Memory device
By configuring multiple global input and output lines and decoders in the memory device, the simultaneous writing or reading of ordinary data and metadata is achieved, and data loss and metadata are solved in the volatile memory device when the power is interrupted and in the metadata mode is solved. The problem of performance degradation improves the performance and reliability of the memory device.
Patent Information
- Application Number
- CN202410708117.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-09-14
- Filing Date
- 2024-06-03
- Publication Date
- 2024-12-06
AI Technical Summary
The existing volatile memory devices lose data when the power is interrupted, and in metadata mode, the simultaneous writing or reading of normal data and metadata leads to performance deterioration.
A memory device is designed, by configuring multiple global input and output lines in the column direction, and activate multiple word lines simultaneously using row decoder and column decoder to realize simultaneous writing or reading of ordinary data and metadata. , reduce data loading time and increase data rate by allocating different GIO line lengths.
It realizes improving the performance of the memory device without losing data, reducing data access time in metadata mode, and improving the overall performance and reliability of the memory device.
Smart Images

Figure CN119091939A_ABST
Abstract
Description
[0001] This application claims the priority of Korean Patent Application No. 10-2023-0072256 filed on June 5, 2023, in the Korean Intellectual Property Office, and Korean Patent Application No. 10-2023-0122784 filed on September 14, 2023, in the Korean Intellectual Property Office, the disclosures of which are incorporated herein by reference in their entirety. Technical Field
[0002] The present disclosure relates to memory devices. Background Art
[0003] Semiconductor memory devices can be classified into volatile memory devices and nonvolatile memory devices. When the power supply of the volatile memory device is interrupted, the volatile memory device loses its stored data. Dynamic random access memory (DRAM) is a nonvolatile memory device and is used in various fields such as mobile systems, servers, and graphic devices. Summary of the invention
[0004] The present disclosure relates to a volatile memory device for allocating specific data.
[0005] In general, one innovative aspect of the subject matter described in this specification is a memory device including: a memory bank connected to a plurality of word lines; a first global input and output (GIO) line; and a second GIO line formed to have a greater length in a column direction than the first GIO line. One of the first GIO line and the second GIO line can be allocated to metadata.
[0006] Another general aspect is a memory device, the memory device including: a memory bank connected to a plurality of word lines; a first global input and output (GIO) line; a second GIO line formed to have a greater length in a column direction than the first GIO line; and a row decoder connected to the plurality of word lines, extending along the column direction, and configured to select the plurality of word lines. One of the first GIO line and the second GIO line may be assigned to metadata.
[0007] Another general aspect is a memory device, the memory device comprising: a memory bank including a plurality of sub-memory banks divided along a column direction; a plurality of word lines, each word line extending along a row direction and connected to the plurality of sub-memory banks; and a plurality of global input and output (GIO) lines extending along a column direction and connected to the plurality of sub-memory banks. The plurality of GIO lines may include two GIO lines connected to different sub-memory banks, and one of the two GIO lines may be allocated to metadata. BRIEF DESCRIPTION OF THE DRAWINGS
[0008] The above and other aspects, features and advantages of the present disclosure will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings.
[0009] Figure 1 is a diagram illustrating an example of a memory system.
[0010] Figure 2 is a plan view of an example of a memory device.
[0011] Figure 3 and Figure 4 is a diagram illustrating an example of the operation of the first memory bank.
[0012] Figure 5 and Figure 6 is a diagram illustrating an example of the operation of the second memory bank.
[0013] Figure 7 is a diagram illustrating an example of a memory bank.
[0014] Figure 8 It is shown Figure 7 FIG. 1 is a diagram of an example of area “A”.
[0015] Fig. 9 is a diagram illustrating an example of the operation of a storage body when the metadata mode is turned off.
[0016] Fig.10 An example of the operation of a storage body when metadata mode is turned on is shown.
[0017] Fig.11 and Fig.12 is a diagram illustrating an example of a word line allocation method.
[0018] Fig.13 and Fig.14 is a diagram showing an example of column selection signals for normal data and metadata.
[0019] Fig.15 and Fig.16 is a diagram showing an example of column selection signals for normal data and metadata. DETAILED DESCRIPTION
[0020] Figure 1 is a diagram illustrating an example of a memory system.
[0021] Reference Figure 1 , the memory system 100 includes a memory controller 110 and a memory device 120 .
[0022] The memory controller 110 may control the overall operation of the memory system 100, and may control data exchange between the memory device 120 and a host device connected to the memory system 100. For example, the memory controller 110 may control the memory device 120 to write or read data based on a request from the host device.
[0023] The memory controller 110 may communicate with the memory device 120 through a memory interface, or communicate with the host device through a host interface. Therefore, the memory controller 110 may mediate signals between the host device and the memory device 120. The memory controller 110 may control the memory device 120 based on a command CMD for controlling the memory device 120. For example, the memory device 120 may include a dynamic random access memory (DRAM), a double data rate 4 (DDR4) synchronous DRAM (SDRAM), a DDR5 SDRAM, a DDR6 SDRAM, a low power DDR4 (LPDDR4) SDRAM, a LPDDR5 SDRAM, a LPDDR6 SDRAM, and the like. However, the present disclosure is not limited thereto, and the memory device 120 may include a non-volatile memory device. However, in some embodiments, an example will be provided where the memory device 120 is a volatile memory.
[0024] The memory controller 110 may send the above-mentioned command CMD, clock signal CLK, and address ADDR to the memory device 120. The memory controller 110 may provide data DQ to the memory device 120, and may receive data DQ from the memory device 120. The memory device 120 may include a memory cell array (MCA) 130 storing the data DQ, a control logic circuit 140, a data input and / or output (I / O) buffer 150, and the like.
[0025] The memory cell array 130 may include a plurality of memory cells. Each memory cell may store data (e.g., 1 bit). The memory cell array 130 may include a plurality of word lines WL1 and WL2 and a plurality of global input and output (GIO) lines, and a plurality of memory cells may be connected to each word line and each GIO line. For example, the word lines and the GIO lines may constitute a single group configured to output a data prefetch size required by the memory device 120.
[0026] The plurality of GIO lines GIO1 and GIO2 may be configured to transmit data written to a plurality of word lines WL1 and WL2 and a selected word line and CSL among a plurality of column select lines (CSL), or transmit data read from a plurality of word lines WL1 and WL2 and a selected word line and CSL among a plurality of CSLs. The memory device may be driven at high speed, and data loading may be reduced by the plurality of GIO lines GIO1 and GIO2. In this case, the memory cells connected to the word lines may be referred to as rows, and the memory cells selected by the plurality of CSLs may be referred to as columns.
[0027] In some embodiments, the plurality of GIO lines GIO1 and GIO2 may include a first GIO line GIO1 and a second GIO line GIO2. The first GIO line GIO1 and the second GIO line GIO2 may have different lengths or pitches. For example, the second GIO line GIO2 may be configured to be longer than the first GIO line GIO1 in a column direction. Since the first GIO line GIO1 is shorter than the second GIO line GIO2, the first GIO line GIO1 may be referred to as a short GIO line, and the second GIO line GIO2 may be referred to as a long GIO line.
[0028] The first GIO line GIO1 and the second GIO line GIO2 may be formed of different metals. Since the second GIO line GIO2 is arranged to be longer than the first GIO line GIO1, the speed of transmitting data may be relatively reduced as the load increases. Therefore, the second GIO line GIO2 may be formed of a metal capable of achieving a higher data rate than the first GIO line GIO1.
[0029] In some embodiments, some of the GIO lines may be allocated to metadata. In some embodiments, the metadata may represent data for improving the performance of the memory device 120 or enhancing security. For example, one of the first GIO line GIO1 and the second GIO line GIO2 may be allocated to metadata. The other of the first GIO line GIO1 and the second GIO line GIO2 may be allocated to normal data corresponding to the metadata. Normal data may be defined as data other than metadata.
