Numerous examples are disclosed of an artificial neural network comprising a reference array used for I-V slope configuration in the main array.
Artificial neural networks mimic biological neural networks (the central nervous systems of animals, in particular the brain) and are used to estimate or approximate functions that can depend on a large number of inputs and are generally unknown. Artificial neural networks generally include layers of interconnected “neurons” which exchange messages between each other.
One of the major challenges in the development of artificial neural networks for high-performance information processing is a lack of adequate hardware technology. Indeed, practical neural networks rely on a very large number of synapses, enabling high connectivity between neurons, i.e., a very high computational parallelism. In principle, such complexity can be achieved with digital supercomputers or specialized graphics processing unit clusters. However, in addition to high cost, these approaches also suffer from mediocre energy efficiency as compared to biological networks, which consume much less energy primarily because they perform low-precision analog computation. CMOS analog circuits have been used for artificial neural networks, but most CMOS-implemented synapses have been too bulky given the high number of neurons and synapses.
Applicant previously disclosed an artificial (analog) neural network that utilizes one or more non-volatile memory arrays as the synapses in U.S. Patent Application Publication 2017/0337466A1, which is incorporated by reference. The non-volatile memory arrays operate as an analog neural memory and comprise non-volatile memory cells arranged in rows and columns. The neural network includes a first plurality of synapses configured to receive a first plurality of inputs and to generate therefrom a first plurality of outputs, and a first plurality of neurons configured to receive the first plurality of outputs. The first plurality of synapses includes a plurality of memory cells, wherein each of the memory cells includes spaced apart source and drain regions formed in a semiconductor substrate with a channel region extending there between, a floating gate disposed over and insulated from a first portion of the channel region and a non-floating gate disposed over and insulated from a second portion of the channel region. Each of the plurality of memory cells store a weight value corresponding to a number of electrons on the floating gate. The plurality of memory cells multiply the first plurality of inputs by the stored weight values to generate the first plurality of outputs.
Non-Volatile Memory Cells
Non-volatile memories are well known. For example, U.S. Pat. No. 5,029,130 (“the '130 patent”), which is incorporated herein by reference, discloses an array of split gate non-volatile memory cells, which are a type of flash memory cells. Such a memory cell 210 is shown in
Memory cell 210 is erased (where electrons are removed from the floating gate) by placing a high positive voltage on the word line terminal 22, which causes electrons on the floating gate 20 to tunnel through the intermediate insulation from the floating gate 20 to the word line terminal 22 via Fowler-Nordheim (FN) tunneling.
Memory cell 210 is programmed by source side injection (SSI) with hot electrons (where electrons are placed on the floating gate) by placing a positive voltage on the word line terminal 22, and a positive voltage on the source region 14. Electron current will flow from the drain region 16 towards the source region 14. The electrons will accelerate and become heated when they reach the gap between the word line terminal 22 and the floating gate 20. Some of the heated electrons will be injected through the gate oxide onto the floating gate 20 due to the attractive electrostatic force from the floating gate 20.
Memory cell 210 is read by placing positive read voltages on the drain region 16 and word line terminal 22 (which turns on the portion of the channel region 18 under the word line terminal). If the floating gate 20 is positively charged (i.e., erased of electrons), then the portion of the channel region 18 under the floating gate 20 is turned on as well, and current will flow across the channel region 18, which is sensed as the erased or “1” state. If the floating gate 20 is negatively charged (i.e., programmed with electrons), then the portion of the channel region under the floating gate 20 is mostly or entirely turned off, and current will not flow (or there will be little flow) across the channel region 18, which is sensed as the programmed or “0” state.
Table No. 1 depicts typical voltage and current ranges that can be applied to the terminals of memory cell 210 for performing read, erase, and program operations:
Other split gate memory cell configurations, which are other types of flash memory cells, are known. For example,
Table No. 2 depicts typical voltage and current ranges that can be applied to the terminals of memory cell 310 for performing read, erase, and program operations:
Table No. 3 depicts typical voltage and current ranges that can be applied to the terminals of memory cell 410 for performing read, erase, and program operations:
Table No. 4 depicts typical voltage ranges that can be applied to the terminals of memory cell 510 and substrate 12 for performing read, erase, and program operations:
The methods and means described herein may apply to other non-volatile memory technologies such as FINFET split gate flash or stack gate flash memory, NAND flash, SONOS (silicon-oxide-nitride-oxide-silicon, charge trap in nitride), MONOS (metal-oxide-nitride-oxide-silicon, metal charge trap in nitride), ReRAM (resistive ram), PCM (phase change memory), MRAM (magnetic ram), FeRAM (ferroelectric ram), CT (charge trap) memory, CN (carbon-tube) memory, OTP (bi-level or multi-level one time programmable), and CeRAM (correlated electron ram), without limitation.
