A probabilistic computing device and a method of random data generation

Abstract

The application discloses a kind of probability computing device and random data generation method, it is related to probability calculation field, this probability computing device includes: magnetic tunnel junction and adjusting circuit, magnetic tunnel junction at least includes: first port, second port and third port;Adjusting circuit at least includes: first signal source, second signal source, third signal source, first reference element, second reference element and reading element, first signal source connects first port and provides input signal for magnetic tunnel junction, third signal source connects third port and is grounded for magnetic tunnel junction, second signal source connects second port and continues to adjust the output of voltage to this tunnel junction, realize accurate control using adjusting voltage to magnetic tunnel junction probability calculation, reduce the bit error rate of device, and reference element can also be added in the control circuit, increase the difference between reading voltage and reference voltage in a certain range, improve the signal-to-noise ratio of output signal, reduce the influence of noise on output data.

Description

Technical Field

[0001] This invention relates to the field of probability calculation, specifically to a probability calculation device and a method for writing random data. Background Technology

[0002] With the rapid development of the integrated circuit industry, the problems that need to be solved are becoming increasingly complex, and probabilistic computing is being used more and more widely in the fields of artificial intelligence, big data, and the Internet of Things. At present, quantum computing systems are commonly used for probabilistic calculations, but the state of the basic unit of quantum computing systems, the qubit, is difficult to maintain, and it has high requirements for the working environment, making large-scale integration difficult.

[0003] Magnetic tunnel junctions (MTJs) possess excellent characteristics such as high response speed, low power consumption, and long durability, making them a fundamental unit of magnetic random access memory (MRAM) and widely used. Applying them to probabilistic computation, by controlling the probability through the adjustment of bias voltage, can effectively solve the speed and reliability problems in probabilistic calculations.

[0004] However, in practical applications of MTJ, it was found that the MTJ probabilistic device has a relatively low switching ratio and weak current regulation capability. When the current fluctuates or when reading the device signal, the output has a large bit error rate. In addition, due to the presence of a large amount of noise in the signal, the overall signal-to-noise ratio of the device is low, which affects the accuracy of the data. Summary of the Invention

[0005] This invention provides a probability calculation device and a random data generation method to improve the current regulation capability of the MTJ itself, reduce the output bit error rate, improve the signal-to-noise ratio of the device, and increase the accuracy of the data.

[0006] To address the aforementioned problems, a first aspect of the present invention provides a probability calculation device.

[0007] The probability calculation device includes:

[0008] The regulating circuit and the magnetic tunnel junction, wherein the magnetic tunnel junction includes at least: a first port, a second port and a third port;

[0009] The adjustment circuit includes at least: a first signal source, a second signal source, a third signal source, a first reference element, a second reference element, and a readout element.

[0010] The first signal source is connected in parallel with one end of the second reference element and the first end of the magnetic tunnel junction, and the first signal source is used as the input terminal of the adjustment circuit.

[0011] The third signal source is used as the output terminal of the adjustment circuit, and the first terminal of the third signal source is connected in parallel with one terminal of the first reference element and the other terminal of the second reference element.

[0012] The second signal source is connected in parallel with the other end of the first reference element and one end of the readout element;

[0013] The second end of the third signal source is connected in parallel with the other end of the reading element and the second port of the magnetic tunnel junction;

[0014] The third port of the magnetic tunnel junction is grounded.

[0015] In some embodiments, the first signal source provides an electrical signal to the magnetic tunnel junction. When the electrical signal changes, the probability of the magnetic tunnel junction changing from a first resistive state to a second resistive state in response to the electrical signal changes. The change in the electrical signal includes at least: a change in current, a change in voltage, and a change in frequency.

[0016] In some embodiments, the second signal source provides an adjustment voltage to the magnetic tunnel junction, and in response to the adjustment voltage, the probability of the magnetic tunnel junction changing from the first resistive state to the second resistive state changes, and the probability of change corresponds one-to-one with the adjustment voltage.

[0017] In some embodiments, the output voltage of the third signal source includes: the difference between the read voltage and the reference voltage.