[0030] In some embodiments, a ratio of normal data to metadata accessed by the memory device 120 may be required. For example, the required ratio of normal data and metadata accessed by the memory device 120 (hereinafter, the required normal / meta ratio) may be set to n:1 (where n is a positive integer). For example, n may be 8 or 16, but the present disclosure is not limited thereto. For example, in the case of n=8, when accessing 64 bytes of normal data, 8 bytes of metadata may need to be accessed. Therefore, the ratio of metadata input or output from one of the first GIO line GIO1 and the second GIO line GIO2 to normal data input or output from the other of the first GIO line GIO1 and the second GIO line GIO2 may be 1:n.
[0031] The control logic circuit 140 may receive a command CMD and an address ADDR from the memory controller 110. The control logic circuit 140 may control the memory cell array 130 based on the command CMD and the address ADDR. For example, the control logic circuit 140 may decode the command CMD and generate a control signal corresponding to the decoded command CMD. In addition, the control logic circuit 140 may generate an internal command (CMD) signal (such as a refresh signal, an activation signal, a precharge signal, a read signal, or a write signal). In addition, the control logic circuit 140 may decode the address ADDR and select or activate a word line and CSL corresponding to the decoded address ADDR.
[0032] In some embodiments, the control logic circuit 140 may control the memory device to simultaneously activate a first word line WL1 corresponding to the first GIO line GIO1 and a second word line WL2 corresponding to the second GIO line GIO2 among a plurality of word lines. Based on the first word line WL1 and the second word line WL2 being simultaneously activated under the control of the control logic circuit 140, the first GIO line GIO1 and the second GIO line GIO2 corresponding to the word lines may simultaneously input or output normal data and metadata. For example, metadata (or normal data) written or read in response to the activated first word line WL1 may be input or output from the first GIO line GIO1, and normal data (or metadata) written or read in response to the activated second word line WL2 may be simultaneously input or output from the second GIO line GIO2.
[0033] The data input and / or output buffer 150 may provide normal data and / or metadata to the memory cell array 130 based on the clock signal CLK when a write operation is performed, or may provide normal data and / or metadata provided from the memory cell array 130 to the memory controller 110 based on the clock signal CLK when a read operation is performed.
[0034] For example, the data input and / or output buffer 150 may provide normal data and metadata simultaneously input to the memory cell array 130 to the memory cell array 130 , or provide normal data and metadata output from the memory cell array 130 to the memory controller 110 .
[0035] In this example, an operation mode of a memory system that simultaneously writes or reads normal data and metadata may be defined as a metadata mode. For example, the metadata mode may be an operation mode for inputting or outputting metadata. In the metadata mode, the memory controller 110 and the control logic circuit 140 may control the memory device 120 to simultaneously write or read normal data and metadata through a first GIO line GIO1 assigned to normal data (or metadata) and a second GIO line GIO2 assigned to metadata (or normal data).
[0036] In this example, for example, a portion of the GIO lines connected to the memory cell array 130 included in the memory device 120 may be allocated to metadata. When the word lines and the GIO lines constitute a single group, writing or reading normal data and metadata may require two accesses, resulting in performance degradation. In some embodiments, a portion of the first GIO line GIO1 and the second GIO line GIO2 may be allocated to metadata, and the number of activated word lines may be doubled to allow normal data and metadata to be written or read simultaneously.
[0037] Figure 2 is a plan view of an example of a memory device.
[0038] Reference Figure 2 , the memory device 120 includes a memory body unit BU and a peripheral circuit region PERI.
[0039] The memory cell unit BU (also referred to as a "memory cell group") may be set to a plurality according to the capacity of the memory device 120. A portion and a remaining portion of the plurality of memory cell units BU may be arranged along the row direction d2, and the peripheral circuit region PERI may be inserted therebetween. Each of the memory cell units BU may include a plurality of memory cells. Each of the plurality of memory cells may include a plurality of mats including a plurality of memory cells.
[0040] In some embodiments, each of the plurality of memory banks may include a plurality of sub-memory banks. The plurality of sub-memory banks included in each memory bank may correspond to the same memory bank address. The plurality of sub-memory banks provided for each memory bank may be i sub-memory banks (where i is a positive integer). The plurality of sub-memory banks may be divided and arranged along the column direction d1 in a single memory bank. Each of the plurality of sub-memory banks may be connected to a plurality of word lines.
[0041] For example, the first memory bank B1 may include sub-memory banks SB1-1 to SB1-i, and the sub-memory banks SB1-1 to SB1-i may be arranged along the column direction d1. In addition, the second memory bank B2 arranged along the row direction d2 from the first memory bank B1 may include sub-memory banks SB2-1 to SB2-i, and the sub-memory banks SB2-1 to SB2-i may be arranged along the column direction d1.
[0042] The row decoders RD1 and RD2 may extend along the column direction d1 and may be provided in a number corresponding to the number of banks included in a single bank unit BU. In some embodiments, the row decoders RD1 and RD2 may select word lines connected to different banks B1 and B2 based on decoding a row address.
[0043] In some embodiments, each of the row decoders RD and RD2 may simultaneously select at least two word lines connected to different sub-memory banks among a plurality of word lines connected to the sub-memory banks included in the memory banks B1 and B2 to read or write the normal data ND and the metadata MD. For example, the first row decoder RD1 may simultaneously select at least two word lines connected to at least two sub-memory banks among 1 to i sub-memory banks SB1-1 to SB1-i, and the second row decoder RD2 may simultaneously select at least two word lines connected to at least two sub-memory banks among the sub-memory banks SB2-1 to SB2-i.
[0044] The column decoders CD1 and CD2 may extend along the row direction d2 and may be provided in a number corresponding to the number of memory banks included in a single memory bank unit BU. In some embodiments, the column decoders CD1 and CD2 may select a sub-memory bank included in the same memory bank and a CSL connected to the first GIO line and the second GIO line based on decoding a column address. For example, the first column decoder CD1 may select a CSL connected to the sub-memory banks SB1-1 to SB1-i, and the second column decoder CD2 may select a CSL connected to the sub-memory banks SB2-1 to SB2-i.
[0045] Figure 3 and Figure 4 is a diagram illustrating an example of the operation of the first memory bank.
[0046] Reference Figure 3 and Figure 4 The memory unit BU includes a first memory bank B1 connected to a first row decoder RD1 and a first column decoder CD1, and a second memory bank B2 connected to a second row decoder RD2 and a second column decoder CD2. The first memory bank B1 includes a sub-memory bank SB1-1 and a sub-memory bank SB1-2, and the second memory bank B2 includes a sub-memory bank SB2-1 and a sub-memory bank SB2-2.
[0047] The first GIO line GIO1 and the second GIO line GIO2 may be arranged in the first memory bank B1. For example, the first GIO line GIO1 and the second GIO line GIO2 may be shared and arranged in the first memory bank B1. In this case, the first GIO line GIO1 may be connected to the sub-memory bank SB1-1, and the second GIO line GIO2 longer than the first GIO line GIO1 may be connected to the sub-memory bank SB1-2. For example, the first GIO line GIO1 having a relatively small length may be connected to the sub-memory bank SB1-1 closer to the first column decoder CD1, and the second GIO line GIO2 having a relatively large length may be connected to the sub-memory bank SB1-2 away from the first column decoder CD1. The first word line WL1 may be connected to the sub-memory bank SB1-1, and the second word line WL2 may be connected to the sub-memory bank SB1-2.
[0048] In some embodiments, Figure 3 As shown in , the first GIO line GIO1 connected to the sub-bank SB1-1 can be used for normal data ND, and the second GIO line GIO2 connected to the sub-bank SB1-2 can be used for metadata MD. In this case, the first GIO line GIO1 for normal data ND of the sub-bank SB1-1 can correspond to the second GIO line for metadata MD of the sub-bank SB1-2. A portion of the memory cells included in the sub-bank SB1-2 connected to the second GIO line GIO2 can be used to store metadata MD corresponding to “normal data ND stored in a portion of the memory cells included in the sub-bank SB1-1”.
