1 (en) * 2008-01-25 2015-01-01 Micron Technology, Inc. Content addressable memory
US20140347933A1 (en) * 2012-08-31 2014-11-27 Aplus Flash Technology, Inc. Nor-based bcam/tcam cell and array with nand scalability
US20200211640A1 (en) * 2018-12-26 2020-07-02 Western Digital Technologies, Inc. On-chip non-volatile memory (nvm) search
US20210280250A1 (en) * 2020-03-06 2021-09-09 Micron Technology, Inc. Architecture for ternary content-addressable memory search

Also Published As

Publication number Publication date
TW202336763A (en) 2023-09-16

Similar Documents

Publication Publication Date Title
KR101954502B1 (en) Apparatuses and methods for identifying an extremum value stored in an array of memory cells
US6064591A (en) Memory system
KR100538883B1 (en) Semiconductor memory apparatus
US10832746B2 (en) Non-volatile in-memory computing device
US6836431B2 (en) Method of programming/reading multi-level flash memory using sensing circuit
US9548120B2 (en) Content addressable memory
JP3251297B2 (en) Non-volatile associative memory
Hu et al. In-memory computing with associative memories: A cross-layer perspective
US9001545B2 (en) NOR-based BCAM/TCAM cell and array with NAND scalability
US11289162B2 (en) Analog content addressable memory utilizing three terminal memory devices
US8054662B2 (en) Content addressable memory array
EP1376607A2 (en) Content addressable memory device and method of operating same
US20140321207A1 (en) Determining soft data for combinations of memory cells
JP3828376B2 (en) Storage system
TWI807668B (en) Memory device with high content density
JP2001202785A (en) Associative memory
US12057179B2 (en) Memory device with high content density
JP2004288347A (en) Associative memory
JP2023044545A (en) Information processor and memory system
CN114816330A (en) Multiply-add operation device and control method of multiply-add operation thereof
US11875850B2 (en) Content addressable memory device, content addressable memory cell and method for data searching with a range or single-bit data
US20230238037A1 (en) Content addressable memory device and method for data searching and comparing thereof
TW202343252A (en) Content addressable memory device, content addressable memory cell and method for data searching and comparing thereof
CN112885392B (en) Storage unit, ternary content addressing memory and electronic equipment
KR101136578B1 (en) Ternary content addressable memory