Arm Limited

Apical Limited

Apical Limited and Arm Limited

Sharjeel Saeed

Daren Croxford

Jayavarapu Srinivasa Rao

Dominic Hugo Symes

Ozgur Ozkurt

Kushan Vijaykumar Vyas

Mathieu Jean Joseph Robart

Aaron DeBattista

Rachel Jean Trimble

John Wakefield Brothers

Jakob Axel Fries

Michael Andrew Campbell

Sean Tristram LeGuay Ellis

Graeme Leslie Ingram

Michal Bogusz

Piotr Chrobak

Jared Smolens

Oskar Lars Flordal

Jens Olson

Quinn Carter

Davide Marani

Henrik Nils-Sture Olsson

Alexey Lopich

Timothy Fawcett Milner

Chi-wen Cheng

A processing unit is provided which comprises volatile storage for storing machine learning data in binary representation, and a data processing engine communicatively coupled to the volatile storage. The processing unit is configured to selectively invert the bit values in binary representations of portions of the machine learning data when performing storage operations using the volatile storage. A computer-implemented method, and non-transitory computer-readable storage medium comprising instructions for executing the method are also provided. The method comprises receiving a request to perform a storage operation on the volatile storage using the machine learning data and performing the storage operation, including, selecting a portion of the machine learning data and inverting bit values in a binary representation of the selected portion. A computer-implemented method comprising receiving a request to store machine learning data on volatile storage and storing the machine learning data is also provided. Storing the machine learning data includes operating on at least a portion of the machine learning data to prioritize one of two potential bit values.

Selective bit inversion in storage operations for machine learning

There is provided a data processing apparatus for performing machine learning. The data processing apparatus includes convolution circuitry for convolving a plurality of neighbouring regions of input data using a kernel to produce convolution outputs. Max-pooling circuitry determines and selects the largest of the convolution outputs as a pooled output and prediction circuitry performs a size prediction of the convolution outputs based on the neighbouring regions, wherein the size prediction is performed prior to the max-pooling circuitry determining the largest of the convolution outputs and adjusts a behaviour of the convolution circuitry based on the size prediction.

Convolution size prediction to reduce calculations

An information processing apparatus is described for processing a workload. The information processing apparatus comprises a processor and a memory element connected to the processor via a data link. In advance of processing a workload, the information processing apparatus estimates an access time required to transfer an amount of the workload that is to be transferred from the external memory element to the processor, and estimates a processing time for the processor to process the workload. A processing rate characteristic of the processor and/or a data transfer rate between the memory and the processor is set in dependence upon the estimated processing time and estimated access time. Methods for varying a quality of service (QoS) value of requests to the external memory element are also described.

Information processing apparatus and a method for processing a workload

A method of compressing kernels comprising detecting a plurality of replicated kernels. The plurality of replicated kernels comprise kernels. The method also comprises generating a composite kernel from the replicated kernels. The composite kernel comprises kernel data and meta data indicative of the rotations applied to the composite kernel data. The method also comprises storing a composite kernel.

System and method for compressing kernels

A data processing system including storage. The data processing system also includes at least one processor to generate output data using at least a portion of a first neural network layer and generate a key associated with at least the portion of the first neural network layer. The at least one processor is further operable to obtain the key from the storage and obtain a version of the output data for input into a second neural network layer. Using the key, the at least one processor is further operable to determine whether the version of the output data differs from the output data.

Data processing system and method

In a graphics processing system comprising a graphics processor, a main memory, and a memory management unit, when rendering a frame that represents a view of a scene comprising one or more objects using a ray tracing process and the ray tracing process requires a traversal of a ray tracing acceleration data structure indicative of the distribution of geometry for the scene being rendered to determine geometry for the scene that may be intersected by a ray, at least part of the traversal of the ray tracing acceleration data structure is performed by the memory management unit (MMU).

Graphics processing

A method apparatus and computer readable medium for processing input data using a neural network comprising at least a first layer and a second layer. The method comprising the steps of applying a partitioning scheme to the input data, to partition the input data into a plurality of blocks, each block representing a portion of the input data. At the first layer of the neural network, the blocks of the input data are processed in a first order to generate intermediary data, wherein the intermediary data is partitioned into a plurality of intermediary blocks. At the second layer of the neural network, the intermediary blocks are processed in a second order, wherein the second order differs from the first order.