[0049] In some embodiments, normal data ND and metadata MD corresponding to the normal data ND may be simultaneously written to or read from the first memory bank B1. When a read operation is performed on the normal data ND and metadata MD written in the first memory bank B1, the normal data ND and metadata MD may be read based on enabling two word lines simultaneously for each sub-memory bank. For example, the first row decoder RD1 may simultaneously select the first word line WL1 and the second word line WL2 to simultaneously read or write the normal data and metadata. In addition, the first column decoder CD1 may select CSL.
[0050] Then, normal data ND and metadata MD may be read from the memory cell corresponding to the selected first word line WL1, the selected second word line WL2, and the selected CSL, and may be sent to the first GIO line GIO1 and the second GIO line GIO2. For example, normal data ND may be read from the sub-storage bank SB1-1 connected to the first word line WL1 and the first GIO line GIO1, and metadata may be read from the sub-storage bank SB1-2 connected to the second word line WL2 and the second GIO line GIO2. In this case, the first GIO line GIO1 assigned to normal data ND may output normal data ND, and the second GIO line GIO2 assigned to metadata MD may output metadata MD.
[0051] Optionally, in some embodiments, write operations of normal data ND and metadata MD corresponding to the normal data ND may also be performed simultaneously. When write operations of normal data ND and metadata MD are performed on the first memory bank B1, two word lines may be activated simultaneously for each sub-memory bank, and normal data ND and metadata MD may be written simultaneously through the two word lines. For example, the first row decoder RD1 may select the first word line WL1 and the second word line WL2 simultaneously, and the first column decoder CD1 may select CSL.
[0052] In response to the selected first word line WL1, the selected second word line WL2, and the selected CSL, normal data ND may be transmitted to a portion of the memory cells included in the sub-memory bank SB1-1 through the first GIO line GIO1, and metadata MD may be written to a portion of the memory cells included in the sub-memory bank SB1-2 through the second GIO line GIO2. In this case, the first GIO line GIO1 allocated to the normal data ND may input the normal data ND, and the second GIO line GIO2 allocated to the metadata MD may input the metadata MD.
[0053] Alternatively, if Figure 4 As shown in , the first GIO line GIO1 connected to the sub-storage bank SB1-1 can be used for metadata MD, and the second GIO line GIO2 connected to the sub-storage bank SB1-2, which is longer than the first GIO line GIO1, can be used for normal data ND. In this case, the first GIO line GIO1 used for metadata MD of the sub-storage bank SB1-1 can correspond to the second GIO line GIO2 used for normal data ND of the sub-storage bank SB1-2. A portion of the memory cells included in the sub-storage bank SB1-1 connected to the first GIO line GIO1 can be used to store metadata MD corresponding to “normal data ND stored in a portion of the memory cells included in the sub-storage bank SB1-2 connected to the second GIO line GIO2”.
[0054] In some embodiments, when normal data ND and metadata MD corresponding to the normal data ND are simultaneously written to or read from the first memory bank B1, the normal data ND and metadata MD may be read from memory cells corresponding to the first word line WL1, the second word line WL2, and the CSL selected simultaneously through the first row decoder RD1 and the first column decoder CD1, and then may be sent to the first GIO line GIO1 and the second GIO line GIO2. For example, the metadata MD may be read from the sub-memory bank SB1-1 connected to the first word line WL1 and the first GIO line GIO1, and the normal data ND may be read from the sub-memory bank SB1-2 connected to the second word line WL2 and the second GIO line GIO2. In this case, the second GIO line GIO2 assigned to the normal data ND may output the normal data ND, and the first GIO line GIO1 assigned to the metadata MD may output the metadata MD.
[0055] Optionally, in some embodiments, write operations of normal data ND and metadata MD corresponding to the normal data ND may also be performed simultaneously. When write operations of normal data ND and metadata MD are performed on the first memory bank B1, two word lines may be activated simultaneously for each sub-memory bank, and normal data ND and metadata MD may be written simultaneously through the two word lines. In response to the first word line WL1, the second word line WL2, and CSL selected by the first row decoder RD1 and the first column decoder CD1, metadata MD may be written into a portion of the memory cells included in the sub-memory bank SB1-1 through the first GIO line GIO1, and normal data ND may be written into a portion of the memory cells included in the sub-memory bank SB1-2 through the second GIO line GIO2. In this case, the first GIO line GIO1 assigned to metadata MD may input metadata MD, and the second GIO line GIO2 assigned to normal data ND may input normal data ND.
[0056] Figure 5 and Figure 6 is a diagram showing an example of the operation of the second storage body. Figure 3 and Figure 4 A detailed description of those parts that overlap with those described in .
[0057] Reference Figure 5 and Figure 6, similar to the first memory bank B1, the first GIO line GIO1 and the second GIO line GIO2 may also be arranged in the second memory bank B2. In this case, the first GIO line GIO1 having a relatively small length may be connected to the sub-memory bank SB2-1, and the second GIO line GIO2 having a relatively large length may be connected to the sub-memory bank SB2-2. In addition, the first word line WL1 may be connected to the sub-memory bank SB2-1, and the second word line WL2 may be connected to the sub-memory bank SB2-2.
[0058] In some embodiments, Figure 5 As shown in, the first GIO line GIO1 connected to the sub-storage bank SB2-1 can be used for normal data ND, and the second GIO line GIO2 connected to the sub-storage bank SB2-2 can be used for metadata MD. In this case, the first GIO line GIO1 for normal data ND of the sub-storage bank SB2-1 can correspond to the second GIO line GIO2 for metadata MD of the sub-storage bank SB2-2. A portion of the memory cells included in the sub-storage bank SB2-2 connected to the second GIO line GIO2 can be used to store metadata MD corresponding to “normal data ND stored in a portion of the memory cells included in the sub-storage bank SB2-1 connected to the first GIO line GIO1”.
[0059] In some embodiments, normal data ND and metadata MD corresponding to the normal data ND may be simultaneously written to or read from the second memory bank B2. For example, based on enabling two word lines simultaneously for each sub-memory bank, normal data ND and metadata MD may be simultaneously written or read through the two word lines. For example, the second row decoder RD2 may simultaneously select the first word line WL1 and the second word line WL2 to read or write the normal data ND and metadata MD, and the second column decoder CD2 may select CSL.
[0060] Then, normal data ND read from the memory cells corresponding to the selected first word line WL1, the selected second word line WL2, and the selected CSL may be output through the first GIO line GIO1, and the read metadata MD may be output through the second GIO line GIO2. Alternatively, normal data ND input through the first GIO line GIO1 may be written into the memory cells corresponding to the selected first word line WL1, the selected second word line WL2, and the selected CSL, and metadata MD input through the second GIO line GIO2 may be written into the memory cells.
[0061] Alternatively, if Figure 6As shown in, a first GIO line GIO1 connected to the sub-storage bank SB2-1 and having a relatively small length can be used for metadata MD, and a second GIO line GIO2 connected to the sub-storage bank SB2-2 and having a relatively large length can be used for normal data ND. In this case, the first GIO line GIO1 for metadata MD of the sub-storage bank SB2-1 can correspond to the second GIO line GIO2 for normal data ND of the sub-storage bank SB2-2. A portion of the memory cells included in the sub-storage bank SB2-1 connected to the first GIO line GIO1 can be used to store metadata MD corresponding to “normal data ND stored in a portion of the memory cells included in the sub-storage bank SB2-2 connected to the second GIO line GIO2”.
[0062] In some embodiments, when normal data ND and metadata MD corresponding to the normal data ND are simultaneously written to or read from the second memory bank B2, metadata MD read from “memory cells corresponding to the first word line WL1, the second word line WL2, and the CSL simultaneously selected by the second row decoder RD2 and the second column decoder CD2” may be output through the first GIO line GIO1, and the read normal data ND may be output through the second GIO line GIO2. Alternatively, metadata MD input through the first GIO line GIO1 may be written into memory cells corresponding to the selected first word line WL1, the selected second word line WL2, and the selected CSL, and normal data ND input through the second GIO line GIO2 may be written into the memory cells.
