TW200908007A - Memory and 1-bit error checking method thereof - Google Patents

Abstract

A memory 1-bit error checking method is provided. Firstly, at least one piece of data fragment whose side is 2n bits is received. Next, an error correction code, a parity code and a data code are generated and written in the memory. Then, the at least one piece of data fragment is read from the memory and used as at least one piece of read data fragment. Next, a new error correction code, a new parity code and a new data code are generated. Afterwards, a determination as to whether the at least one piece of read data fragment has a 1-bit error is made. After that, if the at least one piece of read data fragment does not have a 1-bit error, then the at least one piece of read data fragment is outputted.

Description

46PA 200908007 IX. Description of the Invention: [Technical Field] The present invention relates to a memory and a 1-bit read error detection method thereof, and particularly relates to a simple hardware architecture and saving "recall" Space memory and its 1-bit read error detection method. σ [Prior Art] The memory system is used in a variety of data storage applications today, in which the integrity of the data stored in the memory is ensured by the memory design. An important part of the loop is that the error correction code (ECC) is usually used to achieve this requirement. Error correction drink is a data integrity test that can detect and correct 1-bit error. Please refer to FIG. 1A, which is a flow chart of a conventional memory 1-bit read error detection method. First, in step 1〇〇, at least one data fragment of y bits and two sizes is received. , η is an integer greater than or equal to 0. Then, in step 110, a 2 η bit size is generated according to at least one data segment of the 2 Π bit size. Error correction code, which is generated by the error correction detection algorithm (ECC Algorithm). In order to avoid the j-bit error occurring in the error correction code, the error correction code includes the error correction parity. The bit error (ECC parity bit) and its complement. The error correction code belongs to the management bits of the at least one piece of data. Then, in step 120, at least one piece of data and error is repaired 200908007

The 6PA positive code is written in the syllabus. At least one piece of information is read in 'f 130' from the memory in step 140, and the data piece is taken less. Then, the correction code, this new error, reads a sample of the Bellows to generate a new error, and the positive code is also produced by the error correction detection algorithm, ‘At least: 2丄50, according to the error correction job correction piece. Step κη: Whether the fragment is identical to at least one piece of data Error correction code = = action ' If the result is . , then to the second refusal or - 〇Γ) - pen data fragment. The call-to-pen read data segment is identical to at least the segment, and the step St (four) segment is identical to at least the pen data slice. The memory output outputs the at least one judgment read operation is correct, and the record segment is different from at least = read. If at least the pen reading data body judges the reading operation, it is in step no. If it is memorized, the pen reading data == less ==_ will read at least one piece of data. '"Capital fragment, and output, for example:: Ming to = the size is 8 digits of the positive detection algorithm generated ♦ ^ ^ section based on the error correction of the same location ρι error correction code 'including error repair, At least - pen and its complement ΡΓ,, *P4,. In the correction code 1, the error two positive 3 also produces a new error of 6-bit size. The error correction parity bits P1, PUP4, and the error

346PA 200908007 The formula for the erroneous correction of the complement ΡΓ, P2, and P4' of the collocated bits is as follows, the ten series represents, mutually exclusive or, '. (Formula 1) (Formula 2) (Formula 3) (Formula 4) (Formula 4) (Formula 5) (Formula 6) P4 = D7 Ten D6 Ten D5 Ten D4 P2 = D7 Ten D6 Ten D3 Ten D2 P1 = D7 Ten D5 Ten D3 Ten D1 P4,=D3 Ten D2 Ten D10D〇P2' =D5 Ten D4 Ten D1®d〇ΡΓ=D6 Ten Μ10 D2 Ten d 〇Operation 1 has; bit erroneous, that is, at least-pen (four) error correction Expected to read the pen = fragment = wrong bribe money red repudiation or (four) results incomplete visceral New ^, the wrong location of the pen to read the data fragment. Incorrectly invert the wrong bit to correct this error. The above-mentioned conventional memory reading error 込Δ1.A, the measurement method, the error correction detection Equivalent Method (ECC Algorithm) can be « , ^ 1 ^ ^ Bay / IJ and correct 1 bit error kernel for 5 own hidden body The error correction code of the 1-bit 疋 read error detection will make the management bit occupy too much: if it is in a large space, and the hardware structure of the memory cannot be simplified: the memory is empty [invention] The present invention has With regard to a kind of recording body and its method, the error reading unit has a simple hardware structure and saves the printing method by using a simpler one-bit reading error reading element and making a printing method according to the present invention. Memory reading error reading 1, ° memory space test