[0018] The reference voltage includes: the voltage obtained by connecting the first terminal of the third signal source in parallel with the first reference element and the second reference element;

[0019] The read voltage includes: the read voltage of the third signal source includes: the voltage after the second terminal of the third signal source is connected in parallel with the read element and the magnetic tunnel junction.

[0020] In some embodiments, the resistance of the first reference element after being connected in parallel with the second reference element is a first resistance value, and the resistance of the reading element after being connected in parallel with the magnetic tunnel junction is a second resistance value, wherein the first resistance value and the second resistance value are equal in magnitude.

[0021] In some embodiments, in response to the magnetic tunnel junction changing from the first resistive state to the second resistive state, the read voltage changes from the first read voltage to the second read voltage, and in response to the read voltage changing from the first read voltage to the second read voltage, the output voltage of the third signal source changes from the first output voltage to the second output voltage.

[0022] In some embodiments, the magnetic tunnel junction further includes a top electrode connected in series with the readout element, the top electrode being made of at least one or a combination of two or more of the following materials: aluminum, tantalum, gold, chromium, copper, molybdenum, tungsten, and platinum.

[0023] In some embodiments, the magnetic tunnel junction further includes: a free layer, a reference layer, and a barrier layer.

[0024] The magnetization direction of the reference layer does not change after fabrication and serves as a reference direction for the magnetization direction of the free layer. When the magnetization direction of the free layer is the same as that of the reference layer, the magnetic tunnel junction exhibits a low-resistance state. When the magnetization direction of the free layer is opposite to that of the reference layer, the magnetic tunnel junction exhibits a high-resistance state.

[0025] In some embodiments, the magnetic tunnel junction further includes a bottom electrode, which is made of at least one or a combination of two or more of the following materials: tantalum, aluminum, gold, chromium, copper, molybdenum, tungsten, and platinum.

[0026] In some embodiments, the material of the barrier layer includes at least one of the following compounds or a mixture of compounds, the compounds including: aluminum oxide (Al2O3) or magnesium oxide (MgO).

[0027] In another aspect of this application, a random data generation method is also proposed for use in probability calculation devices.

[0028] The probability calculation device includes an adjustment circuit and a magnetic tunnel junction, the adjustment circuit including at least a first signal source, a second signal source, and a third signal source, and the method includes:

[0029] The electrical signal input from the first signal source is acquired, and in response to the electrical signal, the resistance state of the magnetic tunnel junction may change from the first resistance state to the second resistance state.

[0030] The adjustment voltage input from the second signal source is obtained. In response to the adjustment voltage, the probability of the magnetic tunnel junction changing from the first resistance state to the second resistance state changes, and the probability corresponds one-to-one with the adjustment voltage.

[0031] The output voltage of the third signal source is obtained. If the output voltage is greater than or equal to a preset threshold, the magnetic tunnel junction output is set to data "1".

[0032] If the output voltage is less than a preset threshold, the output data of the magnetic tunnel junction is set to "0".

[0033] When the electrical signal changes, the probability changes in response to the change in the electrical signal, and the change in the electrical signal includes at least: current change, voltage change and frequency change.

[0034] In some embodiments, the adjustment circuit further includes a first reference element, a second reference element, and a readout element, and the third signal source includes a first terminal and a second terminal.

[0035] The output voltage of the third signal source includes: the difference between the read voltage of the third signal source and the reference output voltage of the third signal source.

[0036] The reference voltage of the third signal source includes: the voltage after the first terminal of the third signal source is connected in parallel with the first reference element and the second reference element;

[0037] The reading voltage of the third signal source includes the voltage after the second terminal of the third signal source is connected in parallel with the reading element and the magnetic tunnel junction.

[0038] In some embodiments, in response to the magnetic tunnel junction changing from the first resistive state to the second resistive state, the read voltage changes from the first read voltage to the second read voltage.

[0039] In response to the first read voltage changing to the second read voltage, the output voltage changes from the first output voltage to the second output voltage.

[0040] In some embodiments, the voltage adjustment input method includes: inputting after the electrical signal and inputting synchronously with the electrical signal.

[0041] In some embodiments, the magnetic tunneling membrane structure may be constructed by sputtering.

