PROCESSING DEVICE COMPRISING BITCELL CIRCUITS MADE OF A PAIR OF VARIABLE RESISTORS IN PARALLEL

Abstract

 A bitcell circuit includes: a first variable resistor as a resistive memory device having a resistance value that is set based on the resistive memory device switching between different resistance states; a second variable resistor as another resistive memory device connected to the first variable resistor in parallel, wherein the second variable resistor is set with a resistance value complementary to the resistance value of the first variable resistor; a first switch serially connected to the first variable resistor and configured to switch application of a voltage or current to the first variable resistor; and a second switch serially connected to the second variable resistor and configured to perform a switching operation for applying the voltage or current to the first variable resistor, complementarily to the first switch.

Description

BACKGROUND

1. Field

[0001] The following description relates to a processing device and an electronic device having the processing device.

2. Description of the Related Art

[0002] Neural network devices may perform a multiply-accumulate (MAC) operation of repeating multiplications and additions. A neural network may repeatedly perform a MAC operation of multiplying the values of nodes of a previous layer with weights mapped to the nodes and adding multiplication results at a specific node, and may perform an operation of applying an activation function to a result of the MAC operation. To this end, a memory access operation of loading an appropriate input and weight at a desired or determined time point may also be performed. However, such neural network operations, such as the MAC operation, may not be efficiently performed using other hardware architecture instead of a generally known digital computer.

SUMMARY

[0003] This Summary is provided to introduce a selection of concepts in simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

[0004] In one general aspect, a bitcell circuit (or a bitcell) includes: a first variable resistor as a resistive memory device having a resistance value that is set based on the resistive memory device switching between different resistance states; a second variable resistor as another resistive memory device connected to the first variable resistor in parallel, wherein the second variable resistor is set with a resistance value complementary to the resistance value of the first variable resistor; a first switch serially connected to the first variable resistor and configured to switch application of a voltage or current to the first variable resistor; and a second switch serially connected to the second variable resistor and configured to perform a switching operation for applying the voltage or current to the first variable resistor, complementarily to the first switch.

[0005] The first variable resistor may be set with either one of a first resistance value and a second resistance value.

[0006] The second variable resistor may be set with the second resistance value when the first variable resistor is set with the first resistance value, and may be set with the first resistance value when the first variable resistor is set with the second resistance value.

[0007] The resistance values of the first variable resistor and the second variable resistor may be set with values corresponding to weights for a multiply and accumulate (MAC) operation.

[0008] The first switch and the second switch may complementarily perform on/off operations according to an input value applied to the bitcell circuit to perform a MAC operation of a neural network.

[0009] A resistance value of the bitcell circuit may be equal to the resistance value of the first variable resistor when the first switch is closed by application of a first input value, and may be equal to the resistance value of the second variable resistor when the second switch is closed by application of a second input value.

[0010] A resistance value of the bitcell circuit may be a first resistance value when the first variable resistor is set with the first resistance value corresponding to a first weight and a first input value is applied to the bitcell circuit.

[0011] A resistance value of the bitcell circuit may be the first resistance value when the second variable resistor is set with the second resistance value corresponding to a second weight and a second input value is applied to the bitcell circuit.

[0012] A resistance value of the bitcell circuit may be a second resistance value when the second variable resistor is set with the second resistance value corresponding to the second weight and the first input value is applied to the bitcell circuit.

[0013] A resistance value of the bitcell circuit may be the second resistance value when the first variable resistor is set with the first resistance value corresponding to the first weight and the second input value is applied to the bitcell circuit.

[0014] The bitcell circuit may have a resistance value corresponding to a result of an exclusive-NOR (XNOR) operation between the input value applied to the bitcell circuit and a weight set for the bitcell circuit.

[0015] The first variable resistor and the first switch serially connected to each other may be parallelly connected with the second variable resistor and the second switch serially connected to each other,

[0016] One end of the first variable resistor and one end of the second variable resistor may be commonly connected to a first bit data line.

[0017] One end of the first switch and one end of the second switch may be commonly connected to a second bit data line.

[0018] Another bitcell circuit may be connected to either one of: the one end of the first variable resistor and the one end of the second variable resistor commonly connected to the first bit data line; and the one end of the first switch and the one end of the second switch commonly connected to the second bit data line is connected to.

[0019] The first variable resistor and the second variable resistor are magnetic tunnel junction (MTJ) devices.

[0020] A processing device includes a bitcell array including a plurality of bitcells including the bitcell circuit.

[0021] In another general aspect, a bitcell circuit includes: a pair of variable resistors as resistive memory devices set to have different resistance values and connected to each other in parallel; and a pair of switches serially connected to the variable resistors, respectively, and configured to complementarily switch applications of a voltage or current to the variable resistors.

[0022] In another general aspect, a processing device includes: a bitcell array including a plurality of bitcells (or a plurality of bitcell circuits) each including a pair of variable resistors and a pair of switches, wherein at least one of the plurality of bitcells includes: a first variable resistor as a resistive memory device having a resistance value set based on the resistive memory device switching between different resistance states; a second variable resistor as another resistive memory device connected to the first variable resistor in parallel, wherein the second variable resistor is set with a resistance value complementary to the resistance value of the first variable resistor; a first switch serially connected to the first variable resistor and configured to switch application of a voltage or current to the first variable resistor; and a second switch serially connected to the second variable resistor and configured to perform a switching operation for applying the voltage or current to the first variable resistor, complementarily to the first switch.

[0023] The plurality of bitcells may form the bitcell array including a plurality of bitcell lines.

[0024] Each of the plurality of bitcell lines may include serially connected bitcells from among the plurality of bitcells.

[0025] A first bitcell line from among the plurality of bitcell lines may be configured to perform processing of a MAC operation of a first node from among a plurality of nodes of a neural network.

[0026] The pair of variable resistors included in each of bitcells of the first bitcell line may be set with resistance values corresponding to weights for the MAC operation of the first node.

[0027] In each of the bitcells included in the first bitcell line, one switch of the pair of switches may be closed and another switch of the pair may be opened, based on input values of the MAC operation of the first node.

[0028] A result of the MAC operation of the first node may correspond to a value of a voltage drop of the first bitcell line that occurs due to a predetermined value of current applied to the first bitcell line, in response to the setting of the resistance values corresponding to the weights and switching operations of the switches corresponding to the input value.

[0029] A value of the voltage drop of the first bitcell line may correspond to a sum of values of voltage drops occurring in the bitcells included in the first bitcell line.

[0030] The processing device may further include a pair of bit data lines connected between one or more of the bitcells and configured to apply a voltage or current to both ends of the one or more of the bitcells, to set resistance values of the pair of variable resistors.

[0031] The processing device may be a device that performs in-memory processing.

[0032] In another general aspect, a processing device including a bitcell array includes: a plurality of bitcells (or a plurality of bitcell circuits); and a pair of bit data lines connected between one or more of the bitcells and configured to apply a voltage or current to both ends of the one or more of the bitcells, wherein at least one of the plurality of bitcells includes: a pair of variable resistors as resistive memory devices set to have different resistance values according to switching between different resistance states and connected to each other in parallel; and a pair of switches serially connected to the variable resistors, respectively, and configured to complementarily switch applications of a voltage or current to the variable resistors.

[0033] Other features and aspects will be apparent from the following detailed description, the drawings, and the claims.

BRIEF DESCRIPTION OF THE DRAWINGS

[0034] 

FIG. 1 illustrates a neural network node model according to one or more embodiments.

FIG. 2 illustrates an example of a neural network;

FIG. 3 illustrates an example of a circuit diagram of a bitcell used in a processing device;

FIGS. 4A and 4B illustrate examples of a structure and operation of a variable resistor applied to a bitcell;

FIG. 5A illustrates an example of a neural network;

FIG. 5B illustrates an example of a circuit diagram of a processing device configured to perform an operation of a neural network;

FIG. 6 illustrates an example of a circuit diagram of a processing device configured to perform an operation of a neural network;

FIG. 7 illustrates an example of a connection relationship between a variable resistor and a switch;

FIG. 8 illustrates an example of a horizontally cross-sectional view illustrating a variable resistor layer;

FIG. 9A illustrates an example of a vertical cross-sectional view of bitcells;

FIG. 9B illustrates an example of a vertical cross-sectional view of bitcells;

FIG. 10 illustrates an example of a plan view of an active layer of a processing device;

FIG. 11 illustrates an example of a plan view of a seventh metal layer of a processing device;

FIG. 12 illustrates an example of a circuit diagram of a processing device having a connection structure of a switch and a variable resistor that is different from that of another processing device;

FIG. 13 illustrates an example of a connection relationship between a variable resistor and a switch;

FIG. 14 illustrates an example of a processing device; and

FIG. 15 illustrates an example of an electronic device.

DETAILED DESCRIPTION

[0035] The following detailed description is provided to assist the reader in gaining a comprehensive understanding of the methods, apparatuses, and/or systems described herein. However, various changes, modifications, and equivalents of the methods, apparatuses, and/or systems described herein will be apparent after an understanding of the disclosure of this application. For example, the sequences of operations described herein are merely examples, and are not limited to those set forth herein, but may be changed as will be apparent after an understanding of the disclosure of this application, with the exception of operations necessarily occurring in a certain order. Also, descriptions of features that are known in the art, after an understanding of the disclosure of this application, may be omitted for increased clarity and conciseness.

[0036] Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the like elements throughout. In this regard, the one or more embodiments may have different forms and should not be construed as being limited to the descriptions set forth herein. Accordingly, the embodiments are merely described below, by referring to the figures, to explain aspects of the present description. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items. Expressions such as "at least one of," when preceding a list of elements, modify the entire list of elements and do not modify the individual elements of the list.

[0037] The terminology used herein is for the purpose of describing particular examples only, and is not to be used to limit the disclosure. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. As used herein, the term "and/or" includes any one and any combination of any two or more of the associated listed items. As used herein, the terms "include," "comprise," and "have" specify the presence of stated features, numbers, operations, elements, components, and/or combinations thereof, but do not preclude the presence or addition of one or more other features, numbers, operations, elements, components, and/or combinations thereof. The use of the term "may" herein with respect to an example or embodiment (for example, as to what an example or embodiment may include or implement) means that at least one example or embodiment exists where such a feature is included or implemented, while all examples are not limited thereto.