[0063] In this example, the memory bank may include a single sub-memory bank connected to a first GIO line GIO1 having a relatively small length and a single sub-memory bank connected to a second GIO line GIO2 having a relatively large length. Each GIO line may be connected to each sub-memory bank, but may be shared and arranged according to the memory bank. Each GIO line may be assigned to any one of normal data ND and metadata MD, and the word lines connected to each sub-memory bank may be activated at the same time. Therefore, the memory bank is able to write or read normal data ND and metadata MD at the same time through a single access, and a write or read operation may be performed without time loss or limitation.
[0064] Figure 7 is a diagram illustrating an example of a memory bank.
[0065] Reference Figure 7 , memory bank B can be connected with the above Figures 2 to 6The memory bank B may be connected to a first word line WL1 and a second word line WL2 extending along a row direction d2, a plurality of first GIO lines L_GIO1-1 to U_GIO1-k extending along a column direction d1, a plurality of second GIO lines L_GIO2-1 to U_GIO2-k, a row decoder RD, and an input / output sense amplifier stage IOSAL. A plurality of word lines including the first word line WL1 and the second word line WL2 may be connected to the memory bank B, but in Figure 7 , only the first word line WL1 and the second word line WL2 selected simultaneously by the row decoder RD are shown. In addition, for the sake of simplicity of the drawings, the first GIO line and the second GIO line corresponding to each column are each shown as one, but a plurality of GIO lines may be set in response to the size of the burst length that may be the input and output of each column. For example, N first GIO lines and N second GIO lines (where N is a positive integer) may be set for each. For example, 2N GIO lines may be set for a single memory bank B.
[0066] The memory bank B may include a first sub-memory bank SB1 and a second sub-memory bank SB2. The first sub-memory bank SB1 and the second sub-memory bank SB2 may be divided by a memory bank boundary BB. In some embodiments, the memory bank boundary BB may include a plurality of dummy memory cells.
[0067] A plurality of first GIO lines L_GIO1-1 to U_GIO1-k arranged in the memory bank B may be connected to the first sub-memory bank SB1, and a plurality of second GIO lines L_GIO2-1 to U_GIO2-k may be connected to the second sub-memory bank SB2. In this case, the plurality of second GIO lines L_GIO2-1 to U_GIO2-k may be commonly arranged in the first sub-memory bank SB1 and the second sub-memory bank SB2, but may be connected to the second sub-memory bank SB2 to input data written to the pad MAT included in the second sub-memory bank SB2 or output data read from the pad MAT included in the second sub-memory bank SB2.
[0068] Each sub-storage bank may include a plurality of MATs. Some word lines and some GIOs may be connected to each MAT. The MAT may include a two-dimensionally arranged memory cell array, and each memory cell may be connected to one of the word lines connected to the MAT and one of the GIOs connected to the MAT. For example, the MAT included in the first sub-storage bank SB1 and the memory cells included in the MAT are connected to the first GIO line, and the MAT included in the second sub-storage bank SB2 and the memory cells included in the MAT may be connected to the second GIO line.
[0069] In some embodiments, a plurality of GIO lines may be classified into one of a first side, a second side, and a third side. Each side may correspond to at least one column. In this case, the MAT included in the second side may be used for an error correction code (ECC). For example, the MAT included in the second side may store parity bits for ECC. Therefore, among the plurality of GIO lines, the first GIO line M_GIO1 and the second GIO line M_GIO2 of the second side may input or output parity bits.
[0070] The first GIO lines L_GIO1-1 to L_GIO1-k of the first side L may be allocated to normal data ND (or metadata MD), and the second GIO lines L_GIO2-1 to L_GIO2-k of the first side may be allocated to metadata MD (or normal data ND). Similarly, the first GIO lines U_GIO1-1 to U_GIO1-k of the third side U may be allocated to normal data ND (or metadata MD), and the second GIO lines U_GIO2-1 to U_GIO2-k of the third side may be allocated to metadata MD (or normal data ND). In this case, when the first GIO lines L_GIO1-1 to L_GIO1-k of the first side and the first GIO lines U_GIO1-1 to U_GIO1-k of the third side may be allocated to normal data ND, the second GIO lines L_GIO2-1 to L_GIO2-k of the first side and the second GIO lines U_GIO2-1 to U_GIO2-k of the third side may be allocated to metadata MD. When the first GIO lines L_GIO1-1 to L_GIO1-k and U_GIO1-1 to U_GIO1-k are allocated to metadata MD, the first side second GIO lines L_GIO2-1 to L_GIO2-k and the third side second GIO lines U_GIO2-1 to U_GIO2-k may be allocated to normal data ND.
[0071] The first and second GIO lines L_GIO1-1 to L_GIO1-k and L_GIO2-1 to L_GIO2-k are each provided with k lines (where k is a positive integer). Similarly, the first and second GIO lines U_GIO1-1 to U_GIO1-k and U_GIO2-1 to U_GIO2-k of the third side may each be provided with k lines.
[0072] In some embodiments, when the first word line WL1 and the second word line WL2 are simultaneously activated by the row decoder RD, data may be written to or read from a plurality of MATs corresponding to the first word line WL1 and a plurality of MATs corresponding to the second word line WL2. For example, the first word line WL1 and the second word line WL2 are simultaneously activated so that a single column may be selected for each of the sub-memory bodies SB1 and SB2 (in addition, the CSL may be selected by a column decoder (not shown)). By selecting the first word line WL1 and the CSL, a portion of the memory cells connected to the column corresponding to the first word line WL1 may be activated. In addition, by selecting the second word line WL2 and the CSL, a portion of the memory cells connected to the column corresponding to the second word line WL2 may be activated.
[0073] Then, normal data ND or metadata MD may be input or output from the first GIO line connected to the column corresponding to the first word line WL1, and metadata MD or normal data ND may be input or output from the second GIO line connected to the column corresponding to the second word line WL2. In this case, parity data may be input or output from the first GIO line M_GIO1 and the second GIO line M_GIO2 on the second side.
[0074] The input / output sense amplifier stage IOSAL may sense and amplify voltages of the plurality of GIO lines based on data output from the plurality of GIO lines. Alternatively, the input / output sense amplifier stage IOSAL may transmit data to the memory bank B through the plurality of GIO lines.
[0075] Figure 8 It is shown Figure 7 FIG. 1 is a diagram of an example of area “A”.
[0076] Reference Figure 8 , in some embodiments, the region 'A' may correspond to a single column. The single column may include a plurality of first mats MAT1 included in the first sub-memory bank SB1 and a plurality of second mats MAT2 included in the second sub-memory bank SB2. The plurality of first mats MAT1 and the plurality of second mats MAT2 may be separated by a memory bank boundary BB. As described above, the plurality of first mats MAT1 and the plurality of second mats MAT2 may be divided into one of the first side to the third side.
[0077] Each of the plurality of first pads MAT1 and the plurality of second pads MAT2 may be connected to a plurality of CSLs.
[0078] Writing data to each of the plurality of first pads MAT1 or reading data from each of the plurality of first pads MAT1 may be performed through a plurality of CSLs, a plurality of first local input / output (LIO) lines LIO1<1:a> connected to the plurality of CSLs, and a plurality of first GIO lines GIO1<1:a> connected to the plurality of first LIO lines LIO1<1:a> (where “a” is a positive integer greater than or equal to 8). Similarly, writing data to each of the plurality of second pads MAT2 or reading data from each of the plurality of second pads MAT2 may be performed through a plurality of CSLs, a plurality of second LIO lines (LIO2<1:a>) connected to the plurality of CSLs, and a plurality of second GIO lines GIO2<1:a> connected to the plurality of second LIO lines. In some embodiments, a plurality of first GIO lines GIO1<1:a> connected to a plurality of first pads MAT1 and having a relatively small length may be allocated to normal data ND (or metadata MD), and a plurality of second GIO lines GIO2<1:a> connected to a plurality of second pads MAT2 and having a relatively large length may be allocated to metadata MD (or normal data ND).