In order to utilize the memory arrays comprising one of the types of non-volatile memory cells described above in an artificial neural network, two modifications are made. First, the lines are configured so that each memory cell can be individually programmed, erased, and read without adversely affecting the memory state of other memory cells in the array, as further explained below. Second, continuous (analog) programming of the memory cells is provided.
Specifically, the memory state (i.e., charge on the floating gate) of each memory cell in the array can be continuously changed from a fully erased state to a fully programmed state, and vice-versa, independently and with minimal disturbance of other memory cells. This means the cell storage is effectively analog or at the very least can store one of many discrete values (such as 16 or 64 different values), which allows for very precise and individual tuning of all the memory cells in the memory array, and which makes the memory array ideal for storing and making fine tuning adjustments to the synapsis weights of the neural network.
Neural Networks Employing Non-Volatile Memory Cell Arrays
S0 is the input layer, which for this example is a 32×32 pixel RGB image with 5 bit precision (i.e. three 32×32 pixel arrays, one for each color R, G and B, each pixel being 5 bit precision). The synapses CB1 going from input layer S0 to layer C1 apply different sets of weights in some instances and shared weights in other instances and scan the input image with 3×3 pixel overlapping filters (kernel), shifting the filter by 1 pixel (or more than 1 pixel as dictated by the model). Specifically, values for 9 pixels in a 3×3 portion of the image (i.e., referred to as a filter or kernel) are provided to the synapses CB1, where these 9 input values are multiplied by the appropriate weights and, after summing the outputs of that multiplication, a single output value is determined and provided by a first synapse of CB1 for generating a pixel of one of the feature maps of layer C1. The 3×3 filter is then shifted one pixel to the right within input layer S0 (i.e., adding the column of three pixels on the right, and dropping the column of three pixels on the left), whereby the 9 pixel values in this newly positioned filter are provided to the synapses CB1, where they are multiplied by the same weights and a second single output value is determined by the associated synapse. This process is continued until the 3×3 filter scans across the entire 32×32 pixel image of input layer S0, for all three colors and for all bits (precision values). The process is then repeated using different sets of weights to generate a different feature map of layer C1, until all the features maps of layer C1 have been calculated.
In layer C1, in the present example, there are 16 feature maps, with 30×30 pixels each. Each pixel is a new feature pixel extracted from multiplying the inputs and kernel, and therefore each feature map is a two dimensional array, and thus in this example layer C1 constitutes 16 layers of two dimensional arrays (keeping in mind that the layers and arrays referenced herein are logical relationships, not necessarily physical relationships—i.e., the arrays are not necessarily oriented in physical two dimensional arrays). Each of the 16 feature maps in layer C1 is generated by one of sixteen different sets of synapse weights applied to the filter scans. The C1 feature maps could all be directed to different aspects of the same image feature, such as boundary identification. For example, the first map (generated using a first weight set, shared for all scans used to generate this first map) could identify circular edges, the second map (generated using a second weight set different from the first weight set) could identify rectangular edges, or the aspect ratio of certain features, and so on.
An activation function P1 (pooling) is applied before going from layer C1 to layer S1, which pools values from consecutive, non-overlapping 2×2 regions in each feature map. The purpose of the pooling function P1 is to average out the nearby location (or a max function can also be used), to reduce the dependence of the edge location for example and to reduce the data size before going to the next stage. At layer S1, there are 16 15×15 feature maps (i.e., sixteen different arrays of 15×15 pixels each). The synapses CB2 going from layer S1 to layer C2 scan maps in layer S1 with 4×4 filters, with a filter shift of 1 pixel. At layer C2, there are 22 12×12 feature maps. An activation function P2 (pooling) is applied before going from layer C2 to layer S2, which pools values from consecutive non-overlapping 2×2 regions in each feature map. At layer S2, there are 22 6×6 feature maps. An activation function (pooling) is applied at the synapses CB3 going from layer S2 to layer C3, where every neuron in layer C3 connects to every map in layer S2 via a respective synapse of CB3. At layer C3, there are 64 neurons. The synapses CB4 going from layer C3 to the output layer S3 fully connects C3 to S3, i.e. every neuron in layer C3 is connected to every neuron in layer S3. The output at S3 includes 10 neurons, where the highest output neuron determines the class. This output could, for example, be indicative of an identification or classification of the contents of the original image.