[0038] In addition, terms such as first, second, A, B, (a), (b), and the like may be used herein to describe components. Each of these terminologies is not used to define an essence, order, or sequence of a corresponding component but used merely to distinguish the corresponding component from other component(s). Although terms of "first" or "second" are used herein to describe various members, components, regions, layers, or sections, these members, components, regions, layers, or sections are not to be limited by these terms. Rather, these terms are only used to distinguish one member, component, region, layer, or section from another member, component, region, layer, or section. Thus, a first member, component, region, layer, or section referred to in examples described herein may also be referred to as a second member, component, region, layer, or section without departing from the teachings of the examples.

[0039] Throughout the specification, when an element, such as a layer, region, or substrate, is described as being "on," "connected to," or "coupled to" another element, it may be directly "on," "connected to," or "coupled to" the other element, or there may be one or more other elements intervening therebetween. In contrast, when an element is described as being "directly on," "directly connected to," or "directly coupled to" another element, there can be no other elements intervening therebetween. Likewise, expressions, for example, "between" and "immediately between" and "adjacent to" and "immediately adjacent to" may also be construed as described in the foregoing.

[0040] The embodiments described below relate to a technical field of a processing device, for example, a neuromorphic processor or a neural processor, and detailed descriptions on items that are well-known may be omitted.

[0041] According to embodiments, a processing device of one or more embodiments may be provided with an analog circuit for processing multiplication and addition operations, unlike a general digital computer in which information is exchanged by using a common data bus. In other words, the processing device may perform in-memory processing or internal processing. Accordingly, the processing device may be referred to as various terms such as an in-memory processing device, a processor in memory (PIM), and a function in memory (FIM).

[0042] FIG. 1 illustrates a neural network node model according to one or more embodiments.

[0043] The neural network node model 11 may include, as an example of neuromorphic computations, a multiplication computation that multiplies information from a plurality of neurons or nodes by a synaptic weight, an addition computation Σ for values ω₀x₀, ω₁x₁, ω₂x₂ obtained by multiplying the synaptic weight, and a computation for applying a characteristic function b and an activation function f to a result of the addition computation. A neuromorphic computation result may be provided by a neuromorphic computation. Here, values like x₀, x₁, x₂, and so on correspond to axon values, and values like ω₀, ω₁, ω₂, and so on correspond to synaptic weights. While the nodes, values, and weights of the neural network node model 11 may be respectively referred to as "neurons," "axon values," and "synaptic weights," such reference is not intended to impart any relatedness with respect to how the neural network architecture computationally maps or thereby intuitively recognizes information and how a human's neurons operate. I.e., the terms are merely terms of art referring to the hardware implemented nodes, values, and weights of the neural network node model 11.

[0044] FIG. 2 illustrates an example of a neural network (e.g., a neural network 20).

[0045] Referring to FIG. 2, the neural network 20 is an example of a neural network that implements the above-described neural network node model, and may correspond to a deep neural network (DNN). Although, for convenience of explanation, the neural network 20 is illustrated as including two hidden layers (e.g., hidden layer 1 and hidden layer 2), the neural network 20 may include various numbers of hidden layers (e.g., three or more hidden layers). Furthermore, although FIG. 2 illustrates that the neural network 20 separately includes an input layer 21 for receiving input data, the input data may be directly input to the hidden layers.

[0046] In the neural network 20, nodes of layers, except an output layer, may be connected to nodes of a next layer through links for transmitting an output signal. Values obtained by multiplying node values of nodes included in the previous layer by weights respectively allocated to the links may be input to one node through the links. The node values of the previous layer may correspond to axon values and the weights correspond to synaptic weights. Each weight may be referred to as a parameter of the neural network 20. An activation function may include a sigmoid, a hyperbolic tangent (tanh), and a rectified linear unit (ReLU), and non-linearity may be formed in the neural network 20 by the activation function.

[0047] The output of any one node 22 included in the neural network 20 may be expressed as in Equation 1 below, for example.

[0048] Equation 1 may represent an output value y_i of the i-th node 22 with respect to m input values in any layer. In Equation 1 x_j may denote an output value of the j-th node of a previous layer, and w_j,i may denote a weight applied to a connection between the j-th node of the previous layer and the i-th node 22 of a current layer. Also, f() may denote an activation function. As shown in Equation 1, an accumulated result of multiplications of the input value x_j and the weight w_j,i may be used with respect to the activation function. In other words, a multiply-accumulate (MAC) operation of multiplying and adding an appropriate input value x_j and weight w_j,i at a desired or determined time point may be repeated. In addition to the above use, there are various application fields using the MAC operation. To this end, a processing device capable of processing the MAC operation in an analog circuit domain may be used.

[0049] FIG. 3 illustrates an example of a circuit diagram of an embodiment of a bitcell (or a bitcell circuit) BC used in a processing device. A structure and operation of the bitcell BC will now be described with reference to FIG. 3.

[0050] The bitcell BC of FIG. 3 may have a circuit structure included in a processing device for implementing a neuromorphic processor, a neural processor, or the like. The processing device may be, for example, an in-memory processing unit that stores data in a memory (resistive memory device or the like) and uses the stored data when an operation is necessary.

[0051] The bitcell BC may include a pair of variable resistors R_a and R_b connected to each other in parallel, a pair of switches S_a and S_b serially connected to the variable resistors R_a and R_b, respectively, and switches S_BDLa and S_BDLb connected to first and second bit data lines BDL_a and BDL_b, respectively. However, the circuit structure of the bitcell BC of FIG. 3 is only an example, and the bitcell BC may be implemented by an equivalent circuit using different circuit devices.

[0052] The pair of variable resistors R_a and R_b may be resistors that may be set with different resistance values, and the resistance values of the variable resistors R_a and R_b may be determined by a weight that is applied to the bitcell BC. For example, each of the variable resistors R_a and R_b may have one of two resistance values, for example, a resistance value of 20 Ω or 5 Ω. For example, when a weight applicable to the bitcell BC is -1 or 1, and when the weight of '1' is applied to the bitcell BC, the first variable resistor R_a may be set to be 20 Ω and the second variable resistor R_b may be set to be 5 Ω. On the other hand, when a weight of '-1' is applied to the bitcell BC, the first variable resistor R_a may be set to be 5 Ω and the second variable resistor R_b may be set to be 20 Ω. In other words, the resistance values of the variable resistors R_a and R_b may be complementarily set such that the variable resistors R_a and R_b have different resistance values.

[0053] In detail, the variable resistors R_a and R_b may be resistive memory devices. The resistive memory device may be switchable between different resistance states according to voltages or currents applied to both ends of the resistive memory device, and thus may have a plurality of resistance states. The resistive memory device may have a single-layered structure or multilayered structure including, for example, a transition metal oxide, a metal oxide such as a perovskite material, a phase change material such as a chalcogenide material, a ferroelectric material, a ferromagnetic material, etc. Meanwhile, an operation in which the resistive memory device changes from a high resistance state to a low resistance state may be referred to as a SET operation, and an operation in which the resistive memory device changes from a low resistance state to a high resistance state may be referred to as a RESET operation.

[0054] A method of changing the resistance values of the variable resistors R_a and R_b will now be described. First, one end of a variable resistor to be changed may be connected to the first bit-data line BDL_a and another end of the variable resistor may be connected to the second bit-data line BDL_b. When illustrating and describing the first variable resistor R_a, one end of the first variable resistor R_a (upper end of the variable resistor R_a in FIG. 3) may be connected to the second bit-data line BDL_b through the upper bit-data line switch S_BDLb, and the other end of the first variable resistor R_a (lower end of the variable resistor R_a in FIG. 3) may be connected to the first bit-data line BDL_a through the first switch S_a and the lower bit-data line switch S_BDLa.

[0055] In an example, the upper bit-data line switch S_BDLb may not be included in the bitcell BC of FIG. 3. For example, the upper bit-data line switch S_BDLb may be a switch included in a bitcell (not shown) adjacent to the bitcell BC of FIG. 3, or an independent switch not included in any bitcell.

[0056] When the first variable resistor R_a is connected to the bit data lines BDL_a and BDL_b, a SET operation or a RESET operation may be performed with respect to the first variable resistor R_a by controlling a voltage on both ends of the first variable resistor R_a or a current flowing in the first variable resistor R_a through the bit data lines BDL_a and BDL_b. However, when the second switch S_b and the bit-data line switches S_BDLa and S_BDLb on both ends of the bitcell BC are closed, the second variable resistor R_b may be connected to the bit data lines BDL_a and BDL_b, and thus a SET operation or a RESET operation may be performed with respect to the second variable resistor R_b.

[0057] A voltage and/or current applied to change the resistance values of the variable resistors R_a and R_b may be much greater than a voltage and/or current applied to read the resistance values of the variable resistors R_a and R_b. In other words, the voltage and/or current applied to read the resistance values of the variable resistors R_a and R_b may not change (or have a negligible effect on) the resistance values of the variable resistors R_a and R_b.

[0058] The pair of switches S_a and S_b serially connected respectively to the variable resistors R_a and R_b may perform an ON/OFF operation according to what input is applied to the bitcell BC. The switches S_a and S_b may complementarily operate so that, when one switch is closed, the other switch is open. For example, when an input applicable to the bitcell BC is '-1' or '1', it may be designed that, when the input of '1' is applied, the first switch S_a is closed and the second switch S_b is open, and when the input of '-1' is applied, the first switch S_a is open and the second switch S_b is closed.

[0059] According to the above-described operation methods of a variable resistor and a switch, resistance values measured from both ends of the bitcell BC of FIG. 3 may vary according to weights and inputs applied to the bitcell BC. A relationship between a weight, an input, and a resistance value on both ends of the bitcell BC may be summarized in Table 1 below, for example.

Table 1:

Input	Weight	Input * Weight	Resistance Value (Ω)
1.	1.	1.	20.
1.	-1.	-1.	5.
-1.	1.	-1.	5.
-1.	-1.	1.	20.

[0060] Referring to Table 1, when a product of the input and the weight is 1, the resistance value of the bitcell BC is 20 Ω, and, when the product of the input and the weight is -1, the resistance value of the bitcell BC is 5 Ω. In other words, when the resistance value of the bitcell BC is measured or a voltage drop of the bitcell BC due to a current having a constant value is measured, a product of an input and a weight applied to the bitcell BC may be determined. A processing device (for example, a neuromorphic processor) that obtains a sum of the products of inputs and weights may be implemented by using the characteristics of the bitcell BC.

