Memory device with error correction function and operation method thereof
By introducing ECC circuits into the memory device, error detection and correction of multiple memory bank data is achieved, the problem of increasing bit error rate in semiconductor memory devices is solved, the speed and accuracy of the memory device are improved, and the integration is enhanced.
Patent Information
- Application Number
- CN201910317233.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-04-20
- Filing Date
- 2019-04-19
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2039-04-19
AI Technical Summary
With the reduction of the process scale of semiconductor memory equipment, the bit error rate increases rapidly, and the prior art is difficult to effectively solve the problem of data error correction, especially in the data transmission and correction process between multiple memory banks. , affects the speed and accuracy of memory devices.
A memory device is designed, including a memory cell array, control logic and error correction code (ECC) circuit. Through a calibration sub-generator, a parity generator and a calibration sub-decoder, error detection and correction of memory bank data is realized. Ensure error correction is performed in data transfer and write operations between multiple memory banks, improving the speed and accuracy of memory devices.
By introducing ECC circuits into the memory device, data errors between memory banks can be effectively corrected, speed and accuracy of memory devices can be improved, data errors can be reduced, and integration and processing speed of memory devices can be enhanced.
Smart Images

Figure CN110389851B_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the priority benefit of Korean Patent Application No. 10-2018-0046289 filed on April 20, 2018 in the Korean Intellectual Property Office, the disclosure of which is incorporated herein by reference in its entirety. Technical Field
[0003] Exemplary embodiments of the inventive concept relate to a memory device, and more particularly, to a memory device having an error correction function and an operating method thereof. Background Art
[0004] As technology advances, the capacity and speed of semiconductor memory devices continue to increase. Volatile memory devices are semiconductor memory devices that determine data by charges stored in capacitors therein and the data stored therein are lost when power is interrupted.
[0005] The process scale has been reduced to increase the integration density of semiconductor memory devices. As the process scale is reduced, the bit error rate increases rapidly, which can be solved by error correction code (ECC). Summary of the invention
[0006] According to an exemplary embodiment of the present invention, a memory device includes: a memory cell array including a first memory bank and a second memory bank; control logic configured to receive a command from the outside and control an internal operation of the memory device in response to the received command; and an error correction code (ECC) circuit configured to: retain, in a latch circuit, first read data read from the first memory bank in response to a first mask write (MWR) command for the first memory bank based on a latch control signal from the control logic; generate, in response to a first write control signal received from the control logic, a first parity from data in which the first read data retained in the latch circuit is merged with first write data corresponding to the first MWR command; and after generating the first parity, control the ECC operation based on the latch control signal to retain, in the latch circuit, second read data read from the second memory bank.
[0007] According to an exemplary embodiment of the inventive concept, a memory device includes: a memory cell array including a first memory bank and a second memory bank; a control logic configured to receive a write command for the first memory bank and a read command for the second memory bank from the outside, and control the internal operation of the memory device in response to the received write command and read command; and an error correction code (ECC) circuit configured to correct errors included in data. The ECC circuit includes: a syndrome generator configured to receive first read data read from the first memory bank and generate a syndrome indicating that an error exists in the first read data; a parity generator configured to generate parity of first write data received from the control logic while the syndrome generator generates a syndrome for the first read data, and send the first write data and the parity of the received first write data to the second memory bank; and a syndrome decoder configured to correct the error included in the first read data by decoding the generated syndrome.
[0008] According to an exemplary embodiment of the present invention, a method for operating a memory device includes: receiving first read data read from a first memory body, and retaining the first read data in response to a latch control signal received from control logic; outputting the retained first read data in response to a first write control signal received from the control logic after applying the latch control signal; generating a first parity from data in which the first read data is merged with first write data corresponding to a first mask write (MWR) command in response to the first write control signal; and retaining second read data read from a second memory body in response to the latch control signal. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] The above and other features of the present inventive concept will be more clearly understood by describing in detail exemplary embodiments of the present inventive concept with reference to the attached drawings.
[0010] Figure 1 is a block diagram illustrating a memory device according to an exemplary embodiment of the inventive concept.
[0011] Figure 2 is a diagram showing an exemplary embodiment according to the inventive concept Figure 1 Detailed block diagram of a memory device.
[0012] Figure 3 FIG. 2 is a diagram showing a case where a read command and a write command are applied according to an exemplary embodiment of the inventive concept. Figure 1 A block diagram of a memory device.
[0013] Figure 4 is a data flow diagram illustrating a case where a read command and a write command are applied according to an exemplary embodiment of the inventive concept.
[0014] Figure 5 FIG. 2 is a diagram showing an exemplary embodiment of the inventive concept in the case where a mask write command is applied. Figure 1 A block diagram of a memory device.
[0015] Figure 6 is a data flow diagram in the case where a mask write command is applied according to an exemplary embodiment of the inventive concept.
[0016] Figure 7 FIG. 2 is a diagram showing a global input / output (GIO) line including a plurality of memory banks according to an exemplary embodiment of the inventive concept. Figure 1 A block diagram of a memory device.
[0017] Figure 8 FIG. 2 is a diagram showing a method of including an input / output (IO) line at each of a plurality of memory banks according to an exemplary embodiment of the inventive concept. Figure 1 A block diagram of a memory device.
[0018] Fig. 9 According to an exemplary embodiment of the present inventive concept Figure 1 A flowchart of a method for operating a memory device.
[0019] Fig.10 According to an exemplary embodiment of the present inventive concept Figure 1 A detailed flow chart of an operating method of a memory device.
[0020] Fig.11 is a block diagram illustrating a data processing system according to an exemplary embodiment of the inventive concept. DETAILED DESCRIPTION
[0021] Exemplary embodiments of the inventive concept provide a memory device in which a plurality of memory banks include circuits each having an error correction function, and an operating method of the memory device.
[0022] Hereinafter, exemplary embodiments of the inventive concept are described in detail with reference to the accompanying drawings.Throughout the application, the same reference numerals may refer to the same elements.
[0023] Figure 1 is a block diagram illustrating a memory device according to an exemplary embodiment of the inventive concept.
[0024] refer to Figure 1 , the memory device 1000 may include a memory cell array 100 , a control logic 200 , a data control logic 210 , and an error correction code (ECC) circuit 300 .
[0025] The memory cell array 100 may include a plurality of memory cells provided as a matrix arranged in rows and columns. The memory cell array 100 may include a plurality of word lines and a plurality of bit lines connected to the memory cells. A plurality of word lines may be connected to the rows of the memory cells, and a plurality of bit lines may be connected to the columns of the memory cells. The memory cells of the memory cell array 100 may be dynamic random access memory (RAM) (DRAM) cells, static RAM (SRAM) cells, flash memory cells, phase change RAM (PRAM) cells, resistive RAM (ReRAM) cells, magnetoresistive RAM (MRAM) cells, or other types of memory cells. In this case, DRAM may include SDRAM and low power double data rate (LPDDR) RAM.