Each layer of synapses is implemented using an array, or a portion of an array, of non-volatile memory cells.
Non-volatile memory cell array 33 serves two purposes. First, it stores the weights that will be used by the VMM array 32. Second, the non-volatile memory cell array 33 effectively multiplies the inputs by the weights stored in the non-volatile memory cell array 33 and adds them up per output line (source line or bit line) to produce the output, which will be the input to the next layer or input to the final layer. By performing the multiplication and addition function, the non-volatile memory cell array 33 negates the need for separate multiplication and addition logic circuits and is also power efficient due to its in-situ memory computation.
The output of non-volatile memory cell array 33 is supplied to a differential summer (such as a summing op-amp or a summing current mirror) 38, which sums up the outputs of the non-volatile memory cell array 33 to create a single value for that convolution. The differential summer 38 is arranged to perform summation of positive weight and negative weight.
The summed-up output values of differential summer 38 are then supplied to an activation function block 39, which rectifies the output. The activation function block 39 may provide sigmoid, tan h, or ReLU functions. The rectified output values of activation function block 39 become an element of a feature map as the next layer (e.g. C1 in
The input to VMM array 32 in
The output generated by input VMM array 32a is provided as an input to the next VMM array (hidden level 1) 32b, which in turn generates an output that is provided as an input to the next VMM array (hidden level 2) 32c, and so on. The various layers of VMM array 32 function as different layers of synapses and neurons of a convolutional neural network (CNN). Each VMM array 32a, 32b, 32c, 32d, and 32e can be a stand-alone, physical non-volatile memory array, or multiple VMM arrays could utilize different portions of the same physical non-volatile memory array, or multiple VMM arrays could utilize overlapping portions of the same physical non-volatile memory array. The example shown in
Vector-by-Matrix Multiplication (VMM) Arrays
In VMM array 900, control gate lines, such as control gate line 903, run in a vertical direction (hence reference array 902 in the row direction is orthogonal to control gate line 903), and erase gate lines, such as erase gate line 904, run in a horizontal direction. Here, the inputs to VMM array 900 are provided on the control gate lines (CG0, CG1, CG2, CG3), and the output of VMM array 900 emerges on the source lines (SL0, SL1). In one example, only even rows are used, and in another example, only odd rows are used. The current placed on each source line (SL0, SL1, respectively) performs a summing function of all the currents from the memory cells connected to that particular source line.
As described herein for neural networks, the non-volatile memory cells of VMM array 900, i.e., the memory cells 310 of VMM array 900, may be configured to operate in a sub-threshold region.
The non-volatile reference memory cells and the non-volatile memory cells described herein are biased in weak inversion (sub threshold region):
Ids=Io*e
(Vg-Vth)/nVt
=w*Io*e
(Vg)/nVt,
For an I-to-V log converter using a memory cell (such as a reference memory cell or a peripheral memory cell) or a transistor to convert input current into an input voltage:
Vg=n*Vt*log [Ids/wp*Io]
where, wp is w of a reference or peripheral memory cell.
For a memory array used as a vector matrix multiplier VMM array with the current input, the output current is:
Iout=wa*Io*e(Vg)/nVt, namely
Iout=(wa/wp)*Iin=W*Iin
W=e
(Vthp-Vtha)/nVt
Here, wa=w of each memory cell in the memory array.
Vthp is effective threshold voltage of the peripheral memory cell and Vtha is effective threshold voltage of the main (data) memory cell. Note that the threshold voltage of a transistor is a function of substrate body bias voltage and the substrate body bias voltage, denoted Vsb, can be modulated to compensate for various conditions, on such temperature. The threshold voltage Vth can be expressed as:
Vth=Vth0+gamma(SQRT|Vsb−2*φF)−SQRT|2*φF|)
where Vth0 is threshold voltage with zero substrate bias, φF is a surface potential, and gamma is a body effect parameter.
A wordline or control gate can be used as the input for the memory cell for the input voltage.