[0061] FIGS. 4A and 4B illustrate examples of a structure and operation of a variable resistor applied to a bitcell (e.g., the bitcell BC of FIG. 3). The variable resistors R_a and R_b may be implemented as magnetic tunnel junction (MTJ) devices, and may each have a resistance value that varies according to the magnitude and direction of a current (or voltage) and may exhibit a non-volatile feature of keeping the resistance value unchanged even when an input current (or voltage) is cut off.

[0062] Referring to FIGS. 4A and 4B, an MTJ device may include a pinned layer L3, a free layer L1, and a tunnel layer L2 therebetween. A magnetization direction of the pinned layer L3 is fixed, and a magnetization direction of the free layer L1 may be the same as or different from the magnetization direction of the pinned layer L3 according to conditions.

[0063] FIG. 4A illustrates that the magnetization directions of the free layer L1 and the pinned layer L3 are parallel to each other in the MTJ device. When the magnetization directions are parallel as above, the MTJ device may have a low resistance value (for example, a resistance value of 5 Ω). FIG. 4B illustrates that the magnetization directions of the free layer L1 and the pinned layer L3 in the MTJ device are anti-parallel to each other. When the magnetization directions are anti-parallel as above, the MTJ device may have a high resistance value (for example, a resistance value of 20 Ω). Accordingly, the resistance value of the variable resistor may be changed by changing the magnetization direction of the free layer L1.

[0064] The magnetization direction of the free layer L1 may be changed by an electrical/magnetic factor provided inside and/or outside a resistive memory cell. The free layer L1 may include a material having a changeable magnetization direction (for example, a ferromagnetic material). The free layer L1 may include, for example, CoFeB, FeB, Fe, Co, Ni, Gd, Dy, CoFe, NiFe, MnAs, MnBi, MnSb, CrO₂, MnOFe₂O₃, FeOFe₂O₃, NiOFe₂O₃, CuOFe₂O₃, MgOFe₂O₃, EuO, Y₃Fe₅O₁₂, and/or a combination thereof.

[0065] The tunnel layer L2 may have a thickness thinner than a spin diffusion distance and may include a non-magnetic material (for example, magnesium (Mg), titanium (Ti), aluminum (Al), magnesium-zinc (MgZn), and oxides of magnesium-boron (MgB), titanium (Ti), vanadium (V), and/or a combination thereof).

[0066] The pinned layer L3 may have a magnetization direction fixed by an anti-ferromagnetic layer. The pinned layer L3 may include a ferromagnetic material (for example, CoFeB, FeB, Fe, Co, Ni, Gd, Dy, CoFe, NiFe, MnAs, MnBi, MnSb, CrO₂, MnOFe₂O₃, FeOFe₂O₃, NiOFe₂O₃, CuOFe₂O₃, MgOFe₂O₃, EuO, Y₃Fe₅O₁₂ and/or a combination thereof) and may further include an anti-ferromagnetic layer and/or a synthetic anti-ferromagnetic layer to fix the magnetization direction. The anti-ferromagnetic layer may include an anti-ferromagnetic material (for example, PtMn, IrMn, MnO, MnS, MnTe, MnF₂, FeC_l2, FeO, COC_l2, CoO, NiC_l2, NiO, Cr, and/or a combination thereof). The synthetic anti-ferromagnetic layer may include Cu, Ru, Ir, and/or a combination thereof.

[0067] FIG. 5A illustrates an example of a neural network (e.g., a neural network including a first layer having three nodes and a second layer having two nodes). FIG. 5B illustrates an example of a circuit diagram of a processing device (e.g., a processing device 100 implemented using a bitcell (e.g., the bitcell BC of FIG. 3)) configured to perform an operation of a neural network (e.g., configured to perform an operation of the second layer of the neural network of FIG. 5A).

[0068] Referring to FIG. 5A, in a first node a₁ of a second layer L2 (as in Equation 2 below, for example), a MAC operation of performing three multiplications of multiplying inputs x₁, x₂, and x₃ provided by the nodes of a first layer L1 respectively by weights w₁₁, w₂₁, and w₃₁ and summation of multiplication result values may be performed. In a second node a₂ of the second layer L2 (as in Equation 3 below, for example), a MAC operation of performing three multiplications of multiplying the inputs x₁, x₂, and x₃ provided by the nodes of the first layer L1 by weights w₁₂, w₂₂, and w₃₂ and summation of multiplication result values may be performed.

[0069] The processing device 100 of FIG. 5B may include several bitcells BC of FIG. 3 configured to perform an operation of the second layer L2 of FIG. 5A.

[0070] The processing device 100 of FIG. 5B includes a first bitcell line BCL₁ including three bitcells BC₁₁, BC₁₂, and BC₁₃ serially connected to one another, and a second bitcell line BCL₂ including three bitcells BC₂₁, BC₂₂, and BC₂₃ serially connected to one another. In the first bitcell line BCL₁, an operation of Equation 2 of the first node a₁ of FIG. 5A may be performed. In the second bitcell line BCL₂, an operation of Equation 3 of the second node a₂ of FIG. 5A may be performed.

[0071] As a current I of a certain value is applied to each of the bitcell lines BCL₁ and BCL₂, a sum of products of weights and inputs respectively applied to the bitcells BC₁₁, BC₁₂, BC₁₃, BC₂₁, BC₂₂, and BC₂₃ may be derived from a sum of voltage drops respectively occurring in the bitcells BC₁₁, BC₁₂, BC₁₃, BC₂₁, BC₂₂, and BC₂₃. Inputs and weights will be described below by illustrating a case such as Table 2 below, for example.

Table 2:

Input	Weight (first node a1)	Weight (second node a2)
X1 = 1	W11 = 1	W12 = -1
X2 = 1	W21 = -1	W22 = -1
X3 = -1	W31 = -1	W32 = 1

[0072] The resistance values of variable resistors R_11a, R_11b, R_12a, R_12b,..., R_23a, and R_23b included in the processing device 100 may be set by the weights of Table 2, as in Table 3 below, for example. Opening and closing states of switches S_11a, S_11b, S_12a, S_12b, ..., S_23a, and S_23b serially connected to the variable resistors R_11a, ..., and R_23b may be set by the inputs of Table 2, as in Table 4 below, for example.

Table 3:

First bitcell line BCL1 (Ω)		Second bitcell line BCL2 (Ω)
R11a = 20	R11b = 5	R21a = 5	R21b = 20
R12a = 5	R12b = 20	R22a = 5	R22b = 20
R13a = 5	R13b = 20	R23a = 20	R23b = 5

Table 4:

First bitcell line BCL1		Second bitcell line BCL2
S11a = Close	S11b = Open	S21a = Close	S21b = Open
S12a = Close	S12b = Open	S22a = Close	S22b = Open
S13a = Open	S13b = Close	S23a = Open	S23b = Close

[0073] When a current of 1A is provided to the first and second bitcell lines BCL₁ and BCL₂ set as in Table 3 and Table 4, a voltage drop may occur while the current is flowing toward a variable resistor for which a switch is closed from among the variable resistors included in each bitcell. At this time, when voltages (e.g,, V_T1 and V_T2) are measured from upper ends of the first and second bitcell lines BCL₁ and BCL₂, the measured voltages may correspond to sums of the voltage drops occurring in the bitcells. Voltage drops occurring in the bitcells BC₁₁, BC₁₂, and BC₁₃ included in the first bitcell line BCL₁ may be expressed as in Table 5 below, for example, and voltage drops occurring in the bitcells BC₂₁, BC₂₂, and BC₂₃ included in the second bitcell line BCL₂ may be expressed as in Table 6 below, for example.

Table 5:

	Input	Weight	Input * Weight	Resistance (Ω)	Current (A)	Voltage (V)
First bitcell BC11	1	1	1	20	1	20
Second bitcell BC12	1	-1	-1	5	1	5
Third bitcell BC13	-1	-1	1	20	1	20
First bitcell line BCL1	n/a	n/a	n/a	45	1	45

Table 6:

	Input	Weight	Input * Weight	Resistance (Ω)	Current (A)	Voltage (V)
First bitcell BC21	1	-1	-1	5	1	5
Second bitcell BC22	1	-1	-1	5	1	5
Third bitcell BC23	-1	1	-1	5	1	5
Second bitcell line BCL2	n/a	n/a	n/a	15	1	15

[0074] A voltage drop of 45V in the first bitcell line BCL₁ and a voltage drop of 15V in the second bitcell line BCL₂ may represent MAC operation results of weights and inputs applied to the bitcells included in the bitcell lines BCL₁ and BCL₂, respectively. A relationship between a voltage drop and a MAC operation result may be illustrated as in Table 7 below, for example.

Table 7:

Voltage drop (V)	Sum of products of inputs and weights
15	-3
30	-1
45	1
60	3

[0075] Referring to Table 7, a voltage drop in the first bitcell line BCL₁ was measured to be 45V, and thus, a sum of products of an input and a weight applied to the first bitcell line BCL₁ may be '1'. A voltage drop in the second bitcell line BCL₂ was measured to be 15V, and thus a sum of products of an input and a weight applied to the second bitcell line BCL₂ may be '-3'.

[0076] Although each of the bitcell lines BCL₁ and BCL₂ includes three bitcells in FIG. 5B, various numbers of bitcells equal to or greater than one or more bitcells may be included in a bitcell line. For example, a number of bitcells ranging from 64 to 256 may be included in a bitcell line. In an example, the processing device 100 may include 64 or more serially connected bitcells.

[0077] Although the processing device 100 of FIG. 5B includes only two bitcell lines, the processing device 100 may include various numbers of bitcell lines, namely, one or more bit lines. For example, the processing device 100 may include a number of bitcell lines ranging between 64 and 128. In an example, the bitcells of the processing device 100 may form a bitcell array including 64 or more bitcell lines.

[0078] A configuration of a plurality of bitcell lines each including a plurality of bitcells may be referred to as a bitcell array.

[0079] The processing device 100 of FIG. 5B may be an in-memory processing unit that stores values corresponding to weights in a memory device including variable resistors and performs an operation by using the stored values. In contrast with a Von Neumann structure where a memory and an operation part are separated, the in-memory processing unit may provide fast data transmission and reduced power consumption.

[0080] A processing device that applies a constant current I to serially-connected bitcells and derives an operation result corresponding to a sum of products from a voltage drop occurring in a bitcell line has been described above with reference to the embodiment of FIG. 5B. However, the structure of the bitcell BC of FIG. 3 is also applicable to different types of processing devices.