[0026] The memory cell array 100 may include typical circuits associated with the memory device 1000, such as a row decoder, a column decoder, and a sense amplifier. The row decoder may select any one of a plurality of word lines connected to the memory cell array 100. The row decoder may decode a row address received from the control logic 200 to select any one word line corresponding to the row address, and then activate the selected word line.
[0027] A plurality of memory banks included in the memory cell array 100 may perform various memory operations, such as reading, writing, or masked writing. In addition, a plurality of memory banks may include corresponding sense amplifiers or may share one sense amplifier. In addition, a plurality of memory banks may include input / output lines for reading or writing data. For example, a plurality of memory banks may input and output data via a global input / output (GIO) line. On the other hand, a plurality of memory banks may share one ECC circuit 300.
[0028] The column decoder may select a specific bit line among a plurality of bit lines of the memory cell array 100. The column decoder may decode a column address received from the control logic 200 to select a specific bit line corresponding to the column address. A sense amplifier may be connected to the bit line of the memory cell array 100.
[0029] The sense amplifier can sense the voltage change of a specific bit line among the plurality of bit lines, and amplify and output the sensed voltage change. When the memory cell array 100 includes a GIO line, the data sent via the GIO line can be amplified by the input / output sense amplifier, and the amplified data can be output to the outside via the data input / output pin. Alternatively, the data output from the sense amplifier can be output to the ECC circuit 300.
[0030] The control logic 200 may receive a command / address signal CA from an external device (e.g., a central processing unit (CPU) or a memory controller) of the memory device 1000. An activate command, a read command, a write command, a free charge command, etc. may be issued through a combination of the received command / address signal CA.
[0031] The control logic 200 may receive a command / address signal CA representing a mask write command MWR. The control logic 200 may decode the command / address signal CA corresponding to the mask write command MWR by using a command decoder CMD_DEC, and control the memory device 1000 so that a mask write operation or a read-modify-write operation is performed. As an example, the control logic 200 may control the internal operation of the memory device 1000 by generating an internal command signal. Hereinafter, the read-modify-write operation may be collectively referred to as a mask write operation.
[0032] Control logic 200 may receive a command / address signal CA indicating an address ADDR of a memory cell and a bank to be accessed in memory cell array 100. Address ADDR may be provided to memory cell array 100, and a data access operation of a memory cell addressed by address ADDR may be performed.
[0033] The data access operation may be configured to store a plurality of data words in a storage unit at address ADDR and read the stored data words. Each data word may include, for example, 128 data bits. As another example, each data word may include 32 data bits, 64 data bits, 256 data bits, 512 data bits, etc.
[0034] The data words stored in the memory cell array 100 may have soft data errors or hard data errors. When the data words are stored in the memory cell array 100 or read from the memory cell array 100, one or more data bits in the data words may be flipped, and the value of the data words may be changed. Alternatively, the memory cells in the memory cell array 100 may be damaged, and one or more data bits may be prevented from being accurately stored and read. Therefore, it can be considered that the data words read in the memory cell array 100 may include such data errors.
[0035] In order to reduce such data errors, the memory device 1000 may include an ECC circuit 300 having an ECC to generate parity including one or more parity bits for a data word and store the generated parity in the memory cell array 100. Data output through a data access operation of the memory cell array 100 may be provided to the ECC circuit 300 as a word in which parity is added to the data word.
[0036] The codeword output from the memory cell array 100 to the ECC circuit 300 may be referred to as read data R_DATA associated with the mask write operation. The read data R_DATA may include main data and parity for the main data. For example, the memory cell array 100 may output 136-bit read data R_DATA, the main data may include 128-bit data, and the parity may include 8-bit data.
[0037] The data control logic 210 may receive a data mask signal DM and write data WR_DATA from an external device. The data mask signal DM may indicate a new bit (or byte) to be updated among the received write data WR_DATA and a previous bit to be retained in the read data R_DATA.
[0038] The bits corresponding to the logic low of the data mask signal DM in the write data WR_DATA may be written into the memory cell array 100 as new data. The bits corresponding to the logic high of the data mask signal DM in the write data WR_DATA may be masked, and may instead be retained as bits corresponding to the logic high of the data mask signal DM of the main data constituting the read data R_DATA. Hereinafter, for ease of description, the data output from the first memory bank (BANK 1, memory bank 1) 110 may be referred to as the first read data R_DATA 1, and the data output from the second memory bank (BANK 2, memory bank 2) 120 may be referred to as the second read data R_DATA 2. Each read data R_DATA may include main data indicating substantial data and parity for the substantial data. In addition, the main data may include a first group of data corresponding to the logic high of the data mask signal DM and a second group of data corresponding to the logic low of the data mask signal DM. The number of memory banks is only an example, and may be two or more.
[0039] The second group data of the first read data R_DATA 1 may be written into the memory cell array 100 as the bits corresponding to the logic low of the data mask signal DM in the first write data. In addition, the previous bits to be retained may be the first group data of the first read data R_DATA 1, which are the bits corresponding to the logic high of the data mask signal DM. The first write data may indicate the data that may be written into the first memory bank 110.
[0040] Similarly, assuming that the bits corresponding to the logic low of the data mask signal DM among the plurality of bits of the second read data R_DATA 2 are bits corresponding to the second group data of the second read data, these bits can be written into the memory cell array 100 as bits corresponding to the logic high of the data mask signal DM in the second write data. In addition, the previous bits to be retained can be the first group data of the second read data R_DATA 2, which are bits corresponding to the logic high of the data mask signal DM. The first write data can indicate data that can be written into the first memory bank 110.
[0041] On the other hand, the first read data R_DATA 1 may include a first parity, and the second read data R_DATA 2 may include a second parity.
[0042] Therefore, from the viewpoint of the mask write operation, the second set of data may be data to be updated as the first write data, and the first set of data may be previous data to be retained as is.
[0043] Illustratively, 128 bits of main data may be applied to the memory device 1000. Among the 128 bits of main data, according to the mask write operation, 8 bits of first group data may be retained as is, and 120 bits of second group data may be updated as new data.
[0044] When the control logic 200 does not receive a mask write command from an external device, the ECC circuit 300 may generate parity for the write data WR_DATA. In this case, the write data WR_DATA may be new data to be written to the memory cell array 100. The parity for the write data WR_DATA may be stored in the memory cell array 100 together with the write data WR_DATA. When reading data from the memory cell array 100, the ECC circuit 300 may decode the parity to detect and / or correct any error in the read data R_DATA.