Alternatively, the flash memory cells of VMM arrays described herein can be configured to operate in the linear region:
Ids=beta*(Vgs−Vth)*Vds;beta=u*Cox*Wt/L
W=α(Vgs−Vth)
meaning weight W in the linear region is proportional to (Vgs−Vth)
A wordline or control gate or bitline or sourceline can be used as the input for the memory cell operated in the linear region. The bitline or sourceline can be used as the output for the memory cell.
For an I-to-V linear converter, a memory cell (such as a reference memory cell or a peripheral memory cell) or a transistor operating in the linear region can be used to linearly convert an input/output current into an input/output voltage.
Alternatively, the memory cells of VMM arrays described herein can be configured to operate in the saturation region:
Ids=½*beta*(Vgs−Vth)2;beta=u*Cox*Wt/L
Wα(Vgs−Vth)2, meaning weight W is proportional to (Vgs−Vth)2
A wordline, control gate, or erase gate can be used as the input for the memory cell operated in the saturation region. The bitline or sourceline can be used as the output for the output neuron.
Alternatively, the memory cells of VMM arrays described herein can be used in all regions or a combination thereof (sub threshold, linear, or saturation) for each layer or multi layers of a neural network.
Other examples for VMM array 32 of
Memory array 1003 serves two purposes. First, it stores the weights that will be used by the VMM array 1000 on respective memory cells thereof. Second, memory array 1003 effectively multiplies the inputs (i.e. current inputs provided in terminals BLR0, BLR1, BLR2, and BLR3, which reference arrays 1001 and 1002 convert into the input voltages to supply to wordlines WL0, WL1, WL2, and WL3) by the weights stored in the memory array 1003 and then adds all the results (memory cell currents) to produce the output on the respective bit lines (BL0-BLN), which will be the input to the next layer or input to the final layer. By performing the multiplication and addition function, memory array 1003 negates the need for separate multiplication and addition logic circuits and is also power efficient. Here, the voltage inputs are provided on the word lines WL0, WL1, WL2, and WL3, and the output emerges on the respective bit lines BL0-BLN during a read (inference) operation. The current placed on each of the bit lines BL0-BLN performs a summing function of the currents from all non-volatile memory cells connected to that particular bitline.
Table No. 5 depicts operating voltages and currents for VMM array 1000. The columns in the table indicate the voltages placed on word lines for selected cells, word lines for unselected cells, bit lines for selected cells, bit lines for unselected cells, source lines for selected cells, and source lines for unselected cells. The rows indicate the operations of read, erase, and program.
Table No. 6 depicts operating voltages and currents for VMM array 1100. The columns in the table indicate the voltages placed on word lines for selected cells, word lines for unselected cells, bit lines for selected cells, bit lines for unselected cells, source lines for selected cells, and source lines for unselected cells. The rows indicate the operations of read, erase, and program.
Memory array 1203 serves two purposes. First, it stores the weights that will be used by the VMM array 1200. Second, memory array 1203 effectively multiplies the inputs (current inputs provided to terminals BLR0, BLR1, BLR2, and BLR3, for which reference arrays 1201 and 1202 convert these current inputs into the input voltages to supply to the control gates (CG0, CG1, CG2, and CG3) by the weights stored in the memory array and then add all the results (cell currents) to produce the output, which appears on BL0-BLN, and will be the input to the next layer or input to the final layer. By performing the multiplication and addition function, the memory array negates the need for separate multiplication and addition logic circuits and is also power efficient. Here, the inputs are provided on the control gate lines (CG0, CG1, CG2, and CG3), and the output emerges on the bit lines (BL0-BLN) during a read operation. The current placed on each bitline performs a summing function of all the currents from the memory cells connected to that particular bitline.
VMM array 1200 implements uni-directional tuning for non-volatile memory cells in memory array 1203. That is, each non-volatile memory cell is erased and then partially programmed until the desired charge on the floating gate is reached. If too much charge is placed on the floating gate (such that the wrong value is stored in the cell), the cell is erased and the sequence of partial programming operations starts over. As shown, two rows sharing the same erase gate (such as EG0 or EG1) are erased together (which is known as a page erase), and thereafter, each cell is partially programmed until the desired charge on the floating gate is reached.
