Abstract:
A method of forming a semiconductor device starts with a substrate of silicon, a first insulation layer on the silicon, and a silicon layer on the first insulation layer. The silicon layer and the insulation layer are removed just from a second substrate area. A second insulation layer is formed over the silicon layer in the substrate first area and over the silicon in the second substrate area. A first plurality of trenches is formed in the first substrate area that each extends through all the layers and into the silicon. A second plurality of trenches is formed in the second substrate area that each extends through the second insulation layer and into the silicon. An insulation material is formed in the first and second trenches. Logic devices are formed in the first substrate area, and memory cells are formed in the second substrate area.
Abstract:
A non-volatile memory device has a charge pump for providing a programming current and an array of non-volatile memory cells. Each memory cell of the array is programmed by the programming current from the charge pump. The array of non-volatile memory cells is partitioned into a plurality of units, with each unit comprising a plurality of memory cells. An indicator memory cell is associated with each unit of non-volatile memory cells. A programming circuit programs the memory cells of each unit using the programming current, when fifty percent or less of the memory cells of each unit is to be programmed, and programs the inverse of the memory cells of each unit and the indicator memory cell associated with each unit, using the programming current, when more than fifty percent of the memory cells of each unit is to be programmed.
Abstract:
In one example, a method comprises determining a logarithmic slope factor for a selected analog non-volatile memory cell in an array of analog non-volatile memory cells while the selected analog non-volatile memory cell is operating in a sub-threshold region; storing the logarithmic slope factor; determining a linear slope factor for the selected analog non-volatile memory cell while the selected analog non-volatile memory cell is operating in a linear region; storing the linear slope factor; and utilizing one or more of the logarithmic slope factor and the linear slope factor when programming the selected analog non-volatile memory cell to a target current.
Abstract:
Numerous examples are disclosed of input blocks for an array of non-volatile memory cells and associated methods. In one example, a system comprises a vector-by-matrix multiplication array comprising non-volatile memory cells arranged into rows and columns; and an input block comprising a plurality of row circuits and a global digital-to-analog converter generator to generate 2m different analog voltages, where m is an integer; wherein the row circuits in the plurality of row circuits respectively apply one of the 2m different analog voltages to an associated row in the array.
Abstract:
Numerous examples of an input function circuit block and an output neuron circuit block coupled to a vector-by-matrix multiplication (VMM) array in an artificial neural network are disclosed. In one example, an artificial neural network comprises a vector-by-matrix multiplication array comprising a plurality of non-volatile memory cells organized into rows and columns; an input function circuit block to receive digital input signals, convert the digital input signals into analog signals, and apply the analog signals to control gate terminals of non-volatile memory cells in one or more rows of the array during a programming operation; and an output neuron circuit block to receive analog currents from the columns of the array during a read operation and generate an output signal.
Abstract:
In one example, a method comprises determining a program resolution current value; and setting levels for a programming operation of a plurality of non-volatile memory cells in a neural network array such that a delta current between levels of each pair of adjacent cells in the plurality is a multiple of the program resolution current value.
Abstract:
Numerous examples are disclosed of a row address decoding scheme. In one example, a memory system comprises m banks of non-volatile memory cells, the m banks respectively comprising n or fewer sectors and the sectors respectively comprising p rows, and a row decoder to receive a row address comprising r bits and to identify (i) a row using the least significant t bits in the r bits, (ii) a bank using the next u least significant bits, and (iii) a sector using the next v least significant bits, where m≤2u, n≤2v, and p≤2t.
Abstract:
In one example, a method comprises erasing at the same time a word of non-volatile memory cells in an array of non-volatile memory cells arranged into rows and columns, each non-volatile memory cell comprising a word line terminal, a bit line terminal, and an erase gate terminal, by turning on an erase gate enable transistor coupled to erase gate terminals of the word of non-volatile memory cells.