[0045] When the control logic 200 receives a shield write command from an external device, the write data WR_DATA received from the external device may be combined with the read data R_DATA of the shielded portion. In this case, since one or more errors of the previous data may exist before the write data WR_DATA and the previous data are combined, an error correction operation on the previous data may be required. In addition, it may be necessary to generate a new parity for the data in which the write data WR_DATA and the previous data are combined. In this case, an error correction operation on the previous parity may need to precede the generation of the new parity.
[0046] The control logic 200 may receive a mask write command for each of the plurality of memory banks with a small time difference. In this case, due to the characteristics of the mask write operation, a signal combining new data to the data read from the memory bank may be generated after the retention time of the read data. When the mask write operation is performed continuously on the first memory bank 110 and the second memory bank 120, an undesirable result may occur, such that the new data to be combined to the data read from the first memory bank 110 is written to the data read from the second memory bank 120.
[0047] Therefore, the ECC circuit 300 may delay the output time of the first read data R_DATA1 based on the latch control signal LAT_CTRL received from the control logic 200 so that the first write data to be written in the first memory bank 110 and the data read from the first memory bank 110 may be merged and error correction may be performed.
[0048] In this case, the control logic 200 may control the data control logic 210 to transmit the latch control signal LAT_CTRL to the ECC circuit 300 , or may directly transmit the latch control signal LAT_CTRL to the ECC circuit 300 .
[0049] Thereafter, the data control logic 210 may receive the merged data MERGED_DATA from the ECC circuit 300 in which the mask write operation and the error correction have been performed, and may write the received merged data MERGED_DATA into the memory cell array 100 .
[0050] Figure 2 is a diagram showing an exemplary embodiment according to the inventive concept Figure 1 Detailed block diagram of a memory device.
[0051] refer to Figure 2 , the memory cell array 100 may include a first memory bank 110 and a second memory bank 120 .
[0052] The first memory bank 110 and the second memory bank 120 may share one ECC circuit 300. In other words, operations including syndrome generation, syndrome decoding, and parity generation may be performed to correct errors of data included in the first memory bank 110 and data included in the second memory bank 120.
[0053] The ECC circuit 300 may include a syndrome generator 310, a latch circuit 320, a syndrome decoder 330, and a parity generator 340. The ECC circuit 300 may perform various types of ECC encoding and ECC decoding operations to detect errors and correct errors. As an example, the ECC circuit 300 may use parity check, cyclic redundancy code check (CRCC), checksum check, or Hamming code. In addition, the ECC circuit 300 may use correction techniques such as ×4 single device data correction (×4 SDDC), ×8 single-bit error correction and double-bit error detection (SECDED) or lockstep ×8 SDDC to correct errors of multiple memory chips.
[0054] According to an exemplary embodiment of the inventive concept, the ECC circuit 300 may continuously receive first read data R_DATA 1 included in the first memory bank 110 and second read data R_DATA 2 included in the second memory bank 120 and may retain the first and second read data R_DATA 1 and 2 based on a latch control signal LAT_CTRL.
[0055] Thereafter, the ECC circuit 300 may generate a first parity for data that has merged the first read data R_DATA 1 retained in the latch circuit 320 with the first write data based on a first write control signal received from the control logic 200, and may generate a second parity for data that has merged the second read data retained in the latch circuit 320 with the second write data based on a second write control signal received from the control logic 200.
[0056] The syndrome generator 310 may generate a syndrome indicating that one or more errors exist in the read data R_DATA received from the memory cell array 100. For ease of description, it is assumed that the 136-bit read data R_DATA output from the memory cell array 100 includes 128-bit main data and 8-bit parity for the 128-bit main data. In this case, the syndrome generator 310 may generate an 8-bit syndrome for the 128-bit main data.
[0057] The latch circuit 320 may delay an output time point of data received from the memory cell array 100 (a time when data is output from the latch circuit 320 to the syndrome decoder 330 ) for a certain time based on the latch control signal LAT_CTRL.
[0058] When the latch control signal LAT_CTRL is applied, the latch circuit 320 may retain the first read data R_DATA 1, and when the first write control signal is applied, the latch circuit 320 may output the first read data R_DATA 1 retained in the latch circuit 320 to the syndrome decoder 330. In addition, when the latch control signal LATCH CONTRL is sequentially applied, the latch circuit 320 may retain the second read data R_DATA 2, and when the second write control signal is applied, the latch circuit 320 may output the second read data R_DATA 2 retained in the latch circuit 320 to the syndrome decoder 330.
[0059] In this case, the data for which the parity generator 340 generates parity may be data that has been output from the latch circuit 320 and processed through the syndrome decoding operation and the error correction operation. Therefore, the latch circuit 320 may output the data after a certain delay time so that the parity generator 340 may generate parity for the correct data.
[0060] The syndrome decoder 330 can correct the error of the main data by using the decoded signal. As a result of correcting the error of the main data, the corrected main data can be output to the outside of the memory device 1000 to perform a read operation, or can be output to the parity generator 340 to perform a mask write operation.
[0061] The parity generator 340 may generate parity for the received data. In this case, the parity generator 340 may receive the write data WR_DATA from the outside of the memory device 1000 via the data control logic 210. In addition, the parity generator 340 may receive the error-corrected read data R_DATA from the syndrome decoder 330. Therefore, the parity generator 340 may generate parity for the write data WR_DATA, or may generate parity by merging the write data WR_DATA with the corrected read data R_DATA to correspond to the data mask signal DM.
[0062] According to an exemplary embodiment of the inventive concept, the parity generator 340 may generate the first parity and the second parity after the syndrome decoder 330 corrects an error based on the syndromes included in the first read data R_DATA 1 and the second read data R_DATA 2. In this case, the first read data R_DATA 1 and the second read data R_DATA 2 may include data sequentially output from the first memory bank 110 and the second memory bank 120 and the syndromes of the data, respectively.
[0063] In addition, the write data WR_DATA may include the corrected main data received from the syndrome decoder 330. In this case, the parity generator 340 may generate parity of the corrected main data and write the corrected main data and the generated parity into the memory cell array 100.
[0064] The control logic 200 may transmit first read data R_DATA 1 read from the first memory bank 110 and second read data R_DATA 2 read from the second memory bank 120 to the ECC circuit 300 .
[0065] The control logic 200 may receive a read command or a mask write command for the first memory bank 110 and the second memory bank 120 by using a combination of command / address signals CA received from an external device of the memory device 1000. The read command and the mask write command may include addresses ADDR of the first memory bank 110 and the second memory bank 120. Therefore, the read command may include a read command for the first memory bank 110 and a read command for the second memory bank 110, and the same is true for the mask write command.
[0066] In this case, the control logic 200 may control the memory cell array 100 to output data included in the first and second memory banks 110 and 120 to the ECC circuit 300 via an input / output (GIO) line and an input / output sense amplifier.