[0079] Based on a write or read command, the first word line WL1 and the second word line WL2 may be selected at the same time, and a bit lines connected to a plurality of CSLs may be selected from each of the first pad MAT1 and the second pad MAT2. For example, when a write operation is performed, normal data ND (or metadata MD) to be written through a plurality of first GIO lines GIO1<1:a> may be input to the selected a bit lines via a plurality of first LIO lines LIO1<1:a> and a plurality of CSLs. Therefore, a plurality of first pads MAT1 may store normal data ND (or metadata MD). In addition, metadata MD (or normal data ND) to be written through a plurality of second GIO lines GIO2<1:a> may be input to the selected a bit lines via a plurality of second LIO lines LIO2<1:a> and a plurality of CSLs. Therefore, a plurality of second pads MAT2 may store metadata MD (or normal data ND).
[0080] For example, when a read operation is performed, normal data ND (or metadata MD) read from a bit lines selected for a plurality of first pads MAT1 may be transmitted to the first input / output sense amplifier IOSA1 via a plurality of CSLs, a plurality of first LIO lines LIO1<1:a>, and a plurality of first GIO lines GIO1<1:a>. In addition, metadata MD (or normal data ND) read from a bit lines selected for a plurality of second pads MAT2 may be transmitted to the second input / output sense amplifier IOSA2 via a plurality of CSLs, a plurality of second LIO lines LIO2<1:a>, and a plurality of second GIO lines GIO2<1:a>.
[0081] The first input / output sense amplifier IOSA1 and the first driver DRV1 may be connected to the plurality of first GIO lines GIO1<1:a>, and the second input / output sense amplifier IOSA2 and the second driver DRV2 may be connected to the plurality of second GIO lines GIO2<1:a>. The first input / output sense amplifier IOSA1, the first driver DRV1, the second input / output sense amplifier IOSA2 and the second driver DRV2 may be included in Figure 7 of the input / output sense amplifier stage IOSAL.
[0082] The first input / output sense amplifier IOSA1 may sense and amplify the voltage of the plurality of first GIO lines GIO1<1:a> based on data output from the plurality of first GIO lines GIO1<1:a>, and the second input / output sense amplifier IOSA2 may sense and amplify the voltage of the plurality of second GIO lines GIO2<1:a>. For example, normal data ND may be assigned to the plurality of first GIO lines GIO1<1:a>, and metadata MD may be assigned to the plurality of second GIO lines GIO2<1:a>. In this case, the first input / output sense amplifier IOSA1 may sense and amplify the voltage based on the normal data ND, and the second input / output sense amplifier IOSA2 may sense and amplify the voltage based on the metadata MD.
[0083] The first driver DRV1 may transmit data to the memory bank through the plurality of first GIO lines GIO1<1:a>, and the second driver DRV2 may transmit data to the memory bank through the plurality of second GIO lines GIO2<1:a>. For example, when normal data ND is allocated to the plurality of first GIO lines GIO1<1:a>, and metadata MD is allocated to the plurality of second GIO lines GIO2<1:a>, the first driver DRV1 may transmit normal data ND to the plurality of first pads MAT1 through the plurality of first GIO lines GIO1<1:a>, and the second driver DRV2 may transmit metadata MD to the plurality of second pads MAT2 through the plurality of second GIO lines GIO2<1:a>.
[0084] In some embodiments, the number of each of the input / output sense amplifiers, drivers, and GIO lines is equal. Optionally, when the plurality of first GIO lines GIO1<1:a> are allocated to normal data ND, and the plurality of second GIO lines GIO2<1:a> are allocated to metadata MD, the number of the set first input / output sense amplifiers IOSA1 and the number of the first drivers DRV1 may be a ratio of n:1. Optionally, when the plurality of first GIO lines GIO1<1:a> are allocated to metadata MD, and the plurality of second GIO lines GIO2<1:a> are allocated to normal data ND, the number of the set first input / output readout amplifiers IOSA1 and the number of the first drivers DRV1 may be a ratio of 1:n.
[0085] Fig. 9 is a diagram illustrating an example of operation of a storage body when the metadata mode is turned off.
[0086] Reference Fig. 9 An example of , in metadata mode, includes Figure 2 Each memory bank in the memory bank unit BU can simultaneously write or read normal data ND and metadata MD through a GIO line allocated to normal data ND and a GIO line allocated to metadata MD.
[0087] In some embodiments, when the metadata mode is off, the memory bank may perform a write or read operation of normal data ND. In this case, both the first GIO line GIO1 and the second GIO line GIO2 may be allocated to the normal data ND.
[0088] For example, the row decoder may select at least one word line from among the first word line WL1 and the second word line WL2. When normal data ND is read from the at least one selected word line, at least one of the first GIO line GIO1 and the second GIO line GIO2 may read the normal data ND. Alternatively, when normal data ND is input from at least one of the first GIO line GIO1 and the second GIO line GIO2, the normal data ND may be written into a memory cell included in at least one of the first sub-storage bank SB1 and the second sub-storage bank SB2.
[0089] Fig.10 An example of the operation of a storage body when metadata mode is turned on is shown.
[0090] Reference Fig.10 In the example of FIG. 1 , when the metadata mode is turned on, each of the first GIO line GIO1 and the second GIO line GIO2 may be allocated to normal data ND (or metadata MD). Fig.10 A case is shown where the first GIO line GIO1 is allocated to normal data ND, and the second GIO line GIO2 is allocated to metadata MD.
[0091] Then, the first sub-storage body SB1 may store normal data ND, and the second sub-storage body SB2 may store metadata MD. When the first word line WL1 is activated, the first sub-storage body SB1 may write or read normal data ND, and may perform a write or read operation in units of a first page size PS1 corresponding to the first word line WL1. Similarly, the second sub-storage body SB2 may perform a write or read operation in units of a second page size PS2 corresponding to the second word line WL2.
[0092] In some embodiments, the first page size PS1 and the second page size PS2 may have the same or different sizes. For example, each page size may be 1KB or 2KB, but the present disclosure is not limited thereto.
[0093] In some embodiments, the second page size PS2 of the second sub-storage bank SB2 for writing or reading metadata MD may be smaller than the first page size PS1. For example, the number of memory cells to be selected from the second sub-storage bank SB2 to correspond to the second word line WL2 may be smaller than the number of memory cells to be selected from the first sub-storage bank SB1 to correspond to the first word line WL1. In this case, the number of memory cells to be selected from the second sub-storage bank SB2 to correspond to the first word line WL1 may be smaller than the number of memory cells to be selected from the first sub-storage bank SB1 to correspond to the first word line WL1. Figure 1 The memory controller 110 and the control logic circuit 140 in the embodiment may control the memory device 120 such that when the memory device 120 operates in the metadata mode, the second page size PS2 is smaller than the first page size PS1.
[0094] Fig.11 and Fig.12 is a diagram showing an example of a word line allocation method. For example, Fig.11 The case where the required common / element ratio is 8:1 (e.g., n=8) is shown, and Fig.12 A case where the required normal / meta ratio is 16:1 (eg, n=16) is shown. For example, the required ratio of word lines allocated to normal data and word lines allocated to meta data may be n:1. Fig.11 An example is shown that corresponds to the case where the required common / yuan ratio is 8:1, and Fig.12 This may be an example that corresponds to a situation where the required common / yuan ratio is 16:1.
[0095] Reference Fig.11 and Fig.12, a single memory bank B may be connected to a plurality of word lines. In some embodiments, each r-th word line along the column direction d1 among the plurality of word lines may be allocated to metadata (where r is a positive integer). For example, the remaining word lines except each r-th word line may be allocated to normal data. In some embodiments, the word lines allocated to metadata are not newly added, but a portion of the word lines existing in the existing memory device may be allocated to metadata.