Table No. 7 depicts operating voltages and currents for VMM array 1200. The columns in the table indicate the voltages placed on word lines for selected cells, word lines for unselected cells, bit lines for selected cells, bit lines for unselected cells, control gates for selected cells, control gates for unselected cells in the same sector as the selected cells, control gates for unselected cells in a different sector than the selected cells, erase gates for selected cells, erase gates for unselected cells, source lines for selected cells, and source lines for unselected cells. The rows indicate the operations of read, erase, and program.
Table No. 8 depicts operating voltages and currents for VMM array 1300. The columns in the table indicate the voltages placed on word lines for selected cells, word lines for unselected cells, bit lines for selected cells, bit lines for unselected cells, control gates for selected cells, control gates for unselected cells in the same sector as the selected cells, control gates for unselected cells in a different sector than the selected cells, erase gates for selected cells, erase gates for unselected cells, source lines for selected cells, and source lines for unselected cells. The rows indicate the operations of read, erase, and program.
Long Short-Term Memory
The prior art includes a concept known as long short-term memory (LSTM). LSTM units often are used in neural networks. LSTM allows a neural network to remember information over predetermined arbitrary time intervals and to use that information in subsequent operations. A conventional LSTM unit comprises a cell, an input gate, an output gate, and a forget gate. The three gates regulate the flow of information into and out of the cell and the time interval that the information is remembered in the LSTM. VMMs are particularly useful in LSTM units.
LSTM cell 1500 comprises sigmoid function devices 1501, 1502, and 1503, each of which applies a number between 0 and 1 to control how much of each component in the input vector is allowed through to the output vector. LSTM cell 1500 also comprises tan h devices 1504 and 1505 to apply a hyperbolic tangent function to an input vector, multiplier devices 1506, 1507, and 1508 to multiply two vectors together, and addition device 1509 to add two vectors together. Output vector h(t) can be provided to the next LSTM cell in the system, or it can be accessed for other purposes.
An alternative to LSTM cell 1600 (and another example of an implementation of LSTM cell 1500) is shown in
Whereas LSTM cell 1600 contains multiple sets of VMM arrays 1601 and respective activation function blocks 1602, LSTM cell 1700 contains only one set of VMM arrays 1701 and activation function block 1702, which are used to represent multiple layers in the example of LSTM cell 1700. LSTM cell 1700 will require less space than LSTM 1600, as LSTM cell 1700 will require ¼ as much space for VMMs and activation function blocks compared to LSTM cell 1600.
It can be further appreciated that LSTM units will typically comprise multiple VMM arrays, each of which requires functionality provided by certain circuit blocks outside of the VMM arrays, such as a summer and activation function block and high voltage generation blocks. Providing separate circuit blocks for each VMM array would require a significant amount of space within the semiconductor device and would be somewhat inefficient. The examples described below therefore reduce the circuitry required outside of the VMM arrays themselves.
Gated Recurrent Units
An analog VMM implementation can be utilized for a GRU (gated recurrent unit) system. GRUs are a gating mechanism in recurrent neural networks. GRUs are similar to LSTMs, except that GRU cells generally contain fewer components than an LSTM cell.
An alternative to GRU cell 2000 (and another example of an implementation of GRU cell 1900) is shown in
Whereas GRU cell 2000 contains multiple sets of VMM arrays 2001 and activation function blocks 2002, GRU cell 2100 contains only one set of VMM arrays 2101 and activation function block 2102, which are used to represent multiple layers in the example of GRU cell 2100. GRU cell 2100 will require less space than GRU cell 2000, as GRU cell 2100 will require ⅓ as much space for VMMs and activation function blocks compared to GRU cell 2000.
It can be further appreciated that GRU systems will typically comprise multiple VMM arrays, each of which requires functionality provided by certain circuit blocks outside of the VMM arrays, such as a summer and activation function block and high voltage generation blocks. Providing separate circuit blocks for each VMM array would require a significant amount of space within the semiconductor device and would be somewhat inefficient. The examples described below therefore reduce the circuitry required outside of the VMM arrays themselves.
The input to the VMM arrays can be an analog level, a binary level, a pulse, a time modulated pulse, or digital bits (in this case a DAC is needed to convert digital bits to appropriate input analog level) and the output can be an analog level, a binary level, a timing pulse, pulses, or digital bits (in this case an output ADC is needed to convert output analog level into digital bits).