[0081] FIG. 6 illustrates an example of a circuit diagram of a processing device (e.g., a processing device 200) configured to perform an operation of a neural network (e.g., the neural network of FIG. 5A). Referring to FIG. 6, in contrast with FIG. 5B, capacitors BCL_C1 and BCL_C2 may be provided on respective lower ends of the bitcell lines BCL₁ and BCL₂, and be serially connected to bitcells of the bitcell line BCL₁ and bitcells of the bitcell line BCL₂, respectively. In other words, according to an embodiment of FIG. 6, after a voltage V is applied to each of the bitcell lines BCL₁ and BCL₂, the processing device 200 may be implemented using a Time to Digital Converter (TDC) method which is a method of measuring a time period during which the voltage of each of the capacitors BCL_C1 and BCL_C2 rises to a specific value.

[0082] FIGS. 7 through 15 illustrate examples explaining layouts of a processing device (e.g., either of processing devices 100 or 200) implemented using the bitcell BC of FIG. 3. Examples of the structure of such processing device will be further described below with reference to FIGS. 7 through 15.

[0083] FIG. 7 illustrates an example of a connection structure between variable resistors and switches in a bitcell included in a processing device. By taking the first bitcell line BCL₁ of FIG. 5B as an example, FIG. 7 illustrates a cross-sectional view of a processing device including an active layer, a variable resistor layer, and metal layers.

[0084] Referring to FIG. 7, an active layer L100 including the switches S_11a, S_12a, and S_13b formed therein, first through sixth metal layers M100 through M600, and a variable resistor layer L200 may be sequentially stacked. A relationship between the positions of the active layer L100 and the variable resistor layer L200 may depend on a process sequence in which the active layer L100 is first formed and metal wires and the variable resistors R_11a, R_12a, R_13b, etc. are manufactured on the active layer L100.

[0085] The active layer L100 may be a layer including the switches S_11a, S_12a, S_13b, etc. of FIG. 5B formed therein, and thus may be a layer in which the switches S_11a, S_12a, S_13b, etc. are formed on a silicon wafer through a transistor formation process or the like. For convenience of explanation, the active layer L100 of FIG. 7 shows the switches S_11a, S_12a, and S_13b existing on the path of the current I flowing in the first bitcell line BCL₁ due to the inputs and weights of Table 2 from among the switches S_11a, S_11b, S_12a, S_12b, S_13a, S_13b, S₁, S₁₁, S₁₂, and S₁₃ included in the first bitcell line BCL₁ of FIG. 5B. The remaining switches S_11b, S_12b, S_13a, S₁, S₁₁, S₁₂, and S₁₃ of FIG. 5B are in an open state, and thus the current I applied to the first bitcell line BCL₁ does not flow in the remaining switches S_11b, S_12b, S_13a, S₁, S₁₁, S₁₂, and S₁₃, and thus the remaining switches S_11b, S_12b, S_13a, S₁, S₁₁, S₁₂, and S₁₃ are not shown in FIG. 7. However, the remaining switches S_11b, S_12b, S_13a, S₁, S₁₁, S₁₂, and S₁₃ of FIG. 5B not shown in FIG. 7 may be formed in the active layer L100 and may be arranged on a horizontally different location from the switches S_11a, S_12a, and S_13b. In addition, the switches S_21a, S_21b, S_22a, S_22b, S_23a, S_23b, S₂, S₂₁, S₂₂, and S₂₃ included in the second bitcell line BCL₂ of FIG. 5B may be arranged similarly to the first bitcell line BCL₁.

[0086] The variable resistor layer L200 may be a layer including the variable resistors R_11a, R_12a, R_13b, etc. and vias V₁₁, V₁₂, and V₁₃ formed therein, a seventh metal layer M700 may be located above the variable resistor layer L200, and a sixth metal layer M600 may be located below the variable resistor layer L200.

[0087] The variable resistors R_11a, R_12a, R_13b, etc. may be formed in a vertical direction in the variable resistor layer L200 such that one end of both ends of each variable resistor is connected to the seventh metal layer M700 and the other end is connected to the sixth metal layer M600. Hereinafter, respective ends of the vertically-formed variable resistors R_11a, R_12a, R_13b, etc. on the side of the seventh metal layer M700 may be referred to as upper end portions of the variable resistors, and respective ends thereof on the side of the sixth metal layer M600 may be referred to as lower end portions of the variable resistors. A current flowing through the variable resistors R_11a, R_12a, R_13b, etc. may flow in a direction from the upper end portion of each variable resistor to the lower end portion thereof or from the lower end portion to the upper end portion.

[0088] The variable resistor layer L200 of FIG. 7 is illustrated as showing the variable resistors R_11a, R_12a, and R_13b existing on the path of a current flowing due to the inputs and weights of Table 2 from among the variable resistors R_11a, R_11b, R_12a, R_12b, R_13a, and R_13b included in the first bitcell line BCL₁ of FIG. 5B. However, although not shown in FIG. 7, the remaining variable resistors R_11b, R_12b, and R_13a may also formed in the variable resistor layer L200 and may be arranged at a horizontally different location from the variable resistors R_11a, R_12a, and R_13b.

[0089] The vias V₁₁, V₁₂, and V₁₃ may penetrate through the variable resistor layer L200 and electrically connect the seventh metal layer M700 to the sixth metal layer M600. The variable resistors (e.g.,R_11a, R_12a, and R_13b) existing on the path of a current flowing due to the inputs and weights of Table 2 from among the variable resistors of the bitcells may be serially connected to one another through the vias V₁₁, V₁₂, and V₁₃, and an example connection structure thereof will be further described later in more detail.

[0090] The first through seventh metal layers M100 through M700 may be layers on which conductive wires (such as wires for connecting the variable resistors R_11a, R_12a, R_13b, etc. to the switches S_11a, S_12a, S_13b, etc., and wires for transmitting an ON/OFF signal to the switches S_11a, S_12a, S_13b, etc.) are formed.

[0091] When describing the connection relationship between the switches S_11a, S_12a, and S_13b and the variable resistors R_11a, R_12a, and R_13b, shown in FIG. 7, different variable resistors may be electrically connected to both ends of each of the switches S_11a and S_12a. However, in an example, different variable resistors may not be electrically connected to both ends of the switch S_13b, as the switch S_13b may be included in the third bitcell BC₁₃ arranged last in the first bitcell line BCL₁ (on the lowest end of the first bitcell line BCL₁) as shown in FIG. 5B.

[0092] When illustrating and describing the first switch S_12a of the second bitcell BC₁₂, one end of the switch S_12a is connected to the first variable resistor R_12a of the second bitcell BC₁₂ and the other end of the switch S_12a is connected to the second variable resistor R_13b of the third bitcell BC₁₃. For example, one end of the switch S_12a is connected to the lower end portion of the variable resistor R_12a, and the other end of the switch S_12a is connected to the upper end portion of the variable resistor R_13b through the via V₁₂. As described above, because the active layer L100 is located below the variable resistor layer L200, the via V₁₂ of the second bitcell BC₁₂ penetrating through the variable resistor layer L200 may be used to connect the switch S_12a to the upper end portion of the variable resistor R_13b. Referring to FIG. 5B, a location where the switch S_12a is connected to the upper end portion of the variable resistor R_13b via the via V₁₂ of the second bitcell BC₁₂ may be an 'A' region which is a connection between the second bitcell BC₁₂ and the third bitcell BC₁₃. In other words, the via V₁₂ may be used to connect any switch of the second bitcell BC₁₂ to any variable resistor of the third bitcell BC₁₃. Similarly, the via V₁₁ of the first bitcell BC₁₁ may electrically connect any switch of the first bitcell BC₁₁ to any variable resistor of the second bitcell BC₁₂. In other words, due to such a structure of vias, all of the variable resistors of each bitcell may be serially connected to one another.

[0093] FIG. 8 illustrates an example of a horizontally cross-sectional view illustrating a variable resistor layer (e.g., the variable resistor layer L200 of FIG. 7).

[0094] Referring to FIG. 8, the variable resistor layer L200 may include four bitcell lines each including three bitcells. FIG. 5B illustrates the first and second bitcell lines BCL₁ and BCL₂. Compared with FIG. 5B, FIG. 8 illustrates four bitcell lines further including third and fourth bitcell lines BCL₃ and BCL₄ in addition to the first and second bitcell lines BCL₁ and BCL₂ in order to show a configuration of variable resistors and vias. Because the first bitcell line BCL₁ and the third bitcell line BCL₃ have the same structures and the second bitcell line BCL₂ and the fourth bitcell line BCL₄ have the same structures, the variable resistors and the vias of the third and fourth bitcell lines BCL₃ and BCL₄ are not indicated by reference numerals.

[0095] Referring to FIG. 8, each bitcell may further include an inactive variable resistor in addition to variable resistors and a via. For example, the first bitcell BC₁₁ of the first bitcell line BCL₁ may further include nine inactive variable resistors R_i in addition to the variable resistors R_11a and R_11b and the via V₁₁ of FIG. 5B.

[0096] The inactive variable resistors R_i may be variable resistors not used in processing in contrast with the above-described variable resistors R_11a and R_11b, and may not be electrically connected to the switches S_11a, S_11b, etc. of FIG. 5B. The variable resistors R_11a, R_11b, etc. used in processing may be referred to as active variable resistors to be distinguished from the inactive variable resistors R_i. The inactive variable resistors R_i may be formed to obtain the active variable resistors R_11a, R_11b, etc. having uniform electrical characteristics. For example, due to a manufacturing process of the variable resistor layer L200 and/or the processing device, the variable resistors formed to be surrounded by other variable resistors (e.g., near a center of an array of the variable resistors) may exhibit relatively uniform electrical characteristics compared with the variable resistors formed on an edge of the array, and thus only the variable resistors having good reliability (e.g., due to the relatively uniform electrical characteristics) and being located on the center may be used in processing, and the remaining variable resistors may be inactive variable resistor which are not used in processing.

[0097] Examples of the variable resistors included in the four bitcells BC₁₁, BC₁₂, BC₂₁, and BC₂₂ of FIG. 8 will now be illustrated and described. Eleven variable resistors may be formed in each bitcell, and the four bitcells BC₁₁, BC₁₂, BC₂₁, and BC₂₂ may include a total of 44 variable resistors formed therein. Active variable resistors may be located at the center of a group of bitcells (e.g., the four bitcells BC₁₁, BC₁₂, BC₂₁, and BC₂₂) and thus may be surrounded by other variable resistors. In other words, in the four bitcells BC₁₁, BC₁₂, BC₂₁, and BC₂₂, the eight variable resistors R_11a, R_11b, R_12a, R_12b, R_21a, R_21b, R_22a, and R_22b surrounded by 36 inactive variable resistors R_i may correspond to eight active variable resistors.

