Arm Limited

Other

Adobe, Inc.

Graphics Properties Holdings, Inc.

Graphic Properties Holdings, Inc.

Danny Loh

Mark S Peercy

John Stephen Montrym

Robert A Drebin

Alexander Eugene Chalfin

Christopher J Migdal

David L Dignam

John Milligan Airey

Lingchuan Meng

John Wakefield Brothers

Ian Rudolf Bratt

Eric Kunze

Naveen Suda

Tianmu Li

Ahmet Fatih Inci

Kartikeya Bhardwaj

A method of processing input data for a given layer of a neural network using a data processing system comprising compute resources for performing convolutional computations is described. The input data comprises a given set of input feature maps, IFMs, and a given set of filters. The method comprises generating a set of part-IFMs including pluralities of part-IFMs which correspond to respective IFMs, of the given set of IFMs. The method further includes grouping part-IFMs in the set of part-IFMs into a set of selections of part-IFMs. The method further includes convolving, by respective compute resources of the data processing system, the set of selections with the given set of filters to compute a set of part-output feature maps. A data processing system for processing input data for a given layer of a neural network is also described.

Processing data for a layer of a neural network

The present disclosure advantageously provides a heterogenous system, and a method for generating an artificial neural network (ANN) for a heterogenous system. The heterogenous system includes a plurality of processing units coupled to a memory configured to store an input volume. The plurality of processing units includes first and second processing units. The first processing unit includes a first processor and is configured to execute a first ANN, and the second processing unit includes a second processor and is configured to execute a second ANN. The first and second ANNs respectively include an input layer, at least one processor-optimized hidden layer and an output layer. The second ANN hidden layers are different than the first ANN hidden layers.

Specializing neural networks for heterogeneous systems

The present disclosure advantageously provides a system and a method for convolving data in a quantized convolutional neural network (CNN). The method includes selecting a set of complex interpolation points, generating a set of complex transform matrices based, at least in part, on the set of complex interpolation points, receiving an input volume from a preceding layer of the quantized CNN, performing a complex Winograd convolution on the input volume and at least one filter, using the set of complex transform matrices, to generate an output volume, and sending the output volume to a subsequent layer of the quantized CNN.

Efficient convolutional neural networks

A digital token system and methods to provide user generated digital tokens includes receiving from a plurality of users authorization to create one or more unique tokens without approval from a central authority, wherein each user who created the unique token ('creator') is the only user authorized to increase quantity of the same token-type; and rendering the quantity of each token type visible to recipients of the token.

User generated autonomous digital token system

A floating point rasterization and frame buffer in a computer system graphics program. The rasterization, fog, lighting, texturing, blending, and antialiasing processes operate on floating point values. In one embodiment, a 16-bit floating point format consisting of one sign bit, ten mantissa bits, and five exponent bits (s10e5), is used to optimize the range and precision afforded by the 16 available bits of information. In other embodiments, the floating point format can be defined in the manner preferred in order to achieve a desired range and precision of the data stored in the frame buffer. The final floating point values corresponding to pixel attributes are stored in a frame buffer and eventually read and drawn for display. The graphics program can operate directly on the data in the frame buffer without losing any of the desired range and precision of the data.

Floating point computer system with fog

A system and method for providing Personal Cloud computing and for hosting applications and/or content may employ a network attached storage device on which virtual machine monitors (T-cups) and logical devices (Ts) are instantiated in memory. Each T may include hosted content, application modules, a server module configured to host the modules and/or content, and an interface module configured to provide access to the modules and/or content in response to detecting an authorized key. Detecting an authorized key may include communicating with a name server to determine if a T instantiated on a storage device coupled to the system is associated with a device identifier on a list of device identifiers authorized to access the module(s). The storage device may be a computer, camera, frame, phone, audio/video player, or portable storage device. The name server may be configured to authenticate Ts, define T ownership, and/or establish friend-to-friend networks between Ts.

System and method for personal cloud computing

Display system having floating point rasterization and floating point framebuffering

Growing community of inventors

Menlo Park, CA, United States of America

Danny Loh