The 46PA 200908007 method includes, firstly, receiving at least one piece of data, at least one piece of data having a size of 2η bits, and η being an integer greater than or equal to 〇. Connected = = according to at least - the pen data generated error correction, the same ^ check _ data code, the data code corresponds to the error correction code. Further, at least one of the data, the error correction code, the parity check code, and the (10) code are written into the two memory. Then, at least one piece of data is read from the memory to read the data piece, and at least the size of the piece of data read is 2, pen 疒.Λ 70. Then, a new error correction is generated according to at least one piece of read data = a new parity check code and a new data code 'new data code number corresponding to the new = error correction code. Then, based on the error correction code and the new error correction code, the same, the _ check exhaustion and the new parity check code, as well as the information stone horse and new data exhaustion, at least one reading data segment relative to at least one data segment is, at = 1 bit it error. Then, if at least the pen-reading data segment does not have a 1-bit error with respect to the money segment, at least one of the whistle-funding segments is output. According to the present invention, there is proposed a memory comprising a memory cell array and a chip correction circuit. The memory unit array receives and stores at least one piece of data: the memory element reads at least one piece of data from the memory unit array into the data frame, and takes the data piece 'at least one piece of data and at least one size of the pen reading error The system is 2η bits, and η is an integer greater than or equal to 〇. Negative element: The circuit generates an error correction code, a peer check data, and a data code according to at least one data segment, and the data code corresponds to the error correction, and蛉*ί少~ The pen reads the data fragment to generate a new error correction code, a new epoch humiliation and a _data code, and the new data code corresponds to the new error correction code. 9

346PA 200908007 where 'the memory uses the error correction code and the new error correction code, the parity check code and the new parity check code, and the data code and the new poor material code to determine whether at least one piece of read data is relative to at least one piece of data. Has a 1-bit error. In order to make the above-mentioned contents of the present invention more comprehensible, the following detailed description of the preferred embodiment and the accompanying drawings will be described in detail as follows: [Embodiment] The present invention is a memory and its 1-bit The read error detection method uses an error correction code having only one of the complement correction bits (Ecc parity bit), one of its complements, a parity check code of the bit size, and a data code of one bit size. The simple bit reading error detection method is used to detect whether the data stored in the memory has a bit error, so that the memory has a simple hardware structure and saves memory space. Referring to Figure 2, there is shown a flow chart of a 1-bit it read error detection method in accordance with a preferred embodiment of the present invention. This memory ^ bit read error detection method is applied to - memory. First, in step ·, receive at least a pen data segment, the size of the at least pen data segment being 2" bit 7〇 ' η is an integer greater than or equal to G. Then, in step 21, "according to at least the data segment of the pen, an error correction code, a parity card, and a data code" are associated with the error correction code. Error, the positive code is produced by the error correction detection algorithm (10) AlgQrithm), but this error correction rhyme is the error correction parity bit (ECC parity (1) or its complement number, its size is η bit. Next, at least - pen 200908007

The size of the 46PA data segment is 8 bits (D0-D7) as an example. Among them, the formulas for the error correction of the parity bits E1, E2, and E4, and the complements E1, E2', and E4' of the error-corrected parity bits are as follows, the ten-series representative, 'stone repulsion or'. (Formula 7) ( Equation 8) (Equation 9) (Equation 10) (Equation 11) (Equation 12)

E4 = D7 Ten D6 Ten D5 Ten D4 E2 = D7 Ten D6 Ten D3 Ten D2 E1 = D7 Ten D5 Ten D3 Ten D1 E4' Two D3 Ten D2 Ten D1 Ten DO E2' Ten D4 Ten D1 Ten DO

ΕΓ =D6 十 D4 十 D2 十 DO The parity code (par i ty code) is produced by performing a mutually exclusive or (exclusive sinking) action on at least one piece of information. At least - the pen data fragment has an even number of i-times... Pens and tablets have an odd number of 1's and the homotopic check is 1. The formula of the parity check code P or F is as follows (Formula i3) PU P'Shishi_D1_ Body Mind Reading Miscorrected Ma's information Malay to ensure (4) Error correction with =::: Test integrity. When wrong ~ 瑕 low position 疋. When the error is repaired, the data is exhausted as at least one complement of the complement of the same position 70. D戋IV The highest bit of the pen piece. The formula for the asset m 孑D is as follows. Bellows code D = D0 (Formula 14)