[0067] When the control logic 200 receives a read command, the ECC circuit 300 may perform an error correction operation including syndrome generation and syndrome decoding to detect an error occurring in a process of reading data included in the first memory bank 110 and the second memory bank 120 .
[0068] Similarly, when the control logic 200 receives a mask write command, the ECC circuit 300 can perform a read operation belonging to a read-modify-write operation on the data included in the first storage body 110 and the second storage body 120, and can perform an error correction operation including syndrome generation and syndrome decoding to detect errors occurring in the process of reading data.
[0069] The control logic 200 may receive a write command or a mask write command for the first memory bank 110 and the second memory bank 120 by using a combination of command / address signals CA received from an external device of the memory device 1000. In addition, the control logic 200 may receive write data WR_DATA of a memory bank having a corresponding address ADDR together with the command / address signal CA.
[0070] When the control logic 200 receives the write command, the control logic 200 may send a write control signal having the write data WR_DATA to the parity generator 340. Therefore, the parity generator 340 may generate parity for the received write data WR_DATA, and may write the write data WR_DATA and the parity of the write data WR_DATA to a memory bank having the corresponding address ADDR via the data control logic 210 based on the address ADDR of the memory bank included in the write control signal.
[0071] When the control logic 200 receives the mask write command, the control logic 200 may send a write control signal generated based on the data mask signal together with the write data WR_DATA to the parity generator 340. In other words, the control logic 200 may control the parity generator 340 to generate parity for data in which the error correction data received from the syndrome decoder 330 and the write data WR_DATA are merged.
[0072] According to an exemplary embodiment of the inventive concept, the parity generator 340 may receive first write data and second write data from the control logic 200. When receiving a first write control signal from the control logic 200, the parity generator 340 may merge the first write data with the first error correction data according to the data mask signal DM, and may generate parity for the merged first write data and first error correction data. When receiving a second write control signal from the control logic 200, the parity generator 340 may merge the second write data with the second error correction data according to the data mask signal DM, and may generate parity for the merged second write data and second error correction data.
[0073] In this case, the control logic 200 can receive a data masking signal DM, a first masking write signal MWR 1, and a second masking write signal MWR 2 from an external device, and can control the ECC circuit 300 so that the first write data and the first error correction data are merged based on the data masking signal DM and the first masking write signal MWR 1, and the second write data and the second error correction data are merged based on the data masking signal DM and the second masking write signal MWR 2.
[0074] Therefore, the parity generator 340 can generate parity for the data obtained by merging the error correction data with the write data received from the syndrome decoder 330, and can write the data and the parity of the data into the memory body having the corresponding address ADDR via the data control logic 210 based on the address ADDR of the memory body received from the control logic 200.
[0075] On the other hand, the memory device 1000 according to the exemplary embodiment of the inventive concept may satisfy the formula tCCDMW=N*tCCD, where tCCD is a column access strobe (CAS) to CAS command delay time, tCCDMW is a CAS to CAS command delay time in a read-modify-write operation, and N is a natural number. In other words, when the memory cell array 100 includes N memory banks, the CAS to CAS command delay time in a read-modify-write operation may be a value calculated by multiplying the CAS to CAS command delay time of the memory device 1000 by N.
[0076] In a masked write operation, the memory device 1000 according to an exemplary embodiment of the present inventive concept may require a time of 1*tCCD for receiving data from the memory cell array 100, a time of 1*tCCD for correcting an error of the read data R_DATA, a time of 1*tCCD for correcting an error of the data in which the masked read data and the write data WR_DATA are merged, and a time of 1*tCCD for writing data in the memory cell array 100.
[0077] In this case, since the ECC circuit 300 according to the exemplary embodiment of the inventive concept separately includes the syndrome generator 310 and the syndrome decoder 330 that correct the error in the read data R_DATA, and the parity generator 340 that corrects the error in the merged data, the four memory banks included in the memory cell array 100 can share one ECC circuit 300 to perform error correction together. In other words, the data read from the four memory banks can be processed by the syndrome generator 310, the latch circuit 320, the syndrome decoder 330, and the parity generator 340 in units of tCCD, respectively. Therefore, the data of the four memory banks can be processed simultaneously in one ECC circuit 300.
[0078] In this case, the four banks may include a first bank 110, a second bank 120, a third bank, and a fourth bank, wherein the first bank 110 is disposed on a first side surface of the second bank 120, the ECC circuit 300 is disposed on a second side surface of the second bank 120 and on a first side surface of the third bank, and the fourth bank is disposed on a second side surface of the third bank. In this case, the first side surface may refer to an opposite side surface of the second side surface, and general elements for implementing a memory device such as an input / output line, a sense amplifier, a column decoder, and a row decoder may be included between corresponding configuration components.
[0079] However, this is merely an example embodiment, and it is apparent that only two memory banks may share one ECC circuit 300 .
[0080] On the other hand, according to an exemplary embodiment of the inventive concept, the syndrome generator 310 may receive the first read data R_DATA 1 included in the first memory bank 110, and generate a syndrome indicating that an error exists in the first read data R_DATA 1. In addition, the parity generator 340 may generate parity for the first write data received from the control logic 200 while the syndrome generator 310 generates the syndrome for the first read data R_DATA1, and may transmit the first write data and the generated parity to the second memory bank 120. In this case, the syndrome decoder 330 may correct the error included in the first read data R_DATA 1 based on a result of decoding the generated syndrome and the first read data R_DATA 1.
[0081] Therefore, reference Figure 2 , when a mask write operation is performed on each of a plurality of memory banks, the memory device 1000 according to an exemplary embodiment of the inventive concept may retain read data for a certain period of time, write new data corresponding to each memory bank together with existing data, and generate parity to improve the speed and accuracy of the memory device 1000. In addition, even if a read operation and a write operation are simultaneously performed on a plurality of memory banks, syndrome generation and parity generation operations may be performed simultaneously.
[0082] Figure 3 FIG. 2 is a diagram showing an exemplary embodiment of the inventive concept when a read command and a write command are applied. Figure 1 A block diagram of a memory device.
[0083] According to an exemplary embodiment of the present invention, after the read data R_DATA of the second memory bank 120 has been sent to the ECC circuit 300 to generate a syndrome and the write data WR_DATA has been received from the control logic 200 to generate the parity of the write data WR_DATA, the memory device 1000 may write the write data WR_DATA and its parity into the first memory bank 110 and correct an error by decoding the generated syndrome of the read data R_DATA.
[0084] The control logic 200 may apply a voltage to the memory cell array 100 to read data of the second memory bank 120. On the other hand, the control logic 200 may transmit write data WR_DATA to be written in the first memory bank 110 to the ECC circuit 300 with a certain time difference thereafter.
[0085] Unlike a general ECC circuit, when the syndrome generator 310 and the parity generator 340 are provided as separate circuits, the ECC circuit 300 can generate syndromes for data read from the second memory bank 120 and can generate parity for data to be written in the first memory bank 110 substantially at the same time.
