International Business Machines Corporation

Bernard V Brezzo

Jose A Tierno

Dharmendra S Modha

Leland Chang

Daniel Joseph Friedman

Robert K Montoye

Jae-sun Seo

John Vernon Arthur

Paul A Merolla

Bipin Rajendran

Steven Kyle Esser

Yong Liu

Sameh W Asaad

Balakrishna Raghavendra Iyer

Hong Min

Parijat Dube

Bharat Sukhwani

Mathew S Thoennes

Mohit Kapur

Donna N Eng Dillenberger

Benjamin D Parker

Charles Louis Haymes

Ralph E Bellofatto

Thomas Roewer

Embodiments of the invention relate to a time-division multiplexed neurosynaptic module with implicit memory addressing for implementing a neural network. One embodiment comprises maintaining neuron attributes for multiple neurons and maintaining incoming firing events for different time steps. For each time step, incoming firing events for said time step are integrated in a time-division multiplexing manner. Incoming firing events are integrated based on the neuron attributes maintained. For each time step, the neuron attributes maintained are updated in parallel based on the integrated incoming firing events for said time step.

Time-division multiplexed neurosynaptic module with implicit memory addressing for implementing a neural network

A reconfigurable neural network circuit is provided. The reconfigurable neural network circuit comprises an electronic synapse array including multiple synapses interconnecting a plurality of digital electronic neurons. Each neuron comprises an integrator that integrates input spikes and generates a signal when the integrated inputs exceed a threshold. The circuit further comprises a control module for reconfiguring the synapse array. The control module comprises a global final state machine that controls timing for operation of the circuit, and a priority encoder that allows spiking neurons to sequentially access the synapse array.

Reconfigurable and customizable general-purpose circuits for neural networks

Embodiments of the invention relate to a time-division multiplexed neurosynaptic module with implicit memory addressing for implementing a universal substrate of adaptation. One embodiment comprises a neurosynaptic device including a memory device that maintains neuron attributes for multiple neurons. The module further includes multiple bit maps that maintain incoming firing events for different periods of delay and a multi-way processor. The processor includes a memory array that maintains a plurality of synaptic weights. The processor integrates incoming firing events in a time-division multiplexing manner. Incoming firing events are integrated based on the neuron attributes and the synaptic weights maintained.

Time-division multiplexed neurosynaptic module with implicit memory addressing for implementing a universal substrate of adaptation

Methods and arrangements for facilitating accelerations of database functions. A field programmable gate array is incorporated. At least one query control block is incorporated in the field programmable gate array, and database management system operations are accelerated via the field programmable gate array. The accelerating includes employing the at least one query control block to execute a query without reconfiguring the field programmable gate array.

Facilitating field programmable gate array accelerations of database functions

Techniques are disclosed for increasing the throughput of a multiplexed electrical bus by exploiting available pipeline stages of a computer or other system. For example, a method for increasing a throughput of an electrical bus that connects at least two devices in a system comprises introducing at least one signal hold stage in a signal-receiving one of the two devices, such that a maximum frequency at which the two devices are operated is not limited by a number of cycles of an operating frequency of the electrical bus needed for a signal to propagate from a signal-transmitting one of the two devices to the signal-receiving one of the two devices. Preferably, the signal hold stage introduced in the signal-receiving one of the two devices is a pipeline stage re-allocated from the signal-transmitting one of the two devices.

Increasing throughput of multiplexed electrical bus in pipe-lined architecture

A plurality of target field programmable gate arrays are interconnected in accordance with a connection topology and map portions of a target system. A control module is coupled to the plurality of target field programmable gate arrays. A balanced clock distribution network is configured to distribute a reference clock signal, and a balanced reset distribution network is coupled to the control module and configured to distribute a reset signal to the plurality of target field programmable gate arrays. The control module and the balanced reset distribution network are cooperatively configured to initiate and control a simulation of the target system with the plurality of target field programmable gate arrays. A plurality of local clock control state machines reside in the target field programmable gate arrays. The local clock control state machines are coupled to the balanced clock distribution network and obtain the reference clock signal therefrom. The plurality of local clock control state machines are configured to generate a set of synchronized free-running and stoppable clocks to maintain cycle-accurate and cycle-reproducible execution of the simulation of the target system. A method is also provided.

Method and infrastructure for cycle-reproducible simulation on large scale digital circuits on a coordinated set of field-programmable gate arrays (FPGAs)

Growing community of inventors

Somers, NY, United States of America

Bernard V Brezzo