[0096] In some embodiments, a single memory bank (e.g., a first memory bank B1) may include two sub-memory banks (a first sub-memory bank SB1 and a second sub-memory bank SB2) divided along a column direction d1. The first sub-memory bank SB1 may be connected to a plurality of first word lines WL1-0 to WL1-(32k-1) among a plurality of word lines, and the second sub-memory bank SB2 may be connected to a plurality of second word lines WL2-0 to WL2-(32k-1) among a plurality of word lines. The number of first word lines and second word lines may be 32k (where k is 1024). In some embodiments, every r-th word line connected to a sub-memory bank within a single memory bank B may be assigned to metadata of another sub-memory bank, and every r-th word line connected to another sub-memory bank may be assigned to metadata of a sub-memory bank. In this case, the word line assigned to metadata in one sub-memory bank may correspond to the word line assigned to normal data in another sub-memory bank. For example, among the multiple first word lines WL1-0 to WL1-(32k-1) of the first sub-storage body SB1, every rth first word line can be assigned to the metadata of the second sub-storage body SB2 along the column direction d1, and among the multiple second word lines WL2-0 to WL2-(32k-1) of the second sub-storage body SB2, every rth second word line can be assigned to the metadata of the first sub-storage body SB1 along the column direction d1.
[0097] Assigning a word line connected to one sub-memory bank to metadata of another sub-memory bank may mean assigning a word line to metadata corresponding to normal data to be written or read in another sub-memory bank. Therefore, in some embodiments, word lines for normal data and word lines for metadata corresponding to normal data may be arranged to cross each other in terms of sub-memory banks.
[0098] Then, a ratio of first word lines allocated to metadata to the remaining first word lines among the first word lines may be 1:(r-1), and a ratio of second word lines allocated to metadata to the remaining second word lines among the second word lines may be 1:(r-1).
[0099] For example, every r-th first word line may be allocated to metadata corresponding to normal data in every 1st second word line WL2-0 to every r-th second word line WL2-(r-1) among the plurality of second word lines WL2-0 to WL2-(32k-1). Every r-th second word line may be allocated to metadata corresponding to normal data in every 1st word line WL1-0 to every r-1st word line WL1-(r-1) among the plurality of first word lines WL1-0 to WL1-(32k-1). For example, among the multiple first word lines WL1-0 to WL1-(32k-1) of the first sub-storage body SB1, every rth first word line can be allocated to read or write metadata corresponding to the normal data of every 1st second word line to every r-1th second word line of the second sub-storage body SB2, and among the multiple second word lines WL2-0 to WL2-(32k-1) of the second sub-storage body SB2, every rth second word line can be allocated to read or write metadata corresponding to the normal data of every 1st first word line to every r-1th first word line of the first sub-storage body SB1.
[0100] For example, when n is 8, Fig.11 As shown in , r can be 8. For example, referring to Fig.11 , when the required ratio of word lines allocated to normal data and word lines allocated to metadata is 8:1, every eighth word line may be allocated to metadata. Then, every eighth word line may be allocated to metadata, and every first word line to every seventh word line may be allocated to normal data. Therefore, the ratio of normal data word lines to metadata word lines may be 7:1.
[0101] according to Fig.11 , to support the case where the required ratio of word lines allocated to normal data to word lines allocated to metadata is 8:1, every eighth word line of one sub-memory bank may be allocated to read or write metadata corresponding to normal data of every first to seventh word lines of another sub-memory bank. For example, word lines WL1-0 to WL1-6 of the first sub-memory bank SB1 may be allocated to normal data, and word lines WL2-7 may be allocated to read and write metadata corresponding to “normal data written to or read from word lines WL1-0 to WL1-6 of the first sub-memory bank SB1”. In this case, normal data “to be written into or read from the word lines WL1-0 to WL1-6 of the first sub-memory bank SB1” can be read through the first GIO line GIO1, and metadata written into the word line WL2-7 of the second sub-memory bank SB2 corresponding to the normal data can be read through the second GIO line GIO2 (see Figure 3 and Figure 5 ).
[0102] Conversely, the word lines WL2-0 to WL2-6 of the second sub-memory bank SB2 may be allocated to normal data, and the word lines WL1-7 of the first sub-memory bank SB1 may be allocated to read and write metadata corresponding to “normal data written to or read from the word lines WL2-0 to WL2-6 of the second sub-memory bank SB2”. In this case, normal data “to be written to or read from the word lines WL2-0 to WL2-6 of the second sub-memory bank SB2” may be read through the second GIO line GIO2, and metadata written in the word lines WL1-7 of the first sub-memory bank SB1 corresponding to the normal data may be read through the first GIO line GIO1 (see Figure 4 and 6 ).
[0103] Therefore, in the word lines WL1-0 to WL1-7 of the first sub-memory bank SB1 (or the word lines WL2-0 to WL2-7 of the second sub-memory bank SB2), the ratio of normal data to metadata may be 7: 1. The above allocation relationship may be repeated along the column direction from the last word line (e.g., word line WL1-(32k-1) and word line WL2-(32k-1)).
[0104] Reference Fig.12 , when n is 16, r can be 16. For example, referring to Figure 6 , when the required ratio of the word lines allocated to normal data and the word lines allocated to metadata is 16:1, every sixteenth word line may be allocated to metadata. Then, every sixteenth word line may be allocated to metadata, and every first word line to every fifteenth word line may be allocated to normal data. Therefore, the ratio of normal data word lines and metadata word lines may be 15:1.
[0105] according to Fig.12, to support a situation where the required ratio of word lines allocated to normal data to word lines allocated to metadata is 16:1, every sixteenth word line of one sub-memory bank may be allocated to read or write metadata corresponding to “normal data of every first word line to every fifteenth word line of another sub-memory bank”. For example, word lines WL1-0 to WL1-14 of a first sub-memory bank SB1 may be allocated to normal data, and word lines WL2-15 of a second sub-memory bank SB2 may be allocated to read and write metadata corresponding to “normal data written to or read from word lines WL1-0 to WL1-14 of the first sub-memory bank SB1”. In this case, normal data “to be written into or read from the word lines WL1-0 to WL1-14 of the first sub-memory bank SB1” can be read through the first GIO line GIO1, and metadata written into the word line WL2-15 of the second sub-memory bank SB2 corresponding to the normal data can be read through the second GIO line GIO2 (see Figure 3 and Figure 5 ).
[0106] Conversely, the word lines WL2-0 to WL2-14 of the second sub-memory bank SB2 may be allocated to normal data, and the word lines WL1-15 of the first sub-memory bank SB1 may be allocated to read and write metadata corresponding to “normal data written to or read from the word lines WL2-0 to WL2-14 of the second sub-memory bank SB2”. In this case, normal data “to be written to or read from the word lines WL2-0 to WL2-14 of the second sub-memory bank SB2” may be read through the second GIO line GIO2, and metadata written in the word lines WL1-15 of the first sub-memory bank SB1 corresponding to the normal data may be read through the first GIO line GIO1 (see Figure 4 and Figure 6 ).
[0107] Therefore, in the word lines WL1-0 to WL1-15 of the first sub-memory bank SB1 (or the word lines WL2-0 to WL2-15 of the second sub-memory bank SB2), the ratio of normal data to metadata may be 15: 1. The above allocation relationship may be repeated along the column direction from the last word line (e.g., word line WL1-(32k-1) and word line WL2-(32k-1)).
[0108] In this example, a portion of the existing word lines can be allocated to metadata to significantly meet the required normal / meta ratio without adding new lines. For example, the allocation relationship between the normal data word lines and the metadata word lines can be established to cross between sub-memory banks, so that normal data and metadata can be output simultaneously.
[0109] Fig.13 is a diagram illustrating an example of a sub-bank.