[0086] The syndrome generator 310 may generate a syndrome indicating an error of the read data R_DATA received from the second memory bank 120, and then transmit the generated syndrome to the latch circuit 320. When the latch circuit 320 receives a signal related to a simple read operation or a simple write operation other than the MWR operation from the control logic 200, the latch circuit 320 may directly transmit the output time of the received data to the syndrome decoder 330 without delay. The syndrome decoder 330 may correct the received data for an error of the received data, and output the corrected received data to the outside of the memory device 1000. In addition, the parity generator 340 may generate parity of the write data WR_DATA received from the data control logic 210. The write data WR_DATA and the generated parity may be written into the first memory bank 110 via the data control logic 210.
[0087] According to another embodiment of the inventive concept, when the time when the syndrome generator 310 generates the syndrome for the received read data R_DATA and the time when the parity generator 340 generates the parity at least partially overlap, the control logic 200 may separately control the syndrome generator 310 and the parity generator 340 in parallel.
[0088] According to an exemplary embodiment of the inventive concept, the control logic 200 may control the syndrome generator 310 and the parity generator 340 in parallel. The control logic 200 may determine whether the times at which the syndrome generator 310 and the parity generator 340 generate the syndrome and the parity, respectively, overlap at least partially based on the internal command that the control logic 200 has issued and sent. In this case, the control logic 200 may send a first control signal for controlling the syndrome generator 310, the latch circuit 320, and the syndrome decoder 330, and a second control signal for controlling the parity generator 340, respectively, to the ECC circuit 300.
[0089] In other words, even if the first memory bank 110 and the second memory bank 120 respectively perform a write operation and a read operation at the same time, it is possible to improve the integration of the memory device 1000 by performing error correction with one ECC circuit 300 and reducing the area occupied by the ECC circuit 300. In addition, since errors occurring during a read or write operation of data of a plurality of memory banks can be corrected at the same time, the processing speed can be increased.
[0090] Figure 4 1 is a data flow diagram showing an exemplary embodiment of the present invention in the case where a read command and a write command are applied. Figure 3 Descriptions are given with reference to the reference numerals in FIG.
[0091] According to an exemplary embodiment of the inventive concept, commands for read operations and write operations of the first memory bank 110 and the second memory bank 120 may be provided from an external device to the memory device 1000. In this case, first read commands READ (BANK 1) and second read commands READ (BANK 2) and first write commands WRITE (BANK 1) and second write commands WRITE (BANK 2) may be applied to the memory device 1000.
[0092] The control logic 200 may apply a voltage to a specific bit line after applying the voltage to a specific word line of the memory cell array 100. When the voltage (e.g., the first read column selection line RCSL (BANK 1)) is applied to the specific bit line, the first memory bank 110 may send the first read data GIO_BANK1 to the ECC circuit 300 to perform ECC calculation. Similarly, when the control logic 200 applies a voltage (e.g., the second read column selection line RCSL (BANK 2)) to a bit line different from the specific bit line, the second memory bank 120 may send the second read data GIO_BANK2 to the ECC circuit 300.
[0093] The syndrome generator 310 may generate a syndrome for the first read data GIO_BANK1 received from the first memory bank 110, perform error correction based on the syndrome, and transmit the generated syndrome to the latch circuit 320. Upon receiving a signal for a simple read operation instead of a signal for mask writing, the latch circuit 320 may output the error-corrected first read data RDIO_BANK1 to the outside of the memory device 1000.
[0094] The above operation may be similarly performed on the second memory bank 120 , and the latch circuit 320 may transmit the error-corrected second read data RDIO_BANK2 received from the syndrome generator 310 to the outside of the memory device 1000 .
[0095] Since the memory device 1000 according to an exemplary embodiment of the inventive concept includes the syndrome generator 310 and the parity generator 340 which are separated from each other, while a read operation is performed in the second memory bank 120 and a syndrome calculation is performed, a parity calculation according to a write operation may be performed in the first memory bank 110. In other words, even when a syndrome calculation of the second memory bank 120 is performed, a parity calculation for data to be written in the first memory bank 110 may be performed substantially at the same time.
[0096] The second read command READ(BANK 2) for the second memory bank 120 and the first write command WRITE(BANK 1) for the first memory bank 110 may be input from an external device to the memory device 1000 with a small time difference or in the same time zone. In this case, the syndrome generator 310, the latch circuit 320, and the syndrome decoder 330 may be used to correct errors of data in the second memory bank 120 and output the error-corrected data to the outside. At the same time, the parity generator 340 may be used to generate parity for the data to be written in the first memory bank 110.
[0097] The control logic 200 may transmit a first write control signal WC(BANK 1) to the parity generator 340 based on a first write command WRITE(BANK 1) for the first memory bank 110. In addition, the control logic 200 may transmit first write data to be written to the first memory bank 110 to the parity generator 340.
[0098] The parity generator 340 may generate parity for the first write data based on the first write control signal WC(BANK 1). Similarly, for the data to be written into the second memory bank 120, the parity for the second write data may be generated based on the second write control signal WC(BANK2). Data (e.g., WDIO_BANK1) including the first write data and its parity and data (e.g., WDIO_BANK2) including the second write data and its parity may be sent to the memory cell array 100 and may be written into the first memory bank 110 and the second memory bank 120, respectively. In this case, data GIO_BANK1 to be written into the first memory bank 110 and data GIO_BANK2 to be written into the second memory bank 120 may be written into the first memory bank 110 and the second memory bank 120, depending on the voltages WCSL(BANK 1) and WCSL(BANK 2) respectively applied to the specific bit lines received from the control logic 200.
[0099] Therefore, since the parity generator 340 can be provided to the memory device 1000 in parallel with the syndrome generator 310, the latch circuit 320 and the syndrome decoder 330 included in the ECC circuit 300, the memory device 1000 can perform error correction on the data to be written to the first memory bank 110 even if error correction is being performed on the data of the second memory bank 120.
[0100] Figure 5 FIG. 2 is a diagram showing a case where a mask write command is applied according to an exemplary embodiment of the inventive concept. Figure 1 A block diagram of a memory device.
[0101] According to an exemplary embodiment of the inventive concept, different mask write operations performed on data included in each of the first memory bank 110 and the second memory bank 120 may be continuously performed at short time intervals. Therefore, the control logic 200 may sequentially apply voltages to the first memory bank 110 and the second memory bank 120 to transmit the data included in the memory cell array 100 to the ECC circuit 300.
[0102] The memory cell array 100 may continuously transmit the first read data included in the first memory bank 110 and the second read data included in the second memory bank 120 to the syndrome generator 310. The syndrome generator 310 may generate a syndrome for each of the first read data and the second read data, and sequentially transmit the generated syndrome to the latch circuit 320.