[0042] Embodiments of the present invention provide a probability calculation device and a random data generation method. By adding an adjustment circuit to the storage unit of the probability calculation device, and using a second signal source in the adjustment circuit to send a control voltage to the magnetic tunnel junction, the probability of the magnetic tunnel junction changing from a first resistive state to a second resistive state changes in response to the control voltage. In response to the change in the resistive state of the magnetic tunnel junction, the read voltage of the magnetic tunnel junction changes, and in response to the change in the read voltage, the output voltage of the adjustment circuit changes. The output voltage is acquired and compared with a preset threshold, and the output voltage is judged as either "0" or "1". Through precise control of the control voltage, the accuracy of the output data of the probability device is improved. Furthermore, a reference element can be added to the adjustment circuit to increase the difference between the read voltage and the reference voltage within a certain range, thereby improving the signal-to-noise ratio of the output signal and reducing the impact of noise on the output data. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application.

[0044] Figure 1 This is a schematic diagram of the adjustment circuit of a probability device according to an embodiment of this application;

[0045] Figure 2 This is a schematic diagram of the impedance analysis structure of the adjustment circuit of a probability device according to an embodiment of the present invention.

[0046] Figure 3 This is a schematic diagram of the magnetic tunnel junction bottom electrode structure according to an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the port structure of a probability calculation device after packaging according to an embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of the array structure of probability calculation devices according to an embodiment of the present invention. Detailed Implementation

[0049] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] Those skilled in the art will understand that the terms "first," "second," etc., in this application are only used to distinguish different devices, modules, or parameters, and do not represent any specific technical meaning, nor do they indicate any necessary logical order between them.

[0051] With the rapid development of the integrated circuit industry, the problems that need to be solved are becoming increasingly complex, and probabilistic computing is being used more and more widely in the fields of artificial intelligence, big data, and the Internet of Things. At present, quantum computing systems are commonly used for probabilistic calculations, but the state of the basic unit of quantum computing systems, the qubit, is difficult to maintain, and it has high requirements for the working environment, making large-scale integration difficult.

[0052] Magnetic tunnel junctions (MTJs) possess excellent characteristics such as high response speed, low power consumption, and long durability, making them a fundamental unit of magnetic random access memory (MRAM) and widely used. Applying them to probabilistic computation, by controlling the probability through the adjustment of bias voltage, can effectively solve the speed and reliability problems in probabilistic calculations.

[0053] However, in practical applications of MTJ, it was found that the MTJ probabilistic device has a relatively low switching ratio and weak current regulation capability. When the current fluctuates or when reading the device signal, the output has a large bit error rate. In addition, due to the presence of a large amount of noise in the signal, the overall signal-to-noise ratio of the device is low, which affects the accuracy of the data.

[0054] In one embodiment of this application, a probability calculation device is proposed, the structure of which is as follows: Figure 1 As shown, the probability calculation device includes:

[0055] The regulating circuit and the magnetic tunnel junction, wherein the magnetic tunnel junction includes at least: a first port, a second port, and a third port.

[0056] The adjustment circuit includes at least: a first signal source, a second signal source, a third signal source, a first reference element, a second reference element, and a readout element, wherein the readout element can be a resistive element.

[0057] The first signal source is connected in parallel with one end of the second reference element and the first end of the magnetic tunnel junction, and the first signal source is used as the input terminal of the adjustment circuit.

[0058] The third signal source is used as the output terminal of the adjustment circuit, and the first terminal of the third signal source is connected in parallel with one terminal of the first reference element and the other terminal of the second reference element.

[0059] The second signal source is connected in parallel with the other end of the first reference element and one end of the readout element;

[0060] The second end of the third signal source is connected in parallel with the other end of the reading element and the second port of the magnetic tunnel junction;

[0061] The third port of the magnetic tunnel junction is grounded.

[0062] Typically, a magnetic tunnel junction structure consists of a free layer, a reference layer, a barrier layer, and a bottom electrode, and usually has two ports: a bottom electrode input port, used to connect the input signal (equivalent to the first port), and a ground port (equivalent to the third port).