146PA 200908007 DJ = D7 r^ ^ C Equation 15) In step 220, at least the _ pen parity check code and data code are written in the batten segment, the error correction code, and the &~ memory. 1 Snap code, parity check code and data gram system are used in the 'error header bit' and the quilt: in the management of the eve data segment, read at least one time from the memory, after the heat 'in step 230 Early buckwheat segment rabbit segment 'This at least one reading piece of information ρ Ma to ^ ^ a reading data in step 240 'based at least: = small: 2n bits. The error correction code, a new parity check code, and the second::: fragment generate a new system corresponding to the new error correction drink. New Iraq =. Ah poor material code 'new data code measurement algorithm generated, its size is also a erroneous correction code is also corrected by error correction error correction of the same bit, new error ^ _ ^ error correction code is wrong error correction code for error correction The parity bit 3 is also the wrong parity bit. When the complement of the complement of the same bit is also corrected for the error, the new error correction code is generated by at least one operation of reading the ^1 bit check code (Parity or -Gr). When the element is performed - the mutual code is the error correcting the same bit, then = error:: the lowest bit of the positive data segment. When the new $ is the complement of a reading element to J, then the data code is; less: two more positive bits. The highest data fragment is captured in step 250, and the bit check code and the new parity check code are corrected according to the error, and the code is stabbed at least one time such as u 夂Beike code and new data code, And there is 1 bit #t组-μ 掌贝贝科片 疋又疋... After everything, in step 260, if at least one read 12

46PA 200908007 The data segment does not have a 1-bit error with respect to at least one piece of data, and at least one piece of read data is output. In step 250, substantially the error correction code and the new error correction stone, the parity check code and the new parity check code, and the data code and the new data code are mutually exclusive or mutually determined to determine at least one read. Whether the data fragment has a 1-bit error. Please refer to Table 1, which is a look-up table of a memory 1-bit read error detection method in accordance with a preferred embodiment of the present invention.

EPD 0 0 X 0 1 0 0 1 1 1 0 X 1 1 X Table 1 In Table 1, “E” is the result of mutual exclusion or action between the error correction code and the new error correction code. “P” is the same position. The result of mutual exclusion or action after the meta-check code and the new parity check code, "D" is the result of mutual exclusion or action of the data code and the new data code, "X" stands for don't care, "1" means that at least one reading data has a 1-bit error, and "0" means that at least one piece of reading data does not have a 1-bit error. When (E, P, D) is (0, 0, X), it means that according to the error correction code 13

346PA 200908007 and new bug fixes. Please at least—(4) fetch (4) none of the bit errors and judge at least based on the same bit check code and the new parity check code—the pen read data >1 segment does not have a bit error, So at least - pen reading: the same is true for at least one piece of data. At this time, if D is i, the two gongs and the new data code are incorrect. , . When (E, P) is (0 ' _, which means that at least the error correction code and the new dream error correction code are judged at least - the pen reading data does not have a 1-bit error: Renyi, the parity check code and the new parity The check code judges that at least one piece of read data has a 1-bit error. At this time, the data is exhausted and the new data code is used for further judgment. When (E, P, D) is (〇, 1, 丨) , that is, the table does not judge that at least one piece of read data has a 1-bit error according to the data code and the new data code, and the 1-bit error is corrected, and the 1-bit error is the corresponding position of the data code and the new data code. When (E, p, W is (〇, 1, 0), it means that at least one item is taken according to the data code and the new data code.] The material fragment does not have a 1-bit error, that is, the error occurs. The parity check code and the new parity check code. When the (EPD) is (1, 0, X), it means that the error correction code is used, and the new error correction code determines that at least the pen read data segment has a 1-bit error. However, based on the parity check code and the new parity check code, at least one 5 bar is determined. The material segment does not have a 1-bit error. In this case, the error-generating code is the error correction code and the new error correction code, and at least one of the read data segments does not have a 1-bit error with respect to at least the pen data segment. When D) is (1,1,X), it means that at least the code is based on the error correction code and the new error correction code segment has a bit error.