[0103] The latch circuit 320 may store the first read data and the second read data for which the syndrome has been generated for a certain time. In this case, when receiving the latch control signal, the latch circuit 320 may sequentially send the first read data and the second read data including the generated syndrome to the syndrome decoder 330.
[0104] The syndrome decoder 330 may perform error detection and error correction based on the decoded syndrome and output the first and second read data DIO_BANK 1 and DIO_BANK 2 in which error detection and correction have been completed, and may transmit the first and second read data DIO_BANK 1 and DIO_BANK 2 to the parity generator 340 .
[0105] The parity generator 340 may receive a first write control signal for performing a mask write operation on the first read data and first write data, the first write data being new data required for the mask write operation of the first memory bank 110. In addition, the parity generator 340 may receive a second write control signal for performing a mask write operation on the second read data and second write data, the second write data being new data required for the mask write operation of the second memory bank 120.
[0106] The parity generator 340 may merge the first write data with the first group of data corresponding to the logic high of the data mask signal DM in the first read data DIO_BANK1 based on the first write control signal. Thereafter, a first parity for the merged data may be generated. In addition, the parity generator 340 may merge the second write data with the first group of data corresponding to the logic high of the data mask signal DM in the second read data DIO_BANK2 based on the second write control signal. Thereafter, a second parity for the merged data may be generated. The parity generator 340 may send the data for which the parity has been generated to the data control logic 230, and the data control logic 230 may write the data into the first memory bank 110 and the second memory bank 120.
[0107] Figure 6 1 is a data flow diagram of an exemplary embodiment of the present invention in the case of applying a shield write command. Figure 5 Descriptions are given with reference to the reference numerals in FIG.
[0108] According to an exemplary embodiment of the inventive concept, a command for a mask write operation for the first and second memory banks 110 and 120 may be provided from an external device to the memory device 1000. In other words, the control logic 200 may receive a first MWR signal MWR(BANK1) for the first memory bank 110 from outside the memory device 1000.
[0109] The control logic 200 may perform a read-modify-write operation on the first memory bank 110 based on the first MWR signal MWR(BANK1). Therefore, the control logic 200 may control the memory cell array 100 to send the first read data GIO_BANK1 of the first memory bank 110 to the syndrome generator 310 by applying a voltage (e.g., a first read column select line RCSL(BANK 1) command) to a specific bit line.
[0110] Similarly, the control logic 200 may control the memory cell array 100 to send the second read data GIO_BANK2 of the second memory bank 120 to the syndrome generator 310 by applying a voltage (eg, a second read column select line RCSL (BANK 2) command) to a specific bit line.
[0111] The syndrome generator 310 may generate syndromes for the received first read data GIO_BANK1 and the second read data GIO_BANK 2 , respectively, and transmit the generated syndromes together with the received first read data GIO_BANK1 and the second read data GIO_BANK 2 to the latch circuit 320 .
[0112] When the latch control signal is applied, the latch circuit 320 may retain the first read data GIO_BANK1 and the second read data GIO_BANK2. When the latch control signal LAT_CTRL is applied to the latch circuit 320, the first read data GIO_BANK1 may be retained, and when the first write control signal WR CTRL (BANK 1) is applied, the latch circuit 320 may transmit the retained first read data DIO_BANK1 to the syndrome decoder 330. In addition, when the latch control signal LAT_CTRL is applied to the latch circuit 320, the second read data GIO_BANK2 may be retained, and when the second write control signal WR CTRL (BANK 2) is applied, the latch circuit 320 may transmit the retained second read data DIO_BANK2 to the syndrome decoder 330. At this time, the content of the received data (eg, the first read data GIO_BANK1 and the second read data GIO_BANK2 ) and the content of the transmitted data (eg, the retained first read data DIO_BANK1 and the second read data DIO_BANK2 ) are the same.
[0113] In this case, when a signal for shielding a write operation is applied from the outside, the control logic 200 may transmit the latch control signal LAT_CTRL to the latch circuit 320. Substantially simultaneously, when the first write control signal WR CTRL (BANK 1) is arranged to be transmitted to the parity generator 340 between a time point when a voltage (e.g., a second read column selection line command RCSL (BANK 2)) is applied to the second memory bank 120 and a time point when the second read data GIO_BANK2 is retained, the control logic 200 may transmit the latch control signal LAT_CTRL to the latch circuit 320 before the first write control signal WR CTRL (BANK 1).
[0114] In detail, the control logic 200 may send the latch control signal LAT_CTRL based on the time point at which the first write control signal WR CTRL (BANK 1) is applied. In other words, the control logic 200 may send the latch control signal LAT_CTRL at a time point calculated by subtracting a value equal to M times the clock signal period from the time point at which the first write control signal WR CTRL (BANK 1) is applied. M may be a natural number. In addition, when the memory device 1000 satisfies the formula tCCDMW=4*tCCD, the control logic 200 may send the latch control signal at a time point calculated by subtracting a value equal to 2*tCCD (which is half of 4*tCCD) from the time point at which the first write control signal WR CTRL (BANK 1) is applied.
[0115] This above-described arrangement is used to generate parity by merging the first write data with the first read data, rather than generating parity by merging the first write data with the second read data.
[0116] According to an exemplary embodiment of the inventive concept, when the first write control signal WR CTRL(BANK 1) is not received by the latch circuit 320 while the first read data GIO_BANK1 is retained in the latch circuit 320, the control logic 200 may send the latch control signal LAT_CTRL to the latch circuit 320 and retain the first and second read data GIO_BANK1 and GIO_BANK2.
[0117] The syndrome decoder 330 may receive the first read data DIO_BANK1 and the second read data DIO_BANK 2 , perform error detection and error correction based on the decoded syndrome, and then transmit the results of the error detection and error correction to the parity generator 340 .
[0118] The parity generator 340 may perform a mask write operation based on the first write control signal WR CTRL(BANK 1) to merge the first write data with the first read data R_DATA 1 corresponding to a logic low of the data mask signal DM.
[0119] Similarly, the parity generator 340 may perform a mask write operation based on the second write control signal WR CTRL(BANK 2) to merge the second write data with the second read data R_DATA 2 corresponding to a logic low of the data mask signal DM.
[0120] Figure 7 FIG. 2 is a diagram showing a method including GIO lines corresponding to a plurality of memory banks according to an exemplary embodiment of the inventive concept. Figure 1 A block diagram of a memory device.
[0121] refer to Figure 7 , the memory device 1000 may include a memory cell array 100 and an ECC circuit 300. The memory cell array 100 may include a first memory bank 110, a second memory bank 120, an input / output sense amplifier (IOSA) 130, a GIO line 140, and a multiplexer (MUX) 150. The configurations of the memory cell array 100 and the ECC circuit 300 are substantially the same as those described above and thus are omitted.