[0063] Optionally, a third port is added to the magnetic tunnel junction, and a regulating voltage is input through this port. The regulating voltage can be an artificially input or program-preset electrical signal. Since the magnetic field direction of the free layer of the magnetic tunnel junction is affected by the surrounding electric field and will change. For example, when the surrounding electrical signal is greater than a, the magnetic field direction of the free layer is the same as that of the fixed layer; when the surrounding electrical signal is less than b, the magnetic field direction of the free layer is opposite to that of the fixed layer (a and b are constants, and a > b). That is, when the magnetic field direction of the free layer is the same as that of the fixed layer in the initial state and the input electrical signal is less than b, affected by the electrical signal, the magnetic field direction of the free layer will change to be opposite to that of the fixed layer. In response to the change in the magnetic field direction, the resistance state of the magnetic tunnel junction will also change. Assume that the magnetic tunnel junction presents a low resistance state when the magnetic field direction of the free layer is the same as that of the fixed layer, and the magnetic tunnel junction presents a high resistance state when the magnetic field direction of the free layer is opposite to that of the fixed layer. Then after the above-mentioned change in the magnetic field direction of the free layer, the resistance state of the magnetic tunnel junction changes from the low resistance state to the high resistance state (also known as resistance state flipping). Further, we define the reading voltage of the magnetic tunnel junction in the low resistance state as "0" and the reading voltage of the magnetic tunnel junction in the high resistance state as "1". Then the flipping of the resistance state of the magnetic tunnel junction can be regarded as the flipping of "0" or "1". When an electrical signal greater than a is input to the magnetic tunnel junction, the change in the resistance state of the magnetic tunnel junction is similar to the above content and will not be elaborated here;

[0064] Further, when the input electrical signal x satisfies b < x < a, that is, the magnitude of the input electrical signal is between the voltage values that induce the resistance state flipping of the magnetic tunnel junction. After actual verification, the magnetic tunnel junction will have a probability of resistance state flipping, and the corresponding flipping probability is measured through actual verification and shows a normal distribution. In other words, when the magnitude of the input electrical signal is x, the data fed back by the magnetic tunnel junction shows "0" with a probability of A and shows "1" with a probability of B, and A + B = 100%;

[0065] Further, since the main influencing factor for the resistance state flipping of the magnetic tunnel junction is the above-mentioned electrical signal, the probability of the resistance state flipping of the magnetic tunnel junction can be changed by adjusting the magnitude of the electrical signal. That is, the probability of the magnetic tunnel junction outputting "0" or "1" can be changed by adjusting the magnitude of the electrical signal. Therefore, the third port here is called the regulating port (equivalent to the second port), and the electrical signal is called the regulating voltage. In this way, when performing probability calculations, the above-mentioned probability calculation device can be selected to affect the data output probability of the magnetic tunnel junction by controlling the regulating voltage.

[0066] In some possible embodiments, the first signal source provides an electrical signal to the magnetic tunnel junction, causing the magnetic tunnel junction to change from a first resistance state to a second resistance state. At the same time, by changing the electrical signal (for example: changing voltage, current, and frequency, etc.), the probability of resistance state flipping of the magnetic tunnel junction can also be adjusted. Exemplarily, by adjusting the electrical signal such that the voltage of the electrical signal applied across the magnetic tunnel junction is as described above x (b < x < a), then the magnetic tunnel junction has a certain probability of flipping, but the flipping probability is not 100%. In this way, it can be ensured that the magnetic tunnel junction is adjusted to a preset resistance state (low resistance state or high resistance state) at the initial stage of calculation, and during the process, the probability of resistance state flipping of the magnetic tunnel junction can be jointly adjusted with the adjustment voltage, and the probability of resistance state flipping of the magnetic tunnel junction can also be pre-adjusted.

[0067] In some possible embodiments, the second signal source provides an adjustment voltage to the magnetic tunnel junction. In response to the adjustment voltage, the probability of the magnetic tunnel junction changing from the first resistance state to the second resistance state changes, and the probability corresponds to the adjustment voltage one by one. In this way, it can be ensured that each adjustment voltage corresponds to the probability output value of each magnetic tunnel junction data, and can be adjusted according to requirements.