In general, for each memory cell in a VMM array, each weight W can be implemented by a single memory cell or by a differential cell or by two blend memory cells (average of 2 cells). In the differential cell case, two memory cells are needed to implement a weight W as a differential weight (W=W+−W−). In the two blend memory cells, two memory cells are needed to implement a weight W as an average of two cells.
Each non-volatile memory cells used in the analog neural memory system is to be erased and programmed to hold a very specific and precise amount of charge, i.e., the number of electrons, in the floating gate. For example, each floating gate should hold one of N different values, where N is the number of different weights that can be indicated by each cell. Examples of N include 16, 32, 64, 128, and 256.
To ensure accuracy of a VMM system, it is important to compensate for natural variations that occur. For example, the I-V slope can vary within the same memory array due to natural process variations, and it can vary as operating temperature changes. The I-V slope refers to the relationship between the current drawn by a memory cell when a voltage is applied to a terminal of the memory cell, such as the control gate terminal. It would be desirable to be able to compensate for such changes to the I-V slope of part of, or all of the, memory array.
Numerous examples are described for providing a neural network system comprising a plurality of reference arrays that contain a range of intentional variations so that an appropriate portion of the reference array that closely approximates the operating array can be selected for operations.
VMM System Architecture
The input circuit 3406 may include circuits such as a DAC (digital to analog converter), DPC (digital to pulses converter, digital to time modulated pulse converter), AAC (analog to analog converter, such as a current to voltage converter, logarithmic converter), PAC (pulse to analog level converter), or any other type of converters. The input circuit 3406 may implement one or more of normalization, linear or non-linear up/down scaling functions, or arithmetic functions. The input circuit 3406 may implement a temperature compensation function for input levels. The input circuit 3406 may implement an activation function such as ReLU or sigmoid.
The output circuit 3407 may include circuits such as an ADC (analog to digital converter, to convert neuron analog output to digital bits), AAC (analog to analog converter, such as a current to voltage converter, logarithmic converter), APC (analog to pulse(s) converter, analog to time modulated pulse converter), or any other type of converters. The output circuit 3407 may implement an activation function such as rectified linear activation function (ReLU) or sigmoid. The output circuit 3407 may implement one or more of statistic normalization, regularization, up/down scaling/gain functions, statistical rounding, or arithmetic functions (e.g., add, subtract, divide, multiply, shift, log) for neuron outputs. The output circuit 3407 may implement a temperature compensation function for neuron outputs or array outputs (such as bitline output) so as to keep power consumption of the array approximately constant or to improve precision of the array (neuron) outputs such as by keeping the IV slope approximately the same.
Physical array 3500 is divided into two types of arrays, VMM array 3401 (as in
Reference memory cell 4403 can specifically be from the reference array selected in operation 4103 of method 4100 in
Op-amp 4402 outputs a voltage on terminal CG, CGbias, of reference memory cell 4403. Based on the inherent nature of operational amplifiers, op-amp 4402 adjusts this output voltage until the voltage on its inverting input terminal is equal to the voltage, VREF (which can be an arbitrary reference value, such as 0.5V), on its non-inverting input terminal. The output voltage of op-amp 4402 is sampled and can be applied to control gate terminals of memory cells in VMM array 3401 as a bias voltage, or alternatively as an analog input to the array. Alternatively, CGbias can be used in systems such as row DAC bias generator 4500 in
Reference cell 4403 can be a single cell or can comprise multiple cells. For example, multiple reference cells can be used to obtain an averaging effect or to reflect the weight distribution of a layer of a neural network. For example, multiple reference cells can be selected such that 50% of the reference cells reflect low weight values, 20% of the reference cells reflect medium weight values, and 30% of the reference cells reflect high weight values, in an attempt to approximate the typical weight distribution in a VMM array in an artificial neural network.
Or, for example, different instantiations of row bias calibration circuit 4400 can be used to generate different CGbias levels, such as high, medium, and low CGbias levels. For example, a low CGbias level (VREFL) can be generated using a number of reference cells (e.g., 1 cell, 2 cells, . . . 32 cells) where each cell has a target current of 3 nA (generated by current digital-to-analog converter 4401) representing the 1st level of 32 levels of the memory cells for a 5-bit cell; a high CGbias level (VREFH) can be generated using a number of reference cells (e.g., 1 cells, 2 cells, . . . 32 cells), where each cell has a target current of 96 nA representing the 32nd level of 32 levels of the memory cells for a 5-bit cell; and a medium CGbias level (VREFM) can be generated using a number of reference cells (e.g., 1 cells, 2 cells, . . . , 32 cells), where each cells has a target current of 48 nA representing the 16th level of 32 levels of the memory cells for 5-bit cell. In this manner, three instances of row bias calibration circuit 4400 can be used to generate a high voltage VREFH, a medium voltage VREFM, and a low voltage VREFL.