[0098] Referring to FIG. 8, each of the vias V₁₁ and V₁₂ may be formed in the variable resistor layer L200 to have a structure of penetrating through the variable resistor layer L200, and the vias V₁₁ or V₁₂ may be used to connect a switch of a bitcell including the vias V₁₁ or V₁₂ to the upper end portion of any variable resistor of a bitcell adjacent the bitcell. The vias V₁₁, V₁₂, etc. may be arranged to be sufficiently spaced apart (e.g., spaced apart by a predetermined distance) from variable resistors. For example, in the first bitcell BC₁₁ of the first bitcell line BCL₁ of FIG. 8, the active and inactive variable resistors R_11a, R_11b, and R_i may be arranged near the right lower corner, and the via V₁₁ may be arranged near the left upper corner which is sufficiently spaced apart from the variable resistors R_11a, R_11b, and R_i. A distance d₁ may be a minimum distance between the via of a bitcell and the active variable resistors of the bitcell (e.g., the distance d₁ may be a minimum distance between the via V₁₁ of the first bitcell BC₁₁ and the active variable resistors R_11a and R_11b of the first bitcell BC₁₁, and may be a minimum distance between the via V₁₂ of the second bitcell BC₁₂ and the active variable resistors R_12a and R_12b of the second bitcell BC₁₂), and a distance d₂ may be a minimum distance between the via of the bitcell and the inactive variable resistors of the bitcell.

[0099] According to the above-described configuration of FIG. 8, the minimum distance d₁ between the via V₁₁, V₁₂, etc. and the active variable resistors R_11a, R_11b, R_12a, R_12b, etc. of a bitcell (e.g., the minimum distance between the via V₁₁ and the active variable resistors R_11a, R_11b, etc of the bitcell BC₁₁ or the minimum distance between the via V₁₂ and the active variable resistors R_12a, R_12b, etc of the bitcell BC₁₂) may be greater than the minimum distance d₂ between the via V₁₁, V₁₂, etc. and the inactive variable resistors R_i of the bitcell (e.g., the minimum distance between the via V₁₁ and the inactive variable resistors R_i of the bitcell BC₁₁ or the minimum distance between the via V₁₂ and the inactive variable resistors R_i of the bitcell BC₁₂). The minimum distance d₁ between the via V₁₁, V₁₂, etc. and the active variable resistors R_11a, R_11b, R_12a, R_12b, etc. may be, for example, 0.50 µm to 1.20 µm, and the minimum distance d₂ between the vias V₁₁, V₁₂, etc. and the inactive variable resistors R_i may be, for example, 0.30 µm to 0.60 µm.

[0100] A minimum distance d₃ between variable resistors of a bitcell (e.g., a distance any of the variable resistors R_11a, R_11b, R_12a, R_12b, and R_i and a closest variable resistor) may be less than the minimum distance d₁ between the via V₁₁, V₁₂, etc. of the bitcell and the active variable resistors R_11a, R_11b, R_12a, R_12b, etc. of the bitcell or the minimum distance d₂ between the via V₁₁, V₁₂, etc. of the bitcell and the inactive variable resistors R_i. of the bitcell, and may be, for example, 0.10 µm to 0.40 µm. A minimum distance d₄ between a via V₁₁, V₁₂, etc. of a bitcell and a closest adjacent via of an adjacent bitcell may be less than the minimum distance d₁ between the via V₁₁, V₁₂, etc. of the bitcell and the active variable resistors R_11a, R_11b, R_12a, R_12b, etc. of the bitcell and the minimum distance d₂ between the via V₁₁, V₁₂, etc. of the bitcell and the inactive variable resistors R_i of the bitcell, and may be, for example, 0.10 µm to 0.40 µm.

[0101] Respective configurations of variable resistors and/or vias in neighboring (or adjacent) bitcells may be vertically and/or bilaterally symmetrical to each other about a boundary line between the neighboring (or adjacent) bitcells. For example, a configuration of the variable resistors R_11a, R_11b, and R_i and the via V₁₁ included in the first bitcell BC₁₁ may be vertically symmetrical to that of the variable resistors R_12a, R_12b, and R_i and the via V₁₂ included in the second bitcell BC₁₂, based on a boundary line between the bitcells BC₁₁ and BC₁₂ adjacent to each other. Similarly, the configuration of the variable resistors R_11a, R_11b, and R_i and the via V₁₁ included in the first bitcell BC₁₁ of the first bitcell line BCL₁ may be bilaterally symmetrical to a configuration of the variable resistors R_21a, R_21b, and R_i and a via included in the first bitcell BC₂₁ of the second bitcell line BCL₂, based on a boundary line between the bitcells BC₁₁ and BC₂₁ adjacent to each other.

[0102] FIG. 9A illustrates an example of a vertical cross-sectional view of bitcells (e.g., a vertical cross-sectional view of the bitcells BC₁₁ and BC₁₂ taken along line X-X' of FIG. 8), and FIG. 9B illustrates an example of a vertical cross-sectional view of bitcells (e.g., a vertical cross-sectional view of the bitcells BC₁₁ and BC₂₁ taken along line Y-Y' of FIG. 8).

[0103] Referring to FIGS. 9A and 9B, the processing device 100 may include an active layer L100 in which a plurality of switches S_11a, S_11b, etc. are formed, a variable resistor layer L200 in which variable resistors R_11a, R_11b, etc. and vias V₁₁, V₁₂, etc. are formed, and a plurality of metal layers M100, M200, M300, M400, M500, M600, and M700.

[0104] The active layer L100 is a layer in which the switches (for example, S_11a, S_11b, etc. of FIG. 5B) included in bitcells (for example, BC₁₁, BC₁₂, BC₂₁, and BC₂₂ of FIG. 5B) are formed. Although FIG. 9A illustrates some switches such as switches S_11a and S_11b in the cross-section taken along line X-X', other not-shown switches may be formed in the active layer L100. The active layer L100 may be a silicon-based semiconductor layer. A detailed structure of the switches S_11a, S_11b, etc. will be further described later with reference to FIG. 10.

[0105] The plurality of metal layers M100, M200, M300, M400, M500, M600, and M700 may include wires for connecting the active variable resistors R_11a, R_11b, etc. to the switches S_11a, S_11b, etc., wires for transmitting an ON/OFF signal to the switches S_11a, S_11b, etc., and bit-data lines. For example, a lower end portion of the active variable resistor R_11a may be connected to a drain S_11aD of the switch S_11a, through vias that connect the metal layers M100, M200, M300, M400, M500, and M600 to one another. Although a path along which the active variable resistor R_11a is connected to the drain S_11aD is not shown in both FIGS. 9A and 9B, the path may be provided through various designs. For example, the third metal layer M300 may include wires for transmitting an ON/OFF signal to the switches S_11a, S_11b, etc., and the fourth metal layer M400 may include the bit-data lines BDL_1a, BDL_1b, BDL_2a and BDL_2b of FIGS. 5B and/or 6. The seventh metal layer M700 may include, for example, wires for respectively connecting the vias V₁₁ and V₁₂ (e.g., upper end portions of the vias V₁₁ and V₁₂), to the upper end portions of the active variable resistors R_12a and R_13b.

[0106] FIG. 10 illustrates an example of a plan view of an active layer (e.g., the active layer L100 of FIG. 9A).

[0107] FIG. 10 illustrates an example of the switches S_11a, S_11b, S₁₁, etc. included in the bitcells BC₁₁, BC₁₂, BC₂₁, and BC₂₂ of FIG. 5B. An example structure of the active layer L100 will now be described in detail with respect to the example of FIG. 10 showing the three switches S_11a, S_11b, and S₁₁ included in the first bitcell BC₁₁ of the first bitcell line BCL₁.

[0108] The switches S_11a, S_11b, and S₁₁ included in the active layer L100 may be Metal Oxide Semiconductor Field Effect Transistors (MOSFETs) each including a source, a drain, and a gate. A switch implemented by a MOSFET may operate in such a way that, when a gate voltage (equal to or greater than a threshold voltage) is applied to a gate, the switch may be turned on and thus a source and a drain may be electrically connected to each other, and, when no gate voltages are applied to the gate, the switch may be turned off and thus the source and the drain may be electrically disconnected from each other. For example, when a gate voltage is applied to a gate S_11aG of the first switch S_11a, the first switch S_11a may be turned on and thus a source S_11CS1 of the first switch S_11a may be electrically connected to a drain S_11aD of the first switch S_11a. For example, when a gate voltage is applied to a gate S_11bG of the second switch S_11b, the second switch S_11b may be turned on and thus a source S_11CS1 of the second switch S_11b may be electrically connected to a drain S_11bD of the second switch S_11b. In FIG. 10, the first switch S_11a and the second switch S_11b may be implemented to have a structure of commonly sharing the source S_11CS1. The bit-data line switch S₁₁ may have a structure including two gates S_11Ga and S_11Gb, two drains S_11Da and S_11Db, and a common source S_11CS2. The bit-data line switch S₁₁ may operate as a single transistor by using two transistors that receive the same signal. The switches S_11a, S_11b, S₁₁,, etc. of the active layer L100 may be connected to the first metal layer M100 in marked regions L110. In an embodiment, the switches (e.g., S_11a, S_11b, and S₁₁) of the bitcell (e.g., the bitcells BC₁₁) may control either one of a voltage to be applied between two ends of each of the active variable resistors (e.g., R_11a and R_11b) and a current flowing to each of the active variable resistors.

[0109] FIG. 11 illustrates an example of a plan view of a seventh metal layer of a processing device (e.g., the seventh metal layer M700 of FIG. 9A).

[0110] Referring to FIG. 11, the seventh metal layer M700 may be stacked above the variable resistor layer L200 and may provide wires for connecting the vias V₁₁, V₁₂, etc. to the upper end portions of the active variable resistors R_12a, R_13b, etc. For example, the via V₁₁ of the first bitcell BC₁₁ may be connected to the upper end portions of the active variable resistors R_12a and R_12b of the second bitcell BC₁₂ through a wire M750 of the seventh metal layer M700.

[0111] Referring to FIG. 11 together with FIG. 7, in an example, the current I flowing from respective upper end portions of the active variable resistors R_11a and R_11b of the first bitcell BC₁₁ to respective lower end portions thereof may flow down to the active layer L100 through the metal layers M100 through M600 and may respectively pass through the switches S_11a and S_11b. Then, the current I may again flow to the upper end portion of each of the active variable resistors R_12a and R_12b of the second bitcell BC₁₂ through the metal layers M100 through M600 and the via V₁₁ and then through the wire M750 of the seventh metal layer M700.