[0068] In some possible embodiments, the output voltage of the third signal source includes: the difference between the read voltage and the reference voltage,

[0069] The reference voltage includes: the voltage after the first end of the third signal source is connected in parallel with the first reference element and the second reference element;

[0070] The read voltage includes: the read voltage of the third signal source includes: the voltage after the second end of the third signal source is connected in parallel with the read element and the magnetic tunnel junction.

[0071] Generally, the bottom electrode of the magnetic tunnel junction is as Figure 3 shown. The width in the middle of the bottom electrode is small. Therefore, when performing resistance analysis, the bottom electrode is usually regarded as two resistors connected in series, as Figure 2 shown. The resistance value of the bottom electrode is regarded as the series connection of the first bottom electrode resistance and the second bottom electrode resistance (the resistance value of the first bottom electrode resistance = the resistance value of the second bottom electrode resistance = the bottom electrode resistance value / 2), and the voltage transmitted to the magnetic tunnel junction through the first bottom electrode resistance is defined as the input voltage of the magnetic tunnel junction. Then, the resistance analysis of the device unit is as follows:

[0072]

[0073]

[0074]

[0075] Where: V + The reading voltage, V - V represents the reference voltage. set V represents the input voltage corresponding to the first signal source. control V represents the adjustment voltage corresponding to the second signal source. out V represents the output voltage corresponding to the third signal source. x R represents the input voltage of the magnetic tunnel junction. ref1 R represents the resistance value of the first reference element. ref2 R represents the resistance value of the second reference element. MTJ R represents the resistance of the magnetic tunnel junction. read The resistance value of the reading element is given. Based on the above formula derivation, the output voltage formula can be obtained as follows:

[0076]

[0077] From the above formula, we can see that the resistance change in this circuit is R. MTJ To distinguish (V) + )-(V - The value in R MTJ The value of V under high and low resistance states, then (V + )-(V - The larger the value (V), the easier it is to distinguish (i.e., easier to read, lower error rate), thus improving the signal-to-noise ratio of the device. Therefore, to achieve this, increasing (V...)... + )-(V - The value of ) can be implemented in at least the following ways:

[0078] Increase R MTJ The difference between the high-resistivity state and the low-resistivity state;

[0079] Adjust the resistance values ​​of the first bottom electrode and the second bottom electrode;

[0080] Adjust V set and V control Size.

[0081] In some possible implementations, the resistance of the first reference element connected in parallel with the second reference element is a first resistance value, and the resistance of the read element connected in parallel with the magnetic tunnel junction is a second resistance value, wherein the first resistance value and the second resistance value are equal in magnitude.

[0082] Optionally, the selection of the first and second resistance values ​​can be adjusted according to the actual requirements for the clarity of the output signal, based on the above formula. It does not depend on the first and second resistance values ​​being equal in magnitude, and a preset ratio can also be satisfied.

[0083] In some possible implementations, in response to the magnetic tunnel junction changing from the first resistive state to the second resistive state, the read voltage changes from the first read voltage to the second read voltage, and in response to the read voltage changing from the first read voltage to the second read voltage, the output voltage of the third signal source changes from the first output voltage to the second output voltage.

[0084] In some possible implementations, the magnetic tunnel junction further includes a top electrode connected in series with the readout element, the top electrode being made of at least one or a combination of two or more of the following materials: aluminum, tantalum, gold, chromium, copper, molybdenum, tungsten, and platinum.

[0085] In some possible implementations, the magnetic tunnel junction further includes a free layer, a reference layer, and a barrier layer. The magnetization direction of the reference layer does not change after fabrication and serves as a reference direction for the magnetization direction of the free layer. When the magnetization direction of the free layer is the same as the magnetization direction of the reference layer, the magnetic tunnel junction exhibits a low-resistance state. When the magnetization direction of the free layer is opposite to the magnetization direction of the reference layer, the magnetic tunnel junction exhibits a high-resistance state.

[0086] In some possible implementations, the magnetic tunnel junction further includes a bottom electrode, which is made of at least one or a combination of two or more of the following materials: tantalum, aluminum, gold, chromium, copper, molybdenum, tungsten, and platinum.