[0110] Reference Fig.13 , the sub-storage body SB may correspond to one of the sub-storage bodies SB included in the single storage body described in the above example, and may be connected to a plurality of word lines WL0 to WL(32k-1) extending along the row direction d2, a plurality of CSL subsets L_CS0 to U_CS(i-1) extending along the column direction d1, a row decoder RD, and an input / output sense amplifier stage IOSAL. The sub-storage body SB may include a plurality of pads MAT. Some word lines and some CSLs may be connected to each pad MAT. The MAT may include a two-dimensionally arranged memory cell array, and each memory cell may be connected to a single word line among some word lines connected to the corresponding MAT and a single CSL among some CSLs.
[0111] A plurality of CSL subsets L_CS0 to U_CS(i-1) may be connected to a plurality of MATs corresponding to a single column. As shown in the figure, during a single access, a plurality of CSL subsets (L_CS0 to U_CS(i-1)) or a pair of CSL subsets may be selected. For example, a pair of CSL subsets L_CS0 and L_CS1 to a pair of CSL subsets L_CS(i-2) and L_CS(i-1) (where i is a positive integer) may be selected on a first side L, and a pair of CSL subsets U_CS0 and U_CS1 to a pair of CSL subsets U_CS(i-2) and U_CS(i-1) may be selected on a third side U. In addition, a pair of CSL subsets M_CS0 and M_CS1 may be selected.
[0112] Fig.14 It is shown Figure 7 FIG. 1 is a diagram of an example of area “A”.
[0113] Reference Fig.14, for example, multiple CSLs that can be classified into CSL groups CS1 and CS2 can be connected to multiple MATs corresponding to a single CSL block CB. A portion of the multiple CSLs can be connected to each of the CSL groups CS1 and CS2. As an example, multiple CSLs (for example, 64 CSLs, CSL0 to CSL63 as shown in the figure) can be connected to each of the CSL groups CS1 and CS2. Therefore, a total of 128 CSLs can be connected to a single CSL block CB. During a single access, a single CSL can be selected from each of the CSL groups CS1 and CS2. Hereinafter, unless otherwise specified, the CSLs selected from the CSL groups CS1 and CS2 for each CSL block CB and ECC block in a single sub-storage body SB will be referred to as CSL subsets, and the CSLs selected for each CSL block CB will be referred to as a pair of CSL subsets.
[0114] return Fig.13 In some embodiments, a pair of CSL subsets or a portion of a pair of CSL subsets corresponding to a single CSL block CB included in a single sub-memory bank SB may be allocated to “a single word line among the word lines connected to the remaining sub-memory banks that constitute a memory bank together with the single sub-memory bank SB” (e.g., a single word line among the word lines for normal data). For example, all CSL subsets included in the single sub-memory bank SB may be each allocated to one of the normal data word lines connected to the remaining sub-memory banks. In some embodiments, referring to Fig. 9 , a single sub-memory bank SB can be connected with Fig. 9 The first sub-memory bank SB1 corresponds to the first sub-memory bank SB1, and the remaining sub-memory banks can be Fig. 9 Corresponding to the second sub-storage body SB2.
[0115] For example, in some embodiments, a single memory bank may include one sub-memory bank SB and another sub-memory bank. In this case, when access to normal data is performed through the remaining word lines except every r-th word line in one sub-memory bank SB, access to metadata may be performed on every r-th word line in another sub-memory bank. In this case, a pair of CSL subsets or a portion of a pair of CSL subsets in another sub-memory bank may be allocated to each of the remaining word lines of the single sub-memory bank SB. Therefore, when access to normal data is performed on one of the remaining word lines of the single sub-memory bank SB, access to metadata may be performed on a pair of CSL subsets or a portion of a pair of CSL subsets allocated to one of the remaining word lines.
[0116] All CSL subsets may be selected in the normal-side sub-bank SB where access to normal data is performed. Therefore, access to normal data may be performed on the normal data word line and all CSL subsets.
[0117] In some embodiments, "a pair of CSL subsets L_CS0 and L_CS1 to L_CS(i-2) and L_CS(i-1) included in the first side L, and a pair of CSL subsets U_CS0 and U_CS1 to U_CS(i-2) and U_CS(i-1) included in the third side U among the CSL subsets connected to a single sub-storage body SB" can be sequentially assigned to normal data word lines connected to the remaining sub-storage bodies SB starting from a pair of CSL subsets with the lowest index.
[0118] In some embodiments, among the CSL subsets connected to a single sub-storage body SB, "the CSL subsets L_CS0 to L_CS(i-1) included in the first side L and the CSL subsets U_CS0 to U_CS(i-1) included in the second side U" can be sequentially assigned to the normal data word lines connected to the remaining sub-storage bodies SB starting from the CSL subset with the lowest index.
[0119] Fig.15 and Fig.16 is a diagram showing an example of column selection signals for normal data and metadata. Fig.15 and Fig.16 An example of a combination of a meta-side sub-bank (MSB) side column selection signal corresponding to a normal-side sub-bank (NSB) side column selection signal is shown. Although the difference between NSB and MSB is based on the location where normal data is stored, each sub-bank can be considered as NSB and MSB based on the relative sub-bank. For example, referring to Figure 6 , when the first sub-memory bank SB1 is NSB, the MSB may be the second sub-memory bank SB2. Also, when SB2 is NSB, the MSB may be the first sub-memory bank SB1.
[0120] exist Fig.15 and Fig.16 In the embodiment, CSLx[m:n] may be defined as a column selection signal, that is, a signal for selecting CSL. The column selection signal may be generated by a column decoder.
[0121] Here, m and n represent the range of CSL subsets to be selected from multiple CSL subsets assigned to a single storage body (or sub-storage body), and m and n are also positive integers. For example, m represents the lowest index among the indexes of the CSL subsets to be selected, and n represents the highest index among the indexes of the CSL subsets to be selected.
[0122] The column decoder may select the mth to nth CSL subsets from the CSL subsets on each of the first and second sides for a single memory bank (or sub-memory bank) based on CSLx[m:n], and may select the xth CSL from the plurality of CSLs corresponding to the selected CSL subset. Fig.15 When m=0 and n=1 in , the column decoder may select L_CS0 and L_CS1 on the first side L, and select U_CS0 and U_CS1 on the second side U.
[0123] Here, x represents the Fig.14 The index of the CSL selected from the plurality of CSLs included in each of the pair of CSL subsets in , where x is a positive integer. Fig.14 , for example, when x=1, CSL1 may be selected from CS1, and CSL1 may be selected from CS2.
[0124] Back to Fig.15 , for example, when the required normal / meta ratio is 8, every eighth wordline may be allocated to metadata. For example, the ratio of wordlines allocated to normal data to wordlines allocated to metadata may actually be 7: 1. This is intended to significantly meet the required 8: 1 ratio without requiring additional chip size overhead.
[0125] As for the normal side sub-memory bank NSB, access to normal data can be performed on each of the first to seventh word lines WL0 to WL6. Fig.15 As shown in , the column selection signal for the normal-side sub-memory bank NSB may be CSLx[0:15] to select all CSL subsets, and the value of x may range from 0 to 63. The eighth word line WL7 of the normal-side sub-memory bank NSB may be allocated to the meta-side sub-memory bank MSB.
[0126] Next, with respect to the meta-side sub-memory bank MSB, the eighth word line WL7 may be assigned to the normal side sub-memory bank NSB. In this case, the column selection signal may be defined based on the word line accessed in the normal side sub-memory bank NSB. In this example, a pair of CSL subsets of the meta-side sub-memory bank MSB may be sequentially assigned to the normal data word lines connected to the normal side sub-memory bank NSB. For example, when the first word line WL0 of the normal side sub-memory bank NSB is accessed, the m and n of the column selection signal may be determined as 0 and 1, respectively, to select a pair of CSL subsets with the lowest index. Next, when the second word line WL1 of the normal side sub-memory bank NSB is accessed, the m and n of the column selection signal may be determined as m=2 and n=3 to select a pair of CSL subsets with the next index. When the last sixth word line WL6 is accessed, the m and n of the column selection signal may be determined as m=12 and n=13 based on the above-mentioned allocation relationship. In this example, the fifteenth and sixteenth CSL subsets of the meta-side may not be allocated.