[0112] The metal layers M100, M200, M500, and M600 may provide wires used for electrical connection between the above-described components. An insulation layer may be arranged between the metal layers M100 through M600.

[0113] In the above-described example embodiments, a structure in which the variable resistor layer L200 includes one via V₁₁, or V₁₂, or the like for one bitcell has been illustrated and described. However, the number of vias included in a bitcell line may be greater than or smaller than the number of bitcells. For example, one via may be included in every two neighboring (or adjacent) bitcells within any bitcell line of the structure.

[0114] FIG. 12 illustrates an example of a circuit diagram of a processing device (e.g., a processing device 300) having a connection structure of a switch and a variable resistor that is different from that of another processing device (e.g., including a bitcell having a different structure from that of FIG. 5B).

[0115] Referring to FIG. 12, the processing device 300 includes bitcells BC₁₁ and BC₁₃ each having a structure in which the current I flowing in the bitcell line BCL₁ flows through a switch (for example, S_11a or S_13b) prior to flowing through a variable resistor (for example, R_11a or R_13b, respectively).

[0116] When the structure of the first bitcell BC₁₁ of FIG. 12 is compared with the first bitcell BC₁₁ of FIG. 5B, the current I flowing in the first bitcell BC₁₁ of FIG. 5B flows to the first switch S_11a through the first variable resistor R_11a, whereas the current I flowing in the first bitcell BC₁₁ of FIG. 12 flows to the first variable resistor R_11a through the first switch S_11a. In other words, in the structure of the first bitcell BC₁₁ of FIG. 12 and the structure of the first bitcell BC₁₁ of FIG. 5B, the locations of the switches S_11a and S_11b are opposite to those of the variable resistors R_11a and R_11b, based on the direction of the current I.

[0117] In the processing device 300, a bitcell in which the locations of the switches S_11a and S_11b and the variable resistors R_11a and R_11b are reversed like the first bitcell BC₁₁ of FIG. 12 may alternate with a bitcell in which the locations of the switches S_12a and S_12b and the variable resistors R_12a and R_12b are not reversed like the second bitcell BC₁₂ of FIG. 5B. In other words, in the processing device 300, a second bitcell BC₁₂ in which the locations of switches and variable resistors are not reversed may be arranged between the first and third bitcells BC₁₁ and BC₁₃ in which the locations of switches and variable resistors are reversed.

[0118] According to the circuit structure of the processing device 300 of FIG. 12, the current I may also flow in the bitcell line BCL₁ without passing through a via formed in the variable resistor layer L200 like the vias V₁₁, V₁₂, etc. of the variable resistor layer L200. A connection relationship between a variable resistor and a switch related to the processing device 300 will now be described in more detail with reference to FIG. 13.

[0119] FIG. 13 illustrates an example of a connection relationship between a variable resistor and a switch (e.g., cross-sections of an active layer, a variable resistor layer, and metal layers of the processing device 300 of FIG. 12).

[0120] Referring to FIG. 13, the current I flowing in the first bitcell line BCL₁ may pass the first variable resistor R_11a through the first switch S_11a within the first bitcell BC₁₁. One end (upper end portion) of the first variable resistor R_11a of the first bitcell BC₁₁ may be connected to one end (upper end portion) of the first variable resistor R_12a of the second bitcell BC₁₂ through the seventh metal layer M700. Accordingly, in contrast with FIG. 7, the current I having passed through the first variable resistor R_11a may flow to the first variable resistor R_12a of the second bitcell BC₁₂ without passing through any switch. The bit-data line switches S₁, S₁₁, S₁₂, and S₁₃ may be closed only when a weight is applied, and, in other cases (e.g., when the weight is not applied), may be maintained in an open state. Accordingly, after a weight is applied, the current I may not flow from the bit-data lines BDL_1a and BDL_1b to the first bitcell line BCL₁. The current I that passed through the first variable resistor R_12a of the second bitcell BC₁₂ may flow to the second variable resistor R_13b of the third bitcell BC₁₃ through the switches S_12a and S_13b. As described above, in FIG. 13, the current I flowing in the first bitcell line BCL₁ may flow without passing through the vias V₁₁ and V₁₃ formed in the variable resistor layer L200.

[0121] The vias V₁₁ and V₁₃ of FIG. 13 may provide a path for changing the resistance values of the variable resistors R_11a, R_12a, etc. For example, the first via V₁₁ may be used while the resistance value of the second variable resistor R_12a of FIG. 13 is being changed. An example method of changing the resistance value of a variable resistor when a weight is applied to a bitcell has been described above in relation to the bitcell BC of FIG. 3. Referring back to FIG. 12, to change the resistance value of the second variable resistor R_12a, both ends of the second variable resistor R_12a may be connected to the first and second bit-data lines BDL_1a and BDL_1b. For example, the upper end portion of the second variable resistor R_12a may be connected to the second bit-data line BDL_1b through the first bit-data line switch S₁₁, and the lower end portion of the second variable resistor R_12a may be connected to the first bit-data line BDL_1a though the first switch S_12a and the second bit-data line switch S₁₂ of the second bitcell BC₁₂. Referring to FIG. 13, the upper end portion of the second variable resistor R_12a may be connected to the first bit-data line switch S₁₁ through the first via V₁₁. In other words, it may be seen from FIG. 13 that the vias V₁₁ or V₁₃ may provide a path that connects any variable resistor, of a bitcell ncluding the via V₁₁ or V₁₃ to corresponding bit-data line switch among bit-data line switches corresponding to the bitcell.

[0122] When the number of vias formed in the variable resistor layer L200 is compared between each of the bitcells of FIG. 13 and each of the bitcells of FIG. 7, FIG. 7 illustrates that each of the bitcells includes one via, whereas FIG. 13 illustrates that every two bitcells include one via. In detail, the first bitcell BC₁₁ and the second bitcell BC₁₂ of FIG. 13 may include only one via V₁₁. Although the bitcell line BCL₁ of FIG. 13 is illustrated as having a structure in which a via is formed in an odd-numbered bitcell like the first and third bitcells BC₁₁ and BC₁₃, the bitcell line BCL₁ of FIG. 13 may have a structure including a via formed in the variable resistor layer of an even-numbered bitcell.

[0123] FIG. 14 illustrates an example of a processing device (e.g., a processing device 700).

[0124] Referring to FIG. 14, the processing device 700 may include a bitcell array 710, a controller 720, a row decoder 730, a column decoder 740, a weight driver 750, a current source controller 760, a data buffer 770, and a voltage meter 780. The processing device 700 illustrated in FIG. 14 may include constituent elements related to the above-described embodiments. However, the disclosure is not limited thereto, and the processing device 700 may further include general-purpose constituent elements other than the constituent elements illustrated in FIG. 14.

[0125] The controller 720 may decode instructions for driving and operation of the processing device 700. For example, the controller 720 may decode instructions such as weight setting, weight setting test, input application, voltage measurement, etc., and transmits signals to elements for executing these instructions.

[0126] The bitcell array 710 may be an array of bitcells including the above-described variable resistors and switches. The variable resistor may be an MTJ device having a magnetic material.

[0127] The row decoder 730 may receive a row address and an input signal and apply an input value to the bitcell array 710. The row decoder 730 may include a digital-to-analog converter (DAC) or an analog-to-digital converter (ADC), and may apply a driving voltage to the switch serially connected to the variable resistor, on the basis of the input value. Furthermore, the row decoder 730 may change the resistance value of a variable resistor included in the bitcell of the bitcell array 710. In this state, the row decoder 730 may apply a driving voltage to related switches so that a target variable resistor is selected.

[0128] The column decoder 740 may receive a column address and a weight setting signal and apply a voltage/current to the variable resistor. The column decoder 740 may select a bitcell line for voltage measurement, and a weight line connected to the bitcell for weight setting.

[0129] The weight driver 750, during weight setting, may transmit weight data to a bitcell selected by the row decoder 730 and the column decoder 740. The weight driver 750 may drive the weight line connected to the column decoder 740, based on the data received from the data buffer 770, and perform setting of a weight and a test of the set weight. The weight driver 750 may include a current source for applying a test current to the weight line, in order to test whether a desired resistance value is set to the variable resistor.

[0130] The current source controller 760 may receive signals from the controller 720 to drive the current source, and apply a current to the bitcell line.

[0131] The voltage meter 780 may measure the voltage of the resistor or a capacitor connected to one terminal of the bitcell line, and store a measurement value in an external memory (not shown). The voltage meter 780 may include an ADC that outputs a measurement value as a digital value. The processing device 700 may be or include any of the processing devices mentioned above, such as the processing device 100, 200, or 300.

[0132] FIG. 15 illustrates an example of an electronic device (e.g., an electronic device 800).

[0133] Referring to FIG. 15, the electronic device 800 may extract valid information by analyzing input data on the basis of a neural network device 830 including (or as) the processing device, determine a situation on the basis of extracted information, or control elements of an electronic system equipped with the electronic device 800. For example, the electronic device 800 may be applied to in drones, robot apparatuses (such as advanced driver-assistance systems (ADASs)), smart televisions (TVs), smartphones, medical devices, mobile devices, image display devices, measuring devices, and loT devices, and may be mounted in at least one of various other kinds of electronic devices.

[0134] The electronic device 800 may include a processing unit 810 (e.g., one or more processors), a random access memory (RAM) 820, a memory 840, a sensor module 850, and a communication (Tx/Rx) module 860 in addition to a neural network device 830. The electronic device 800 may further include an input/output module, a security module, a power control device, etc. Some of the hardware components of the electronic device 800 may be mounted on a semiconductor chip. The neural network device 830 may be a device implemented as an on-chip type of the processing device according to the above-described embodiments, or a device including the processing device according to the above-described embodiments as a part.

[0135] The processing unit 810 may control overall operations of the electronic device 800. The processing unit 810 may be a central processing unit (CPU), and may include a single processor core or a plurality of processor cores. The processing unit 810 may process or execute programs and/or data stored in the memory 840. The processing unit 810 may control the function of the neural network device 830 by executing the programs stored in the memory 840. The processing unit 810 may be implemented by a graphic processing unit (GPU), an application processor (AP), or the like instead of a CPU.

[0136] The RAM 820 may temporarily store programs, data, or instructions. For example, the programs and/or data stored in the memory 840 may be temporarily stored in the RAM 820 under the control of the processing unit 810 or according to a boot code. The RAM 820 may be implemented by a memory device such as DRAM or SRAM.