[0087] In some possible implementations, the material of the barrier layer includes at least one of the following compounds or a mixture of compounds, including: aluminum oxide (Al2O3) or magnesium oxide (MgO).

[0088] Optionally, the probability device can be packaged with the storage device, as shown in the packaging diagram. Figure 4 As shown, the first signal source, the second signal source, and the third signal source can be retained.

[0089] In another embodiment of this application, a random data generation method is also provided, applied to a probability calculation device.

[0090] The probability calculation device includes an adjustment circuit and a magnetic tunnel junction, the adjustment circuit including at least a first signal source, a second signal source, and a third signal source, and the method includes:

[0091] The electrical signal input from the first signal source is acquired, and in response to the electrical signal, the resistance state of the magnetic tunnel junction may change from the first resistance state to the second resistance state.

[0092] The adjustment voltage input from the second signal source is obtained. In response to the adjustment voltage, the probability of the magnetic tunnel junction changing from the first resistance state to the second resistance state changes, and the probability corresponds one-to-one with the adjustment voltage.

[0093] The output voltage of the third signal source is obtained. If the output voltage is greater than or equal to a preset threshold, the magnetic tunnel junction output is set to data "1".

[0094] If the output voltage is less than a preset threshold, the output data of the magnetic tunnel junction is set to "0".

[0095] When the electrical signal changes, the probability changes in response to the change in the electrical signal, and the change in the electrical signal includes at least: current change, voltage change and frequency change.

[0096] In some possible implementations, the adjustment circuit further includes a first reference element, a second reference element, and a readout element, and the third signal source includes a first terminal and a second terminal.

[0097] The output voltage of the third signal source includes: the difference between the read voltage of the third signal source and the reference output voltage of the third signal source.

[0098] The reference voltage of the third signal source includes: the voltage after the first terminal of the third signal source is connected in parallel with the first reference element and the second reference element;

[0099] The reading voltage of the third signal source includes the voltage after the second terminal of the third signal source is connected in parallel with the reading element and the magnetic tunnel junction.

[0100] In some possible implementations, the read voltage changes from a first read voltage to a second read voltage in response to the magnetic tunnel junction changing from a first resistive state to a second resistive state.

[0101] In response to the first read voltage changing to the second read voltage, the output voltage changes from the first output voltage to the second output voltage.

[0102] In some possible implementations, the regulated voltage input method includes: inputting after the electrical signal and inputting synchronously with the electrical signal.

[0103] In some possible implementations, the magnetic tunneling membrane structure may be constructed by sputtering.

[0104] Optionally, the random data generation method can also be implemented through the probability calculation device array, specifically as follows: Figure 5 As shown, Figure 5 The array is represented as a 2*2 array, in which each probability calculation device is connected via a corresponding input word line and input bit line, and the second signal source is connected via a corresponding control word line and control bit line. The output of the true amplifier is read in differential form.

[0105] Embodiments of the present invention provide a probability calculation device and a random data generation method. By adding an adjustment circuit to the storage unit of the probability calculation device, and using a second signal source in the adjustment circuit to send a control voltage to the magnetic tunnel junction, the probability of the magnetic tunnel junction changing from a first resistive state to a second resistive state changes in response to the control voltage. In response to the change in the resistive state of the magnetic tunnel junction, the read voltage of the magnetic tunnel junction changes, and in response to the change in the read voltage, the output voltage of the adjustment circuit changes. The output voltage is acquired and compared with a preset threshold, and the output voltage is judged as either "0" or "1". Through precise control of the control voltage, the accuracy of the output data of the probability device is improved. Furthermore, a reference element can be added to the adjustment circuit to increase the difference between the read voltage and the reference voltage within a certain range, thereby improving the signal-to-noise ratio of the output signal and reducing the impact of noise on the output data.