For the case of a serial DAC input (such as an input comprising a sequence of one digital input bit applied at a time, where the corresponding digital outputs are shifted and added for each binary input bit position) or a timed input (pulse-width modulation input or pulse count input), a CG bias voltage that is applied to the VMM array can be provided by row bias calibration circuit 4400 of
The mapping block 4502 translates the levels from voltage ladder 4501 into a respective analog output, such as a voltage representing an 8-bit output, and the output of mapping block 4502 is fed to output buffer 4503.
The output of output buffer 4503 is depicted as DAC_OUT 4590, which is then applied to a row of memory cells in a VMM array as a bias voltage (such as on a CG terminal) to alter the I-V slope of those cells. Local trimming for each level (represented by trim blocks L0_trm, . . . Ln_trm) is provided within mapping block 4502, at least when the array is operating in the sub-threshold region. Applying this bias voltage to, for example, the control gate line of the row of memory cells will achieve a non-linear I-V slope for the memory cells in the VMM array over a temperature range in the sub-threshold region. It can be appreciated that DAC_Out 4590 is generated using multiple reference arrays (for example, one reference array to generate VREFH and another reference array to generate VREFL).
Voltage ladder 4501 receives a high reference voltage, VREFH (such as VREFH generated by an instantiation of row bias calibration circuit 4400 in
Mapping block 4502 receives digital input DIN [n:0], which is used to select one of the m+1 analog voltages using sub blocks 4563 inside mapping block 4502, where (m+1)=2{circumflex over ( )}(n+1). For example, if (n+1)=8, then (m+1)=256. Mapping block 4502 comprises (m+1) trim blocks 4562 and (m+1) multiplexors 4563. Mapping block 4502 translates the k+1 voltage levels from the DAC 4501 into a respective analog output corresponding to DIN[n:0]. This is achieved with local trimming for each level (represented by trim blocks L0_trm, . . . Lm_trm), which may be useful, for example, when the non-volatile memory cells in the array are operating in the sub-threshold region. This is desirable to achieve a best matching I-V slope for the non-volatile memory cells in the VMM array over temperature in sub threshold region or linear region.
By adjusting reference voltages VREFL and VREFH, the range of the k+1 levels are adjusted as well. This is, for example, to match the output range of this input block with an input range of the memory cells. This also can be performed for temperature compensation by adjusting (such as shifting lower at high temperature and higher at lower temperature) the reference levels VREFL and VREFH to achieve the desired range of gate bias of the memory cells that may be required due to temperature variations. Further individual level adjustment and temperature compensation is done by level trimming circuits of mapping block 4502 for example for best neural network accuracy.
The reference voltages are generated from the CGbias outputs of instantiations of row bias calibration circuits 4400 in
Global DAC 4701 optionally can comprise row DAC bias generator 4500 in
Each set of a row register 4703, digital comparator 4704, and row sample-and-hold buffer 4705 for a particular row can be considered a local digital-to-analog converter for that particular row, where the voltage supply for that local digital-to-analog converter is provided by global DAC 4701, as explained below.
Address row decoders 4702-0 to 4702-n receive a row address, ADD[n:0], and an enable signal, EN. The output of each address row decoder, denoted ENROW, is high when ADD[n:0] is the address for that particular row and when EN is asserted. Row registers 4703-0 through 4703-n are loaded with respective digital input bits DINx (where x is the number of bits, such as 8 or 16 bits) for that particular row, where the loading operation is triggered by a clock signal, CLK, where DINx is the activation input for that particular row, to perform the vector-by-multiplication matrix operation. When the output of a particular address row decoder 4702 is high, the associated row register 4703 is enabled and outputs its digital bits, DINx. Counter 4707 counts the pulses in another clock signal, CLKB, when enabled by signal EN. The output of counter 4707 is a count value. Digital comparator blocks 4704-0 to 4704-n compare the activation value, DINx, stored in the each of the respective row registers 4703 against the count value. If the count value matches the value stored in a particular row register 4702, then the respective digital comparator block 4704 enables the corresponding row S/H buffer 4705 to sample and hold the analog output value from global DAC 4701. Global DAC 4701 performs digital-to-analog conversion on the count value (which also matches the DINx for the row register 4703 for the row in question). Each row S/H buffer 4705 outputs the held analog value as output signal 4706. If, for example, x=8 and DINx is an 8-bit input (meaning that there are 256 different values for DINx), then counter 4707 will count up to 256 and then reset. In doing so, it will have covered all possible values of DINx, and each row S/H buffer 4705 will have been loaded with its associated value of DINx.