[0122] The first memory bank 110 and the second memory bank 120 may transmit the first read data and the second read data to the ECC circuit 300 via the GIO line 140 , respectively.
[0123] The GIO line 140 may be configured to include a plurality of input / output lines. For example, the GIO line 140 may include a first GIO line connected from the first bank 110 to the IOSA 130 and a second GIO line connected from the second bank 120 to the IOSA 130.
[0124] IOSA 130 may receive first read data GIO_BANK1 of first memory bank 110 and second read data GIO_BANK2 of second memory bank 120 from GIO line 140. In this case, MUX 150 included in IOSA 130 may sequentially transmit first and second read data GIO_BANK1 and GIO_BANK2 to ECC circuit 300.
[0125] As an example, the MUX 150 may first transmit the first received data among the first and second read data GIO_BANK1 and GIO_BANK2 to the ECC circuit 300. In this case, the control logic 200 may control a time point for outputting the first and second read data GIO_BANK1 and GIO_BANK2 to match the error correction timing of the ECC circuit 300.
[0126] The ECC circuit 300 may perform ECC calculation on the first read data GIO_BANK1 and the second read data GIO_BANK2 that have been sequentially output from the MUX 150. The detailed operation of the ECC circuit 300 is substantially the same as described above and thus is omitted.
[0127] Figure 8 FIG. 2 is a diagram showing a method of including an input / output (IO) line at each of a plurality of memory banks according to an exemplary embodiment of the inventive concept. Figure 1 A block diagram of a memory device.
[0128] refer to Figure 8, the memory device 1000 may include a memory cell array 100 and an ECC circuit 300. The memory cell array 100 may include a first memory bank 110, a second memory bank 120, a first IOSA 131, a second IOSA 132, and a MUX 151.
[0129] According to an exemplary embodiment of the inventive concept, the first memory bank 110 may output first read data via the first IOSA 131, and the second memory bank 120 may output second read data via the second IOSA 132. The MUX 151 may sequentially transmit the first and second read data to the ECC circuit 300.
[0130] As an example, the control logic 200 may control the MUX 151 to sequentially transmit the first read data and the second read data to the ECC circuit 300. As another example, the MUX 151 may first transmit the first received data among the first read data and the second read data to the ECC circuit 300.
[0131] The ECC circuit 300 may perform ECC calculation on the first read data and the second read data that have been output from the MUX 150. The detailed operation of the ECC circuit 300 is substantially the same as described above and thus is omitted.
[0132] Fig. 9 According to an exemplary embodiment of the present inventive concept Figure 1 A flowchart of a method for operating a memory device.
[0133] The memory device 1000 may receive and retain first read data read from the first memory bank 110 (S410). In this case, the first data read from the first memory bank 110 may be retained based on the latch control signal LAT_CTRL.
[0134] When the latch control signal LAT_CTRL is applied, the memory device 1000 may output the retained first read data (S420). When the first read data has been retained, the first read data may be output based on the latch control signal LAT_CTRL received from the control logic 200, and then the first write control signal is applied after the latch control signal is applied.
[0135] The memory device 1000 may generate parity for the first read data output from the latch circuit 320 based on the first write control signal received from the control logic 200 (S430). Specifically, the first parity may be generated by using the first read data output from the latch circuit 320 and the first write data corresponding to the first MWR command.
[0136] The memory device 1000 may retain the second read data read from the second memory bank 120 based on the latch control signal LAT_CTRL (S440). In other words, the latch circuit 320 may receive and retain the second read data, and when the latch control signal LAT_CTRL is not applied, may retain the second read data without outputting the second read data.
[0137] Fig.10 According to an exemplary embodiment of the present invention Figure 1 A detailed flow chart of an operating method of a memory device.
[0138] refer to Fig.10 , a mask write command for the first memory bank 110 and the second memory bank 120 may be received (S411). In this case, the first mask write command may be a command for performing a mask write operation on data of the first memory bank 110, and the second mask write command may be a command for performing a mask write operation on data of the second memory bank 120.
[0139] A syndrome of the first read data received from the first memory bank 110 may be generated and transmitted to the latch circuit 320 (S412). The latch circuit 320 may retain the data received from the first memory bank 110 until the latch control signal LAT_CTRL is applied.
[0140] The latch control signal LAT_CTRL may be sent after the output time of the first write control signal and the output time of the latch control signal LAT_CTRL are synchronized (S413). Here, synchronization may indicate that the latch control signal LAT_CTRL is output at a time point earlier than the time point at which the first write control signal is scheduled to be output, so that the latch control signal LAT_CTRL is output at a time point corresponding to the first write control signal. The latch control signal LAT_CTRL may include a value obtained by subtracting a time corresponding to M times the clock signal cycle of the memory device 1000 from the output time point of the first write control signal. Alternatively, when the formula tCCDMW=N*tCCD is satisfied, the latch control signal LAT_CTRL may include a value obtained by subtracting a time corresponding to N / 2*tCCD.
[0141] When the received latch control signal LAT_CTRL is applied to the latch circuit 320, the retained first read data may be output (S414). The latch circuit 320 may output the first read data to the syndrome decoder 330 to perform error correction on the data.
[0142] Fig.11 is a block diagram illustrating a data processing system according to an exemplary embodiment of the inventive concept.
[0143] Fig.11 A data processing system 500 including an application processor 510 and a memory device 520 is shown. A memory control module 511 in the application processor 510 and the memory device 520 may constitute a memory system. In addition, the memory control module 511 may include an MWR command generator 511_1, and the memory device 520 may include a memory cell array 521 and a control logic 522. The control logic 522 according to the above exemplary embodiment may correct errors of data read from a plurality of memory banks in response to a mask write command MWR_CMD and a data mask signal DM from the memory control module 511, and may write the error-corrected data into the memory cell array 521.
[0144] The application processor 510 can be implemented as a system on chip (SoC). The SoC can include a system bus that applies a protocol with a specific standard bus specification, and can include various intellectual property rights (IPs) related to the system bus. As an example of the standard specification of the system bus, the Advanced Microcontroller Bus Architecture (AMBA) protocol of Advanced RISC Machine (ARM) Co., Ltd. can be applied. The bus types of the AMBA protocol can include an advanced high-performance bus (AHB), an advanced peripheral bus (APB), an advanced extensible interface (AXI), AXI4, an AXI consistency extension (ACE), etc. In addition, other types of protocols can also be applied, for example, SONICs' uNetwork, IBM's CoreConnect, or the open core protocol international partner (OCP-IP) of the open core protocol.
[0145] The application processor 510 may further include a processor 512 and an operating memory 513 connected via a system bus. The processor 512 may control various operations within the application processor 510, and may control, for example, the memory control module 511 to access the memory device 520. In addition, the processor 512 may control various operations by using software, and the operating memory 513 may store a program for controlling the operations.