[0106] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A probabilistic computing device, characterized by, The probability calculation device comprises: an adjusting circuit and a magnetic tunnel junction, the magnetic tunnel junction at least comprising: a first port, a second port and a third port; The adjusting circuit at least comprises: a first signal source, a second signal source, a third signal source, a first reference element, a second reference element and a reading element, The first signal source is connected in parallel with one end of the second reference element and the first port of the magnetic tunnel junction, and the first signal source serves as an input end of the adjusting circuit; The third signal source serves as an output end of the adjusting circuit, and a first end of the third signal source is connected in parallel with one end of the first reference element and the other end of the second reference element; The second signal source is connected in parallel with the other end of the first reference element and one end of the reading element; A second end of the third signal source is connected in parallel with the other end of the reading element and the second port of the magnetic tunnel junction; The third port of the magnetic tunnel junction is grounded, The second signal source provides an adjusting voltage for the magnetic tunnel junction, and in response to the adjusting voltage, the probability of the magnetic tunnel junction changing from a first resistance state to a second resistance state changes, and the probability corresponds to the adjusting voltage in a one-to-one manner.

2. The probabilistic computing device of claim 1, wherein, The first signal source provides an electrical signal for the magnetic tunnel junction, and in response to the electrical signal, the probability of the magnetic tunnel junction changing from a first resistance state to a second resistance state changes when the electrical signal changes, and the electrical signal changing at least comprises: current change, voltage change and frequency change.

3. The probabilistic computing device of claim 1, wherein, The output voltage of the third signal source comprises: a difference between a reading voltage and a reference voltage, The reference voltage comprises: a voltage after the first end of the third signal source is connected in parallel with the first reference element and the second reference element; The reading voltage comprises: a voltage after the second end of the third signal source is connected in parallel with the reading element and the magnetic tunnel junction.

4. The probabilistic computing device of claim 1, wherein, The resistance value of the first reference element after being connected in parallel with the second reference element is a first resistance value, and the resistance value of the reading element after being connected in parallel with the magnetic tunnel junction is a second resistance value, and the first resistance value is equal to the second resistance value.

5. The probabilistic computing device of claim 1, wherein, In response to the magnetic tunnel junction changing from the first resistance state to the second resistance state, the reading voltage changes from a first reading voltage to a second reading voltage, and in response to the reading voltage changing from the first reading voltage to the second reading voltage, the output voltage of the third signal source changes from a first output voltage to a second output voltage.

6. A method of random data generation, the method comprising: The method is applied to a probability calculation device, the probability calculation device comprises an adjusting circuit and a magnetic tunnel junction, the adjusting circuit at least comprises a first signal source, a second signal source and a third signal source, and the method comprises: Obtaining an electrical signal input by the first signal source, and in response to the electrical signal, the resistance state of the magnetic tunnel junction has a probability of changing from a first resistance state to a second resistance state; Obtaining an adjusting voltage input by the second signal source, and in response to the adjusting voltage, the probability of the magnetic tunnel junction changing from the first resistance state to the second resistance state changes, and the probability corresponds to the adjusting voltage in a one-to-one manner; acquiring an output voltage of the third signal source, if the output voltage is greater than or equal to a preset threshold, setting the data output by the magnetic tunnel junction as "1", if the output voltage is less than the preset threshold, setting the data output by the magnetic tunnel junction as "0", when the electrical signal changes, the probability changes in response to the change of the electrical signal, and the change of the electrical signal at least includes: current change, voltage change and frequency change, in response to the change of the magnetic tunnel junction from the first resistance state to the second resistance state, the read voltage changes from the first read voltage to the second read voltage, in response to the change of the first read voltage to the second read voltage, the output voltage changes from the first output voltage to the second output voltage.

7. The method of claim 6, wherein, Further comprising: the adjustment circuit further comprises a first reference element, a second reference element and a read element, the third signal source comprises a first end and a second end, the output voltage of the third signal source comprises: the read voltage of the third signal source minus the output voltage of the reference of the third signal source, the reference voltage of the third signal source comprises: the voltage after the first end of the third signal source is connected in parallel with the first reference element and the second reference element, the read voltage of the third signal source comprises: the voltage after the second end of the third signal source is connected in parallel with the read element and the magnetic tunnel junction.

8. The method of claim 6, wherein, The adjustment voltage input mode comprises: input after the electrical signal and input synchronously with the electrical signal.

Images