[0137] The neural network device 830 may perform an operation of a neural network, based on the received input data, and may generate an information signal, based on a result of the operation. The neural network device 830 may include the processing device according to the above-described embodiments (e.g., any of the processing devices 100, 200, 300, or 700). The neural network may be, but is not limited to, a convolution neural network (CNN), a recurrent neural network (RNN), a deep belief network, or a restricted Boltzman machine. The neural network device 830 may correspond to a neural network dedicated hardware accelerator.

[0138] The information signal may include various kinds of recognition signals such as a voice recognition signal, an object recognition signal, an image recognition signal, and a bio-information recognition signal. For example, the neural network device 830 may receive frame data included in a video stream as input data, and may generate a recognition signal regarding an object included in an image represented by the frame data. The neural network device 830 may receive various kinds of input data according to the types or functions of electronic system in which the electronic device 800 is mounted, and may generate recognition signals according to the various kinds of input data.

[0139] The memory 840 is a storage for storing data, and may store, for example, an operating system (OS), various kinds of programs, and various kinds of data. The memory 840 may include a volatile and/or nonvolatile memory. Examples of the nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable and programmable ROM (EEPROM), flash memory, phase-change random access memory (PRAM), magnetic random access memory (MRAM), resistive random access memory (RRAM), and ferroelectric random access memory (FeRAM). Examples of the volatile memory may include DRAM, static random access memory (SRAM), synchronous DRAM (SDRAM), PRAM, MRAM, RRAM, and FeRAM. The memory 840 may include, for example, a hard disk drive (HDD), a solid state drive (SSD), a compact flash (CF), a secure digital (SD) card, a micro-secure digital (Micro-SD) card, a mini-secure digital (Mini-SD) card, an extreme digital (xD) card, or a memory stick.

[0140] The sensor module 850 may collect information about the surroundings of the electronic systems in which the electronic device 800 is mounted. The sensor module 850 may sense or receive a signal, such as an image signal, a voice signal, a magnetic signal, a biometric signal, or a touch signal, from outside the electronic device, and may convert the sensed or received signal to data. To this end, the sensor module 850 may include any of various types of sensing devices, such as a microphone, a photographing device, an image sensor, a light detection and ranging (LIDAR) sensor, an ultrasonic sensor, an infrared sensor, a biosensor, or a touch sensor.

[0141] The sensor module 850 may provide the neural network device 830 with the converted data as input data. For example, the sensor module 850 may include an image sensor, and may generate a video stream by photographing the external environment of the electronic device and provide the neural network device 830 with consecutive data frames of the video stream in order as input data. However, the sensor module 850 is not limited thereto, and the sensor module 850 may provide various other types of data to the neural network device 830.

[0142] The Tx/Rx module 860 may include various wired or wireless interfaces capable of communicating with external devices. For example, the Tx/Rx module 860 may include a local area network (LAN), a wireless local area network (WLAN) such as Wi-Fi, wireless fidelity (Wi-Fi), a wireless personal area network (WPAN) such as Bluetooth, a wireless universal serial bus (USB), ZigBee, near-field communication (NFC), radio-frequency identification (RFID), power-line communication (PLC), or a communication interface capable of connecting to a mobile cellular network such as 3rd generation (3G), 4th generation (4G), long-term evolution (LTE), or 5th Generation (5G).

[0143] The electronic device 800 may include a processor, a memory device for storing program data and executing it, a permanent storage such as a disk drive, a communications port for handling communications with external devices, and user interface devices, including a touch panel, keys, buttons, etc. For example, when software modules or algorithms are involved, these software modules may be stored as program instructions or computer readable codes executable on the processor in a computer-readable recording medium.

[0144] The processing devices, neural network devices, electronic devices, bitcells, bitcell lines, switches, variable resistors, layers, pinned layers, tunnel layers, free layers, active layers, metal layers, variable resistor layers, vias, bitcell arrays, controllers, row decoders, column decoders, weight drivers, current source controllers, data buffers, voltage meters, processing units, RAMs, memories, sensor modules, communication (Tx/Rx) modules, processing device 200, processing device 300, processing device 700, bitcell array 710, controller 720, row decoder 730, column decoder 740, weight driver 750, current source controller 760, data buffer 770, voltage meter 780, electronic device 800, processing unit 810, RAM 820, neural network device 830, memory 840, sensor module 850, communication (Tx/Rx) module 860, and other apparatuses, devices, units, modules, and components described herein with respect to FIGS. 1-15 are implemented by or representative of hardware components. Examples of hardware components that may be used to perform the operations described in this application where appropriate include controllers, sensors, generators, drivers, memories, comparators, arithmetic logic units, adders, subtractors, multipliers, dividers, integrators, and any other electronic components configured to perform the operations described in this application. In other examples, one or more of the hardware components that perform the operations described in this application are implemented by computing hardware, for example, by one or more processors or computers. A processor or computer may be implemented by one or more processing elements, such as an array of logic gates, a controller and an arithmetic logic unit, a digital signal processor, a microcomputer, a programmable logic controller, a field-programmable gate array, a programmable logic array, a microprocessor, or any other device or combination of devices that is configured to respond to and execute instructions in a defined manner to achieve a desired result. In one example, a processor or computer includes, or is connected to, one or more memories storing instructions or software that are executed by the processor or computer. Hardware components implemented by a processor or computer may execute instructions or software, such as an operating system (OS) and one or more software applications that run on the OS, to perform the operations described in this application. The hardware components may also access, manipulate, process, create, and store data in response to execution of the instructions or software. For simplicity, the singular term "processor" or "computer" may be used in the description of the examples described in this application, but in other examples multiple processors or computers may be used, or a processor or computer may include multiple processing elements, or multiple types of processing elements, or both. For example, a single hardware component or two or more hardware components may be implemented by a single processor, or two or more processors, or a processor and a controller. One or more hardware components may be implemented by one or more processors, or a processor and a controller, and one or more other hardware components may be implemented by one or more other processors, or another processor and another controller. One or more processors, or a processor and a controller, may implement a single hardware component, or two or more hardware components. A hardware component may have any one or more of different processing configurations, examples of which include a single processor, independent processors, parallel processors, single-instruction single-data (SISD) multiprocessing, single-instruction multiple-data (SIMD) multiprocessing, multiple-instruction single-data (MISD) multiprocessing, and multiple-instruction multiple-data (MIMD) multiprocessing.

[0145] The methods illustrated in FIGS. 1-15 that perform the operations described in this application are performed by computing hardware, for example, by one or more processors or computers, implemented as described above executing instructions or software to perform the operations described in this application that are performed by the methods. For example, a single operation or two or more operations may be performed by a single processor, or two or more processors, or a processor and a controller. One or more operations may be performed by one or more processors, or a processor and a controller, and one or more other operations may be performed by one or more other processors, or another processor and another controller. One or more processors, or a processor and a controller, may perform a single operation, or two or more operations.

[0146] Instructions or software to control computing hardware, for example, one or more processors or computers, to implement the hardware components and perform the methods as described above may be written as computer programs, code segments, instructions or any combination thereof, for individually or collectively instructing or configuring the one or more processors or computers to operate as a machine or special-purpose computer to perform the operations that are performed by the hardware components and the methods as described above. In one example, the instructions or software include machine code that is directly executed by the one or more processors or computers, such as machine code produced by a compiler. In another example, the instructions or software includes higher-level code that is executed by the one or more processors or computer using an interpreter. The instructions or software may be written using any programming language based on the block diagrams and the flow charts illustrated in the drawings and the corresponding descriptions used herein, which disclose algorithms for performing the operations that are performed by the hardware components and the methods as described above.

[0147] The instructions or software to control computing hardware, for example, one or more processors or computers, to implement the hardware components and perform the methods as described above, and any associated data, data files, and data structures, may be recorded, stored, or fixed in or on one or more non-transitory computer-readable storage media. Examples of a non-transitory computer-readable storage medium include read-only memory (ROM), programmable random-access read only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random-access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROMs, CD-Rs, CD+Rs, CD-RWs, CD+RWs, DVD-ROMs, DVD-Rs, DVD+Rs, DVD-RWs, DVD+RWs, DVD-RAMs, BD-ROMs, BD-Rs, BD-R LTHs, BD-REs, blue-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), a card type memory such as multimedia card or a micro card (for example, secure digital (SD) or extreme digital (XD)), magnetic tapes, floppy disks, magnetooptical data storage devices, optical data storage devices, hard disks, solid-state disks, and any other device that is configured to store the instructions or software and any associated data, data files, and data structures in a non-transitory manner and provide the instructions or software and any associated data, data files, and data structures to one or more processors or computers so that the one or more processors or computers can execute the instructions. In one example, the instructions or software and any associated data, data files, and data structures are distributed over network-coupled computer systems so that the instructions and software and any associated data, data files, and data structures are stored, accessed, and executed in a distributed fashion by the one or more processors or computers.

[0148] While this disclosure includes specific examples, it will be apparent after an understanding of the disclosure of this application that various changes in form and details may be made in these examples without departing from the scope of the claims. The examples described herein are to be considered in a descriptive sense only, and not for purposes of limitation. Descriptions of features or aspects in each example are to be considered as being applicable to similar features or aspects in other examples. Suitable results may be achieved if the described techniques are performed in a different order, and/or if components in a described system, architecture, device, or circuit are combined in a different manner, and/or replaced or supplemented by other components or their equivalents.

The following is a list of further preferred embodiments of the invention:

[0149] 

Embodiment 1. A bitcell circuit comprising:

a first variable resistor as a resistive memory device having a resistance value that is set based on the resistive memory device switching between different resistance states;

a second variable resistor as another resistive memory device connected to the first variable resistor in parallel, wherein the second variable resistor is set with a resistance value complementary to the resistance value of the first variable resistor;

a first switch serially connected to the first variable resistor and configured to switch application of a voltage or current to the first variable resistor; and

a second switch serially connected to the second variable resistor and configured to perform a switching operation for applying the voltage or current to the first variable resistor, complementarily to the first switch.

Embodiment 2. The bitcell circuit of embodiment 1, wherein
the first variable resistor is set with either one of a first resistance value and a second resistance value, and
the second variable resistor is set with the second resistance value when the first variable resistor is set with the first resistance value, and is set with the first resistance value when the first variable resistor is set with the second resistance value.