[0146] The mask write operation in the above exemplary embodiment may be performed as determined by the memory control module 511. For example, regardless of the control of the processor 512, the memory control module 511 may determine the need to perform the mask write operation and provide the mask write command MWR_CMD to the memory device 520 depending on the result of the determination.
[0147] While the inventive concept has been shown and described above with reference to exemplary embodiments thereof, it will be apparent to those skilled in the art that modifications and changes in form and details may be made therein without departing from the spirit and scope of the inventive concept as set forth in the following claims.
Claims
1. A memory device, include: A memory cell array includes a first memory bank and a second memory bank; a control logic configured to receive a command from the outside and control an internal operation of the memory device in response to the received command; and The error correction code (ECC) circuit is configured as follows: retaining, in a latch circuit, first read data read from a first memory bank in response to a first mask write MWR command to the first memory bank based on a latch control signal from the control logic; Delaying the time for retaining the first read data in the latch circuit based on the latch control signal; generating a first parity from data in which first read data retained in the latch circuit for the time is merged with first write data corresponding to a first MWR command in response to a first write control signal received from the control logic; as well as After the first parity is generated, an ECC operation is controlled based on the latch control signal to retain second read data read from the second memory bank in the latch circuit.
2. The memory device according to claim 1, in, The ECC circuit is configured to generate a second parity by using the second read data in response to a second write control signal received from the control logic, and the latch circuit is configured to retain the first read data and the second read data in response to receiving the latch control signal and output the retained data to generate the parity of the first read data and the second read data.
3. The memory device according to claim 2, in, The ECC circuit is configured to correct an error of the first read data in response to receiving the first write control signal, and merge the error-corrected first read data with the first write data.
4. The memory device according to claim 2, in, The second read data is data read in response to a second MWR command for the second memory bank, The control logic is configured to further receive a data mask DM signal corresponding to a first MWR command and a second MWR command, respectively, The first parity is parity generated by using data in which the first write data is merged with the error-corrected first read data, and The second parity is parity generated by using data in which the second write data is merged with the error-corrected second read data.
5. The memory device according to claim 1, in, The control logic is configured to transmit a latch control signal to the ECC circuit at a time point calculated by subtracting a value equal to M times a clock signal period of the memory device from a time point when the first write control signal is transmitted, wherein M is a natural number.
6. The memory device according to claim 1, in, The first memory bank and the second memory bank are configured to send first read data and second read data to the ECC circuit via a first global input / output GIO line and a second GIO line, respectively.
7. The memory device according to claim 6, in, The memory cell array further includes an input / output sense amplifier, and the input / output sense amplifier is configured to send the first read data and the second read data to the ECC circuit in response to receiving the first read data and the second read data via the first GIO line and the second GIO line, respectively.
8. The memory device of claim 1 , further comprising a multiplexer MUX configured to sequentially transmit first read data received from a first input / output sense amplifier corresponding to a first memory bank and second read data received from a second input / output sense amplifier corresponding to a second memory bank to the ECC circuit.
9. The memory device according to claim 1, in, When the formula tCCDMW=N*tCCD is satisfied, N / 2 storage bodies share the ECC circuit, where tCCDMW is the column access enable CAS to CAS command delay time in a read-modify-write operation, tCCD is the CAS to CAS command delay time, and N is a natural number.
10. The memory device according to claim 9, in, N is 4, and During tCCD, operations of continuously receiving first read data and second read data, operations of correcting errors of the first read data and second read data, operations of generating parity of the first read data and parity of the second read data, and operations of writing the first read data and second read data including the generated parity into the first storage body and the second storage body, respectively, are separately performed.
11. The memory device according to claim 1, in, The memory cell array includes a third memory bank and a fourth memory bank, The first memory bank is arranged on a first side surface of the second memory bank, The ECC circuit is arranged on the second side surface of the second memory bank and on the first side surface of the third memory bank, The fourth memory bank is arranged on the second side surface of the third memory bank, A second side surface of the second memory body is opposite to the first side surface of the second memory body, and The second side surface of the third memory bank is opposite to the first side surface of the third memory bank.
12. The memory device according to claim 1, in, In response to not receiving a first write control signal in the ECC circuit while the first read data is retained in the ECC circuit, the control logic is configured to send a latch control signal to the ECC circuit to retain the first read data and the second read data.
13. A memory device, include: A memory cell array includes a first memory bank and a second memory bank; a control logic configured to receive a write command for the first memory bank and a read command for the second memory bank from the outside, and to control an internal operation of the memory device in response to the received write command and read command; as well as an error correction code (ECC) circuit configured to correct errors included in the data, The ECC circuit includes: a syndrome generator configured to receive first read data read from a first memory bank and generate a syndrome indicating that an error exists in the first read data; a parity generator configured to generate parity of first write data received from the control logic while the syndrome generator generates a syndrome for the first read data, and to send the first write data and the parity of the received first write data to the second memory bank; and The syndrome decoder is configured to correct an error included in the first read data by decoding the generated syndrome.
14. The memory device according to claim 13, in, The control logic is configured to control the syndrome generator and the parity generator in parallel when a time when the syndrome generator generates syndromes and a time when the parity generator generates parity at least partially overlap.
15. The memory device according to claim 13, in, The first memory bank and the second memory bank are configured to send first read data and second read data to the ECC circuit via a first global input / output GIO line and a second GIO line, respectively. 16 . The memory device of claim 15 , further comprising a sense amplifier configured to send the first read data and the second read data to the ECC circuit in response to receiving the first read data and the second read data via the first GIO line and the second GIO line, respectively. 17 . The memory device of claim 13 , further comprising a multiplexer MUX configured to transmit first read data received from a first sense amplifier corresponding to the first memory bank and second read data received from a second sense amplifier corresponding to the second memory bank to the ECC circuit.
18. A method for operating a memory device, the method comprising: include: receiving first read data read from a first memory bank and retaining the first read data in response to a latch control signal received from control logic; Based on the latch control signal, delaying the time for retaining the first read data in the latch circuit; outputting the retained first read data for the time in response to a first write control signal received from the control logic after receiving the latch control signal; generating a first parity from data in which the first read data is merged with the first write data corresponding to the first mask write MWR command in response to a first write control signal; as well as Second read data read from the second memory bank is retained in response to the latch control signal.
19. The operating method according to claim 18, in, The generating includes: generating a second parity by using the second read data in response to a second write control signal received from the control logic, and The retaining includes retaining the first read data and the second read data in response to receiving a latch control signal, and outputting the first read data and the second read data to generate parities of the first read data and the second read data, respectively.
20. The operating method according to claim 19, in, In response to receiving the first write control signal, generating the second parity includes correcting errors of the first read data and merging the error-corrected first read data with the first write data.