[0127] according to Fig.15The above-mentioned column selection signal, for example, when each CSL is connected to eight bit lines, 256 bits of normal data (8 bits × 32 (the number of CSL subsets selected in the normal side sub-storage bank NSB)), 32 bits of normal data (8 bits × 4 (the number of CSL subsets selected in the meta side sub-storage bank MSB)) and 16 parity bits can be output during access to a single word line.
[0128] For example, refer to Fig.16 , when the required normal / meta ratio is 16, the sixteenth word line may be allocated to metadata. For example, the ratio of word lines allocated to normal data and word lines allocated to metadata may actually be 15: 1. This is intended to significantly meet the required 16: 1 ratio without requiring additional chip size overhead.
[0129] With respect to the normal side sub-memory bank NSB, access to normal data may be performed on each of the first word line WL0 to the fifteenth word line WL14. As described above, the column selection signal for the normal side sub-memory bank NSB may be CSLx[0:15] to select all CSL subsets. Similarly, the sixteenth word line WL15 of the normal side sub-memory bank NSB may be allocated to the meta side sub-memory bank MSB.
[0130] Next, with respect to the meta-side sub-memory bank MSB, the sixteenth word line WL15 may be allocated to the normal-side sub-memory bank NSB. In this example, the CSL subsets of the meta-side sub-memory bank MSB may be sequentially allocated to the “normal data word line for normal data connected to the normal-side sub-memory bank NSB” starting from the CSL subset with the lowest index. Fig.15 , only one CSL subset is selected so that the column select signal of the MSB of the meta-side sub-bank can be defined as CSLx[m].
[0131] For example, when the first word line WL0 of the normal side sub-memory bank NSB is accessed, the m of the column selection signal may be determined to be 0 to select the CSL subset with the lowest index. Next, when the second word line WL1 of the normal side sub-memory bank NSB is accessed, the m of the column selection signal may be determined to be 1 to select the CSL subset with the next index. According to the above allocation relationship, when the last fifteenth word line WL14 is accessed, the m of the column selection signal may be determined to be 14. In this example, the sixteenth CSL subset of the meta side may not be allocated.
[0132] according to Fig.16The above-mentioned column selection signal, for example, when each CSL is connected to eight bit lines, 256 bits of normal data (8 bits × 32 (the number of CSLs selected in the normal side sub-storage bank NSB)), 16 bits of metadata (8 bits × 2 (the number of CSL subsets selected in the meta-side sub-storage bank MSB)) and 16 bits of parity bits can be output during accessing a single word line.
[0133] In this example, when the memory bank operates in the metadata mode and two word lines are activated at the same time, the page size of the sub-memory bank corresponding to the metadata MD can be controlled to be smaller than the page size of the sub-memory bank corresponding to the normal data ND. Therefore, even when the current increases due to the simultaneous activation of the two word lines, the page size corresponding to the metadata MD can be reduced to offset the current generated due to the simultaneous activation of the word lines.
[0134] As described above, in some embodiments, a volatile memory device may be provided for allocating specific data.
[0135] Although the present disclosure includes many specific implementation details, these should not be interpreted as limiting the scope of possible protection. Specific features described in the context of a single embodiment in the present disclosure may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented in multiple embodiments individually or in any suitable sub-combination. In addition, although the features may be described above as working in a specific combination, one or more features from the combination may be deleted from the combination in some cases, and such a combination may point to a sub-combination or a variation of a sub-combination.
[0136] While example embodiments have been shown and described above, it will be apparent to those skilled in the art that modifications and variations may be made without departing from the scope of the inventive concept as defined by the appended claims.
Claims
1. A memory device, include: a memory bank connected to a plurality of word lines, a first global input and output GIO line, and a second GIO line, the first GIO line having a first length in a column direction, and the second GIO line having a second length in the column direction, wherein the second length is greater than the first length, One of the first GIO line and the second GIO line is allocated to metadata.
2. The memory device according to claim 1, in, The memory bank includes a first sub-memory bank and a second sub-memory bank divided along a column direction. The first sub-bank is connected to the first GIO line and a first word line included in the plurality of word lines, and The second sub bank is connected to a second GIO line and a second word line included in the plurality of word lines.
3. The memory device according to claim 2, in, The first GIO line and the second GIO line are configured to simultaneously input or output data based on the first word line and the second word line being activated.
4. The memory device according to claim 1, in, One of the first GIO line and the second GIO line is configured to input or output metadata, and the other of the first GIO line and the second GIO line is configured to simultaneously input or output normal data corresponding to the metadata.
5. The memory device according to claim 4, in, The ratio of metadata to normal data is 1:n, where n is a positive integer.
6. The memory device according to claim 1, in, Each of the first GIO line and the second GIO line is provided with M GIO lines, where M is a positive integer.
7. The memory device according to claim 1, in, The first GIO line and the second GIO line include different metals.
8. The memory device according to claim 1, further comprising: include: a row decoder configured to select the plurality of word lines; as well as The column decoder is configured to select a plurality of column selection lines respectively connected to the first GIO line and the second GIO line.
9. The memory device according to claim 8, in, The row decoder is configured to simultaneously select a first word line and a second word line among the plurality of word lines.
10. The memory device according to claim 1, further comprising: include: a first sense amplifier connected to the first GIO line and configured to sense and amplify data output from the first GIO line; as well as A second sense amplifier is connected to the second GIO line and is configured to sense and amplify data output from the second GIO line.
11. The memory device according to claim 10, in, The first GIO line is a first GIO line among the plurality of first GIO lines, The second GIO line is a second GIO line among the plurality of second GIO lines. The first sense amplifier is one of the plurality of first input / output sense amplifiers, The second sense amplifier is one of the plurality of second input / output sense amplifiers, There are a first same number of the plurality of first input / output sense amplifiers and the plurality of first GIO lines, and There is a second same number of the second plurality of input / output sense amplifiers and the second plurality of GIO lines.
12. A memory device, include: a memory bank connected to the plurality of word lines, a first global input and output GIO line, and a second GIO line, the second GIO line being formed to have a greater length in a column direction than the first GIO line; as well as a row decoder connected to the plurality of word lines, extending along a column direction, and configured to select the plurality of word lines, One of the first GIO line and the second GIO line is allocated to metadata.
13. The memory device according to claim 12, in, The memory bank includes a first sub-memory bank and a second sub-memory bank divided along a column direction. The first sub-bank is connected to the first GIO line and a first word line included in the plurality of word lines, and The second sub bank is connected to a second GIO line and a second word line included in the plurality of word lines.
14. The memory device according to claim 13, in, The row decoder is configured to simultaneously select the first word line and the second word line in an operation mode for inputting or outputting metadata.
15. The memory device according to claim 12, in, In an operation mode for inputting or outputting metadata, one of the first GIO line and the second GIO line is configured to input or output metadata, and the other of the first GIO line and the second GIO line simultaneously inputs or outputs normal data corresponding to the metadata.
16. The memory device according to claim 15, in, When the operation mode is off, one of the first GIO line and the second GIO line and the other of the first GIO line and the second GIO line are configured to input or output normal data.
17. The memory device according to claim 15, in, A ratio of metadata input or output from the one GIO line to normal data input or output from the other GIO line is 1:n, where n is a positive integer.
18. A memory device, include: A memory bank, comprising a plurality of sub-memory banks divided along a column direction; a plurality of word lines, each word line extending along a row direction and connected to the plurality of sub-memory banks; as well as A plurality of global input / output (GIO) lines extending along a column direction and connected to the plurality of sub-memory banks, The plurality of GIO lines include two GIO lines connected to different sub-memory banks, and one of the two GIO lines is allocated to metadata.
19. The memory device according to claim 18, in, The two GIO lines are formed to have different lengths in a column direction.
20. The memory device according to claim 18, in, Two word lines connected to different sub-memory banks among the plurality of word lines are simultaneously activated.