Embodiment 3. The bitcell circuit of embodiment 1, wherein the resistance values of the first variable resistor and the second variable resistor are set with values corresponding to weights for a multiply and accumulate (MAC) operation.

Embodiment 4. The bitcell circuit of embodiment 1, wherein the first switch and the second switch complementarily perform on/off operations according to an input value applied to the bitcell circuit to perform a MAC operation of a neural network.

Embodiment 5. The bitcell circuit of embodiment 4, wherein a resistance value of the bitcell circuit is equal to the resistance value of the first variable resistor when the first switch is closed by application of a first input value, and is equal to the resistance value of the second variable resistor when the second switch is closed by application of a second input value.

Embodiment 6. The bitcell circuit of embodiment 4, wherein
a resistance value of the bitcell circuit is a first resistance value when the first variable resistor is set with the first resistance value corresponding to a first weight and a first input value is applied to the bitcell circuit,
a resistance value of the bitcell circuit is the first resistance value when the second variable resistor is set with the second resistance value corresponding to a second weight and a second input value is applied to the bitcell circuit,
a resistance value of the bitcell circuit is a second resistance value when the second variable resistor is set with the second resistance value corresponding to the second weight and the first input value is applied to the bitcell circuit, and
a resistance value of the bitcell circuit is the second resistance value when the first variable resistor is set with the first resistance value corresponding to the first weight and the second input value is applied to the bitcell circuit.

Embodiment 7. The bitcell circuit of embodiment 4, wherein the bitcell circuit has a resistance value corresponding to a result of an exclusive-NOR (XNOR) operation between the input value applied to the bitcell circuit and a weight set for the bitcell circuit.

Embodiment 8. The bitcell circuit of embodiment 1, wherein
the first variable resistor and the first switch serially connected to each other are parallelly connected with the second variable resistor and the second switch serially connected to each other,
one end of the first variable resistor and one end of the second variable resistor are commonly connected to a first bit data line, and
one end of the first switch and one end of the second switch are commonly connected to a second bit data line.

Embodiment 9. The bitcell circuit of embodiment 8, wherein another bitcell circuit is connected to either one of:
the one end of the first variable resistor and the one end of the second variable resistor commonly connected to the first bit data line; and
the one end of the first switch and the one end of the second switch commonly connected to the second bit data line is connected to.

Embodiment 10. The bitcell circuit of embodiment 1, wherein the first variable resistor and the second variable resistor are magnetic tunnel junction (MTJ) devices.

Embodiment 11. A processing device comprising:
a bitcell array including a plurality of bitcells including the bitcell circuit of embodiment 1.

Embodiment 12. A bitcell circuit comprising:

a pair of variable resistors as resistive memory devices set to have different resistance values and connected to each other in parallel; and

a pair of switches serially connected to the variable resistors, respectively, and configured to complementarily switch applications of a voltage or current to the variable resistors.

Embodiment 13. A processing device comprising:

a bitcell array including a plurality of bitcells each including a pair of variable resistors and a pair of switches,

wherein at least one of the plurality of bitcells comprises:

a first variable resistor as a resistive memory device having a resistance value set based on the resistive memory device switching between different resistance states;

a second variable resistor as another resistive memory device connected to the first variable resistor in parallel, wherein the second variable resistor is set with a resistance value complementary to the resistance value of the first variable resistor;

a first switch serially connected to the first variable resistor and configured to switch application of a voltage or current to the first variable resistor; and

a second switch serially connected to the second variable resistor and configured to perform a switching operation for applying the voltage or current to the first variable resistor, complementarily to the first switch.

Embodiment 14. The processing device of embodiment 13, wherein
the plurality of bitcells form the bitcell array including a plurality of bitcell lines, and
each of the plurality of bitcell lines comprises serially connected bitcells from among the plurality of bitcells.

Embodiment 15. The processing device of embodiment 14, wherein
a first bitcell line from among the plurality of bitcell lines is configured to perform processing of a MAC operation of a first node from among a plurality of nodes of a neural network, and
the pair of variable resistors included in each of bitcells of the first bitcell line are set with resistance values corresponding to weights for the MAC operation of the first node.

Embodiment 16. The processing device of embodiment 15, wherein, in each of the bitcells included in the first bitcell line, one switch of the pair of switches is closed and another switch of the pair is opened, based on input values of the MAC operation of the first node.

Embodiment 17. The processing device of embodiment 15, wherein a result of the MAC operation of the first node corresponds to a value of a voltage drop of the first bitcell line that occurs due to a predetermined value of current applied to the first bitcell line, in response to the setting of the resistance values corresponding to the weights and switching operations of the switches corresponding to the input value.

Embodiment 18. The processing device of embodiment 17, wherein a value of the voltage drop of the first bitcell line corresponds to a sum of values of voltage drops occurring in the bitcells included in the first bitcell line.

Embodiment 19. The processing device of embodiment 13, further comprising a pair of bit data lines connected between one or more of the bitcells and configured to apply a voltage or current to both ends of the one or more of the bitcells, to set resistance values of the pair of variable resistors.

Embodiment 20. The processing device of embodiment 13, wherein the processing device is a device that performs in-memory processing.

Embodiment 21. A processing device including a bitcell array, the processing device comprising:

a plurality of bitcells; and

a pair of bit data lines connected between one or more of the bitcells and configured to apply a voltage or current to both ends of the one or more of the bitcells,

wherein at least one of the plurality of bitcells comprises:

a pair of variable resistors as resistive memory devices set to have different resistance values according to switching between different resistance states and connected to each other in parallel; and

a pair of switches serially connected to the variable resistors, respectively, and configured to complementarily switch applications of a voltage or current to the variable resistors.

Claims

1. A bitcell circuit comprising:

a pair of variable resistors as resistive memory devices set to have different resistance values and connected to each other in parallel; and

a pair of switches serially connected to the variable resistors, respectively, and configured to complementarily switch applications of a voltage or current to the variable resistors.

2. The bitcell circuit of claim 1, wherein the pair of variable resistors comprises
a first variable resistor as a resistive memory device having a resistance value that is set based on the resistive memory device switching between different resistance states, and a second variable resistor as another resistive memory device connected to the first variable resistor in parallel, wherein the second variable resistor is set with a resistance value complementary to the resistance value of the first variable resistor; and
wherein the pair of switches comprises a first switch serially connected to the first variable resistor and configured to switch application of a voltage or current to the first variable resistor, and a second switch serially connected to the second variable resistor and configured to perform a switching operation for applying the voltage or current to the first variable resistor, complementarily to the first switch.

3. The bitcell circuit of claim 2, wherein
the first variable resistor is set with either one of a first resistance value and a second resistance value, and
the second variable resistor is set with the second resistance value when the first variable resistor is set with the first resistance value, and is set with the first resistance value when the first variable resistor is set with the second resistance value.

4. The bitcell circuit of claim 2 or 3, wherein the resistance values of the first variable resistor and the second variable resistor are set with values corresponding to weights for a multiply and accumulate, MAC, operation.

5. The bitcell circuit of one of claims 2 to 4, wherein the first switch and the second switch complementarily perform on/off operations according to an input value applied to the bitcell circuit to perform a MAC operation of a neural network.

6. The bitcell circuit of claim 5, wherein a resistance value of the bitcell circuit is equal to the resistance value of the first variable resistor when the first switch is closed by application of a first input value, and is equal to the resistance value of the second variable resistor when the second switch is closed by application of a second input value, or
wherein a resistance value of the bitcell circuit is a first resistance value when the first variable resistor is set with the first resistance value corresponding to a first weight and a first input value is applied to the bitcell circuit, a resistance value of the bitcell circuit is the first resistance value when the second variable resistor is set with the second resistance value corresponding to a second weight and a second input value is applied to the bitcell circuit, a resistance value of the bitcell circuit is a second resistance value when the second variable resistor is set with the second resistance value corresponding to the second weight and the first input value is applied to the bitcell circuit, and a resistance value of the bitcell circuit is the second resistance value when the first variable resistor is set with the first resistance value corresponding to the first weight and the second input value is applied to the bitcell circuit, or
wherein the bitcell circuit has a resistance value corresponding to a result of an exclusive-NOR, XNOR, operation between the input value applied to the bitcell circuit and a weight set for the bitcell circuit.

7. The bitcell circuit of one of claims 2 to 6, wherein
the first variable resistor and the first switch serially connected to each other are parallelly connected with the second variable resistor and the second switch serially connected to each other,
one end of the first variable resistor and one end of the second variable resistor are commonly connected to a first bit data line, and
one end of the first switch and one end of the second switch are commonly connected to a second bit data line.

8. The bitcell circuit of claim 7, wherein another bitcell circuit is connected to either one of:

the one end of the first variable resistor and the one end of the second variable resistor commonly connected to the first bit data line; and

the one end of the first switch and the one end of the second switch commonly connected to the second bit data line is connected to.

9. The bitcell circuit of one of claims 2 to 8, wherein the first variable resistor and the second variable resistor are magnetic tunnel junction, MTJ, devices.

10. A processing device comprising:
a bitcell array including a plurality of bitcell circuits as claimed in one of claims 1 to 9.

11. The processing device of claim 10, wherein
the plurality of bitcell circuits form the bitcell array including a plurality of bitcell lines, and
each of the plurality of bitcell lines comprises serially connected bitcell circuits from among the plurality of bitcell circuits.

12. The processing device of claim 11, wherein
a first bitcell line from among the plurality of bitcell lines is configured to perform processing of a MAC operation of a first node from among a plurality of nodes of a neural network, and
the pair of variable resistors included in each of bitcell circuits of the first bitcell line are set with resistance values corresponding to weights for the MAC operation of the first node.

13. The processing device of claim 12, wherein, in each of the bitcell circuits included in the first bitcell line, one switch of the pair of switches is closed and another switch of the pair is opened, based on input values of the MAC operation of the first node, or
wherein a result of the MAC operation of the first node corresponds to a value of a voltage drop of the first bitcell line that occurs due to a predetermined value of current applied to the first bitcell line, in response to the setting of the resistance values corresponding to the weights and switching operations of the switches corresponding to the input value.

14. The processing device of claim 13, wherein a value of the voltage drop of the first bitcell line corresponds to a sum of values of voltage drops occurring in the bitcell circuits included in the first bitcell line.

15. The processing device of one of claims 10 to 14, further comprising a pair of bit data lines connected between one or more of the bitcell circuits and configured to apply a voltage or current to both ends of the one or more of the bitcell circuits, to set resistance values of the pair of variable resistors, or
wherein the processing device is a device that performs in-memory processing.

Drawing