46PA 200908007 Error 'and according to the parity check code and _ bit check code Lin at least one reading Ϊ data segment has 1 bit error, in this case, at least one reading bait > | segment has 1 疋 error, Therefore, according to the error correction code and the new error correction code, the Xiang error correction detection algorithm corrects at least one bit error of the pen data segment. w Take at least one piece of data (00101011), and at least one piece of reading the scam is __ as an example. According to the above memory method, at least the error of the fragment is described to::_sign)=(111) 'The parity check code is 0'. The new error correction code of the data code is the data/code. (P4 P2 P1) = ίΓΓ : The parity check code is 1, and the new data code is 〇. Therefore, the lowest bit can be obtained (G ’ 1,^ ’, that is, at least one bit of the read data segment is found to have a 1-bit error. The segment is _01011). d is to 〆<Read the cultivar 3, read the memory of the continuation of the embodiment _ including the memory cell array 枷 and the wrong two transcripts: the slice 32 two remembers the cell array 310 The system uses at least the memory 300 to read the data from the memory unit array 310 to at least one reading of the film.

DataR, to φ - tolerance β ~ 々子

DataR m paragraph shirt and at least one reading data fragment • The small system is 2n bits 'n is greater than or equal to 〇 敕 生 生 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 320 Code and data exhaustion, corresponding to the error correction code, and according to at least - pen read data fragment ', 15

346PA 200908007 New error correction code, - new parity check code and a new capital code, the new shell code corresponds to the new error correction code. The element or its number 2 is corrected by the error correction bit, and the size of the w code is n yuan. The size of the code and the new parity check code are 1 bit. Bit. Wherein, when the error is to correct the parity bit, the data code reads the lowest bit 70 of the data segment, and the new (four) code system is at least one error correction number r:: positive code and new error correction code month ^ The number, the shell code is at least one piece of data ^ μ 'new data code is at least - the pen reading data fragment is still in place. The memory 300 uses the error correction code and the new error correction code, the parity: the code and the new parity check code, as well as the data code and the new data code, and judges whether a small piece of the read data segment is relative to at least the data segment of the pen. Bit 7L error. In addition, the 1-bit read two==detection method used in the memory 3〇〇 is described in detail in the above-described embodiment of the present invention. This will not be repeated. The memory disclosed in the above embodiment and its 1-bit read h error; U / then method 'utilizes error correction with only one of the error correction parity bits (ECC (4) ty u) or one of its complements The code, the 1-bit size of the parity check, and the 1st 7C size data code, with a simple 1 bit read error = double test method to detect whether the data stored in the memory has i bit error: For at least one data segment of the size of 2 digits, the memory of the invention disclosed in the present invention is compared with the management bit of the 2 η bit size required for the 1-bit read error detection method of the memory. The body 1-bit read error detection method only needs the management bit of η+2 bit size, which greatly saves the memory space and makes the memory have a simple hardware architecture. In addition, the Hamming code is also used in the tradition to ensure the integrity of the data. The Hamming code attaches a check bit to a specific location of at least one piece of data. For example, when the size of at least one piece of data is η bits, then k check bits are appended to 2°, 21, 22... and 2k_1 bits according to Equation 7 below, where k is a positive integer. 2kgn + k+l (Equation 7) where the number of the check bits and their corresponding bits must be an even number. The check bit is used to detect and correct at least one bit error of a piece of data. However, since the bits in the memory are adjacent to each other and the Hamming code has to add a check bit at a specific position, the Hamming code is not suitable for the memory. In summary, although the present invention has been implemented in a preferred embodiment It is to be understood that the invention is not intended to limit the invention, and that the invention may be modified and modified without departing from the spirit and scope of the invention. The scope is subject to the definition of the scope of the patent application attached.

346PA 200908007 [Simple description of the diagram] Figure 1A shows a flow chart of the conventional memory 1-bit read error detection method. FIG. 1B is a schematic diagram showing an error correction detection algorithm. FIG. 2 is a flow chart showing a method for detecting a 1-bit read error of a memory according to a preferred embodiment of the present invention. Figure 3 is a block diagram of a memory in accordance with a preferred embodiment of the present invention. [Main component symbol description] 300 Memory 310 Memory cell array 320 Error correction circuit 18

Claims (1)