Output signal 4706 can be applied, for example, to a control gate line or a word line during a programming or read operation in that particular row or during a neural read operation where all rows are read. During a neural read, all S/H buffers 4705 are enabled to drive the array input terminals through respective output signals 4706, resulting in bitline currents being output by the VMM array, which are then processed by output circuits, such as ITV (current to voltage converter) circuits and ADC (analog-to-digital converter) circuits.
Each respective output signal 4706 can be applied, for example, to a respective control gate line or a word line during a programming or read operation in that particular row.
In another example, a row sample-and-hold buffer 4704 can be shared by multiple rows in a time-multiplexed fashion.
Global DAC 4801 optionally can comprise row DAC bias generator 4500 in
Address row decoders 4802-0 to 4802-n receive a row address, ADD[n:0], and an enable signal, EN. The output of each address row decoder 4802, denoted ENROW, is high when ADD[n:0] is the address for that particular row and EN is asserted. Row registers 4803-0 through 4803-n are loaded with respective digital input bits DINx (where x is the number of bits, such as 8 or 16 bits), where the loading is triggered by clock signal, CLK, for that particular row, where DINx is the activation input for that particular row. When the output, ENROW, of a particular address row decoder 4802 is high, the associated row register 4803 is enabled and outputs its digital bits, DINx. Counter 4807, when enabled by signal EN, counts the pulses in another clock signal, CLKB. The output of counter 4807 is a count value. Digital comparator blocks 4804-0 to 4804-n compare the activation value, DINx, stored in the each of the respective row registers 4803 against the count value. If the count value matches the value stored in a particular row register 4802, then the corresponding row S/H buffer 4805 is enabled by the respective digital comparator block 4804 to sample and hold the analog output value from GDAC 4801. As shown, there are two vertical analog output lines from GDAC 4801. For example, for an 8-bit GDAC 4801, one line can deliver outputs for 0 to 127 analog levels (corresponding to 00000000 to 01111111) and the other line can deliver outputs for 128 to 255 analog levels (corresponding to 10000000 to 11111111). Both lines can operate at the same times to cut the row DAC sampling times from 256 (DAC) clocks to 128 (DAC) clocks. Each row S/H buffer 4805 receives only one of the outputs from GDAC 4801. Output signal 4806 can be applied, for example, to a control gate line or a word line during a programming operation in that particular row.
Optionally, global digital-to-analog converter 4801 can comprise a first global DAC for odd rows and a second global DAC for even rows.
It should be noted that, as used herein, the terms “over” and “on” both inclusively include “directly on” (no intermediate materials, elements or space disposed therebetween) and “indirectly on” (intermediate materials, elements or space disposed therebetween). Likewise, the term “adjacent” includes “directly adjacent” (no intermediate materials, elements or space disposed therebetween) and “indirectly adjacent” (intermediate materials, elements or space disposed there between), “mounted to” includes “directly mounted to” (no intermediate materials, elements or space disposed there between) and “indirectly mounted to” (intermediate materials, elements or spaced disposed there between), and “electrically coupled” includes “directly electrically coupled to” (no intermediate materials or elements there between that electrically connect the elements together) and “indirectly electrically coupled to” (intermediate materials or elements there between that electrically connect the elements together). For example, forming an element “over a substrate” can include forming the element directly on the substrate with no intermediate materials/elements therebetween, as well as forming the element indirectly on the substrate with one or more intermediate materials/elements there between.
This application claims priority to U.S. Provisional Patent Application No. 63/328,543, filed on Apr. 7, 2022, and titled, “Artificial Neural Network Comprising Monte Carlo Reference Array for I-V Slope Configuration,” which is incorporated by reference herein.
Number | Date | Country | |
---|---|---|---|
63328543 | Apr 2022 | US |