摘要:
It is contemplated to realize a semiconductor memory with a large memory capacity, high in integration and low in power dissipation. A semiconductor memory is disclosed, comprising a plurality of blocks each having a memory cell array and sense amplifier(s) to differentially amplify signals read out from the array, wherein a common driving line of amplifiers composed of N-channel MOS transistors among said sense amplifiers and a common driving line of amplifiers composed of P-channel MOS transistors among the sense amplifers are connected between different blocks.
摘要:
The neural network processing system according to the present invention includes a memory circuit for storing neuron output values, connection weights, the desired values of outputs, and data necessary for learning; an input/output circuit for writing or reading data in or out of said memory circuit; a processing circuit for performing a processing for determining the neuron outputs such as the product, sum and nonlinear conversion of the data stored in said memory circuit, a comparison of the output value and its desired value, and a processing necessary for learning; and a control circuit for controlling the operations of said memory circuit, said input/output circuit and said processing circuit.
摘要:
Herein disclosed is a data processing system having a memory packaged therein for realizing a largescale and high-speed parallel distributed processing and, especially, a data processing system for the neural network processing. The neural network processing system according to the present invention comprises: a memory circuit for storing neuron output values, connection weights, the desired values of outputs, and data necessary for learning; an input/output circuit for writing or reading data in or out of said memory circuit; a processing circuit for performing a processing for determining the neuron outputs such as the product, sum and nonlinear conversion of the data stored in said memory circuit, a comparison of the output value and its desired value, and a processing necessary for learning; and a control circuit for controlling the operations of said memory circuit, said input/output circuit and said processing circuit. The processing circuit is constructed to include at least one of an adder, a multiplier, a nonlinear transfer function circuit and a comparator so that at least a portion of the processing necessary for determining the neutron output values such as the product or sum may be accomplished in parallel. Moreover, these circuits are shared among a plurality of neutrons and are operated in a time sharing manner to determine the plural neuron output values. Still moreover, the aforementioned comparator compares the neuron output value determined and the desired value of the output in parallel.
摘要:
Herein disclosed is a data processing system having a memory packaged therein for realizing a large-scale and high-speed parallel distributed processing and, especially, a data processing system for the neural network processing. The neural network processing system according to the present invention comprises: a memory circuit for storing neuron output values, connection weights, the desired values of outputs, and data necessary for learning; and input/output circuit for writing or reading data in or out of said memory circuit; a processing circuit for performing a processing for determining the neuron outputs such as the product, sum and nonlinear conversion of the data stored in said memory circuit, a comparison of the output value and its desired value, and a processing necessary for learning; and a control circuit for controlling the operations of said memory circuit, said input/output circuit and said processing circuit. The processing circuit is constructed to include at least one of an adder, a multiplier, a nonlinear transfer function circuit and a comparator so that at least a portion of the processing necessary for determining the neutron output values such as the product or sum may be accomplished in parallel. Moreover, these circuits are shared among a plurality of neutrons and are operated in a time sharing manner to determine the plural neuron output values. Still moreover, the aforementioned comparator compares the neuron output value determined and the desired value of the otuput in parallel.
摘要:
A data processing system has a memory for realizing large-scale and high-speed parallel distributed processing and, especially, a data processing system for neural network processing. The neural network processing system comprises: a memory circuit for storing neuron output values, connection weights, the desired values of outputs, and data necessary for learning; an input/output circuit for writing or reading data in or out of said memory circuit; a processing circuit for performing a processing for determining the neuron outputs such as the product, sum and nonlinear conversion of the data stored in said memory circuit, a comparison of the output value and its desired value, and a processing necessary for learning; and a control circuit for controlling the operation of the memory circuit, the input/output circuit and the processing circuit. The processing circuit includes at least one of an address, a multiplier, a nonlinear transfer function circuit and a comparator so that at least a portion of the processing necessary for determining the neuron output values such as the product of sum may be accomplished in parallel. Moreover, these circuits are shared among a plurality of neurons and are operated in a time sharing manner to determine the plural neuron output values. Still moreover, the aforementioned comparator compares the neuron output value determined and the desired value of the output in parallel.