46PA 200908007 X. Patent application scope: 1. A memory 丨 喊 喊 接收 接收 接收 接收 接收 接收 接收 接收 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误 错误Pen data piece 2 (four) yuan check drink and a data code, the wrong secret code, - parity code; the shame is corresponding to the error correction, the at least one information y code and the data drink written in - memory and body The error correction code, the parity check reads the 5 + data segment from the memory, and the at least the data segment is at least one reading element; the size of the money segment is 2n according to the at least one money. - owing to drink, a new parity check, a new error correction should be generated in the new error correction code I·, 斤 data code, the new data code is relative to the segment does not have i pen read data segment relative to If at least one of the funds has a 7C error, then the pen 苇 科 I I apply for the patent scope 坌 7 pen # take the data fragment. The error detection method, _, the memory 1 bit read as the error correction of the parity bit error correction code and the new error correction code '^ data code is the at least - pen data piece 19 546PA 200908007 • The lowest bit of the segment, the new data code is the lowest bit of the at least one piece of read data. 3. The memory 1-bit read error detecting method according to claim 1, wherein when the error correction code and the new error correction code are error-corrected complement bits, the data code The highest bit of the at least one piece of data, the new item code being the southernmost bit of the at least one piece of read data. 4. The memory 1-bit read f error detecting method according to claim 1, wherein the step of determining whether the at least one read data segment has a 1-bit error with respect to the at least one data segment is further The method includes: determining, according to the error correction code and the new error correction code, that the at least one piece of read data has a 1-bit error, and determining the at least one piece of read data according to the same check code and the new parity check code If there is a 1-bit error, the 1-bit error is corrected based on the error correction code and the new error correction code. 5. The memory 1-bit read error detection method according to claim 1, wherein the step of determining whether the at least one read data segment has a 1-bit error with respect to the at least one data segment is further The method includes: determining, according to the error correction code and the new error correction code, that the at least one piece of read data has a 1-bit error, but determining the at least one piece of read data according to the same check code and the new parity check code If there is no 1-bit error, the error correction code and the new error correction code generate a 20 46PA 200908007 error, and the at least one read data segment does not have a 1-bit error with respect to the at least one piece of data. 6. The memory 1-bit read error detection method of claim 1, wherein the step of determining whether the at least one read data segment has a 1-bit error with respect to the at least one data segment further comprises: Determining, according to the error correction code and the new error correction code, that the at least one piece of read data does not have a 1-bit error, but determining the at least one piece of read data according to the same check code and the new parity check code If there is a 1-bit error, the data code and the new data code are viewed. When the at least one read data segment is judged to have a 1-bit error according to the data code and the new data code, the 1-bit error is corrected. 7. The memory 1-bit read error detection method of claim 1, wherein the step of determining whether the at least one read data segment has a 1-bit error with respect to the at least one data segment further comprises: Determining, according to the error correction code and the new error correction code, that the at least one piece of read data does not have a 1-bit error, and determining, according to the data code and the new data code, that the at least one piece of read data does not have one bit a meta error, but according to the same bit check code and the new parity check code, the at least one read data segment has a 1-bit error, and the same check code and the new parity check code generate an error, the at least one The read data segment does not have a 1-bit error with respect to the at least one piece of data. 21 146PA 200908007 8. If the patent application scope 苐丨 error detection method, wherein it is determined whether the 1 bit of the small memory is read in the at least the data segment of the pen - the data segment of the pen is read: /, 1 bit The step of the erroneous error is further corrected. According to the error, the sequel and the new snippet of the read data segment do not have "li-a, positive code judges the at least check code and the new epoch check brown sentence = error, and according to the same position The meta-test does not have a 1-bit it error, but according to ^ at least - the pen reads the data segment at least - the pen reads the data segment has the exhaustion and the new data code determines that the new data code generates an error, the ^ less j-yuan error' Then the data stone horse and the less data segment do not have a 1-bit read data segment relative to the memory of the memory, including: % a memory cell array, a segment from which the memory is from the memory cell = Storing at least one piece of information for at least a pen to read a piece of data, taking the at least one piece of data - the size of the pen reading piece of data is a piece of pen data and the at least integer; and ', ', ', bit ' η Greater than or equal to 〇-error correction circuit, which is based on error correction code, - parity check, - pen data discrimination - corresponds to the error correction code, and ^ 摅 and a 贲 code, the data code is a new error Correction code, - new ^ The at least one reading data segment produces a new data code corresponding to the two-digit check code and a new data code, the error correction 蝎; 玛, the same-place reading correction mother and the The new error corrects the new parity check code, and the data code 22 46PA 200908007 and the new data code, and determines whether the at least one read data segment has a 1-bit error with respect to the at least one data segment. The memory of claim 9, wherein when the error correction code and the new error correction code are error correction parity bits, the data code is the lowest bit of the at least one data segment, and the new data code The memory of the at least one read data segment. The memory of claim 9, wherein the error correction code and the new error correction code are When the error correction bits of the same bit complement the information code system for at least the highest bit piece of data fragments, the new data code reading system for at least a sum of the highest bit data fragments. 23