Neural network processing

Herein described is a method of operating an accumulation process in a data processing apparatus. The accumulation process comprises a plurality of accumulations which output a respective plurality of accumulated values, each based on a stored value and a computed value generated by a data processing operation. The method comprises storing a first accumulated value, the first accumulated value being one of said plurality of accumulated values, into a first storage device comprising a plurality of single-bit storage elements; determining that a predetermined trigger has been satisfied with respect to the accumulation process; and in response to the determining, storing at least a portion of a second accumulated value, the second accumulated value being one of said plurality of accumulated values, into a second storage device.

Accumulating data values and storing in first and second storage devices

Aspects of the present disclosure relate to a computer-implemented method of processing data portion. The method comprises processing a first data portion in a convolutional neural network to generate a first input to an activation function in the convolutional neural network; providing a first output by applying the activation function to the first input; and storing an indicator, representative of the first input to the activation function, for the first data portion. The method further comprises determining whether to provide a second output by applying the activation function to a second input, generated from a second data portion, based at least in part on an evaluation of the indicator for the first data portion.

Computer vision optimisation for small value changes

A computer implemented method for performing convolutions between subsets of an input data array and a kernel resulting in subsets of an output data array. The method may include receiving an input data array and using positional data indicating the position of elements of the input data array to determine subsets of the input data array which contains at least one non-zero value data element; performing convolutions between the subsets of the input data array containing at least one non-zero value data element and a kernel to produce output data array subsets; and combining the output data subsets with the positional data to generate output data indicative of a completed output data array.

Data processing method and system for performing convolutions

When a programmable execution unit of a graphics processor is executing a graphics processing program to render a frame that represents a view of a scene using a ray tracing process, and the ray tracing process requires the determination of geometry that will be intersected by a ray, the programmable execution unit sends a message to a ray tracing acceleration data structure traversal circuit of the graphics processor, for the ray tracing acceleration data structure traversal circuit to perform a traversal of a ray tracing acceleration data structure for the scene to determine geometry for the scene that may be intersected by the ray. The ray tracing acceleration data structure traversal circuit then returns to the programmable execution unit an indication of geometry that may be intersected by the ray, and the programmable execution unit uses the indicated geometry to determine any geometry that is intersected by the ray.

A data processing system for providing an output surface for display. The data processing system includes rendering circuitry operable to generate one or more input surfaces to be used for providing an output surface for display. The rendering circuitry is operable to generate a peripheral region of an input surface at a lower fidelity than the fidelity at which a central region of the input surface is generated or is operable to generate one of a plurality of input surfaces at a lower fidelity than the fidelity at which another of the plurality of input surfaces is generated. The data processing system also includes display composition circuitry operable to select part of at least one of the one or more generated input surfaces based on received view orientation data to provide an output surface for display.

Method of and data processing system for providing an output surface

A method for managing memory access for implementing at least one layer of a convolutional neural network is provided. The method comprises predicting an access procedure in relation to a portion of memory based on a characteristic of the convolutional neural network. In response to the prediction, the method comprises performing an operation to obtain and store a memory address translation, corresponding to the portion of memory, in storage in advance of the predicted access procedure. An apparatus is provided comprising at least one processor and storage. The apparatus is configured to predict an access procedure in relation to a portion of memory which is external to the processor. In response to the prediction, the apparatus is configured to obtain and store a memory address translation corresponding to the portion of memory in storage in advance of the predicted access procedure.

Managing memory access for convolutional neural networks

A method of operating a data processing system is disclosed for a data processing system that comprises a display and a display controller operable to provide to the display data in respect of output surfaces to be displayed. The method comprises, when an output surface is to be displayed, the display controller providing to the display data in respect of the output surface in the form of image data and image modification data, and the display using the image data and the image modification data when producing an output surface for display.

Providing output surface data to a display in data processing systems

A method of processing video data representative of a video comprising a first and second frame to generate output data representative of at least one feature of the second frame. The method includes identifying a first and second portion of the second frame, which correspond to a first and second portion of the first frame, respectively. First feature data obtained by processing first frame data associated with the first portion of the first frame using a first feature detection operation is retrieved from storage. Second feature data representative of a second feature map is generated by processing second frame data associated with the second portion of the second frame using the first feature detection operation. The first feature data and the second feature data are processed using a second feature detection operation to generate the output data.

Growing community of inventors

Cambridge, United Kingdom

Sharjeel Saeed