摘要:
Herein disclosed is a data processing system having a memory packaged therein for realizing a large-scale and high-speed parallel distributed processing and, especially, a data processing system for the neural network processing. The neural network processing system according to the present invention comprises: a memory circuit for storing neuron output values, connection weights, the desired values of outputs, and data necessary for learning; an input/output circuit for writing or reading data in or out of said memory circuit; a processing circuit for performing a processing for determining the neuron outputs such as the product, sum and nonlinear conversion of the data stored in said memory circuit, a comparison of the output value and its desired value, and a processing necessary for learning; and a control circuit for controlling the operations of said memory circuit, said input/output circuit and said processing circuit. The processing circuit is constructed to include at least one of an adder, a multiplier, a nonlinear transfer function circuit and a comparator so that at least a portion of the processing necessary for determining the neutron output values such as the product or sum may be accomplished in parallel. Moreover, these circuits are shared among a plurality of neutrons and are operated in a time sharing manner to determine the plural neuron output values. Still moreover, the aforementioned comparator compares the neuron output value determined and the desired value of the output in parallel.
摘要:
Herein disclosed is a data processing system having a memory packaged therein for realizing a large-scale and high-speed parallel distributed processing and, especially, a data processing system for the neural network processing. The neural network processing system according to the present invention comprises: a memory circuit for storing neuron output values, connection weights, the desired values of outputs, and data necessary for learning; an input/output circuit for writing or reading data in or out of said memory circuit; a processing circuit for performing a processing for determining the neuron outputs such as the product, sum and nonlinear conversion of the data stored in said memory circuit, a comparison of the output value and its desired value, and a processing necessary for learning; and a control circuit for controlling the operations of said memory circuit, said input/output circuit and said processing circuit. The processing circuit is constructed to include at least one of an adder, a multiplier, a nonlinear transfer function circuit and a comparator so that at least a portion of the processing necessary for determining the neutron output values such as the product or sum may be accomplished in parallel. Moreover, these circuits are shared among a plurality of neutrons and are operated in a time sharing manner to determine the plural neuron output values. Still moreover, the aforementioned comparator compares the neuron output value determined and the desired value of the output in parallel.
摘要:
A semiconductor device comprising a plurality of circuits driven by at least one external power source, and at least one voltage converter transforming the voltage of the external power source into another voltage. At least a part of the plurality of circuits are driven by the output voltage of the at least one voltage converter, which is provided with a controller for controlling its load driving power, corresponding to the operation of the part of the plurality of circuits. The voltage converter includes a voltage limiter which is used exclusively for each of the different natures of the loads, and its operation and load driving power are controlled, depending on the operations of each of the loads.
摘要:
A very highly integrated semiconductor memory which enables the dynamic random access memory to develop less soft error and to eliminate margin for aligning the masks, that hinders the device from being highly integrated. The memory cell capacitor is constituted by a trench which is provided at a position defined by an insulator formed on the side of gate electrode of a MOS transistor that constitutes the memory cell. Therefore, the MOS transistor and the trench capacitor are self-aligned, and no margin is required for alignment.
摘要:
A sense amplifier capable of performing high-speed data sense operation with lower power consumption using a minuscule signal from a memory cell even in a case where a memory array voltage is reduced. A plurality of drive switches for over-driving are distributively arranged in a sense amplifier area, and a plurality of drive switches for restore operation are concentratively disposed at one end of a row of the sense amplifiers. A potential for over-driving is supplied using a meshed power line circuit. Through the use of the drive switches for over-driving, initial sense operation can be performed on data line pairs with a voltage having an amplitude larger than a data-line amplitude, allowing implementation of high-speed sense operation. The distributed arrangement of the drive switched for over-driving makes it possible to dispersively supply current in sense operation, thereby reducing a difference in sense voltage with respect to far and near positions of the sense amplifiers.