Abstract: A processor, method, and non-transitory computer-readable storage medium for processing template data and search data according to a search window applied to the search data. The search window comprising a set of offset positions. The processing is performed by a block matching engine (BME) that produces a tensor with difference values, and a convolutional engine (CE) that performs a convolutional operation on the tensor. The processing is performed in an iterative interleaved fashion, by dividing the set of offset positions into a plurality of subsets of offset positions. In parallel with processing of the first X offset positions by the CE, the BME generates the next X channels of the tensor, and which are subsequently pipelined through to the CE via an internal storage, etc.

Abstract: A processor, method, and non-transitory computer-readable storage medium for performing a block matching between first data and second data are provided. The block matching is performed using an iterative process, wherein, for each iteration, a portion of the first data and a corresponding portion of the second data is selected using a sliding window approach. When differences data, used for block matching, is calculated for a specific subset of first and second data, many of the calculations overlap with those needed for a nearby subset. Summed area table, SAT, data used for determining the difference data is continuously stored and updated in a buffer, such that overlapping computations can be avoided.

Abstract: Apparatuses, methods, and non-transitory computer-readable media are disclosed. One example concerns a reduce interpolation channel-wise instruction to trigger a reduce interpolation channel-wise operation. The reduce interpolation channel-wise operation comprises: selecting a pair of vectors from a range of source vectors in dependence on a first portion of each element of an interpolation vector; a weighted addition of that element from the pair of vectors, wherein a weighting of the weighted addition is dependent on a second portion of that element of the interpolation vector; and storing a result of the weighted addition in that element of a destination vector.

Abstract: Provided is a data stream processor comprising a streamed data transceiver interface, a structure of processing units configurable to transform data received from a data source over the streamed data transceiver interface according to a specified output requirement, and a configuration unit operable in electronic communication with a data consumer to receive an output requirement and to configure the operation and linkage of a processing unit in the structure of processing units to transform input data to output data according to the specified output requirement; wherein the structure of processing units is further operable to provide the output data for output over the streamed data transceiver interface.

Abstract: Provided is a graphics processing unit comprising a texture unit, an execution unit, and a machine-learning neural network engine, all configured in a pipeline in electronic communication with an integrated cache memory; and a visual data processing engine comprising a configurable stencil processor integrated into the pipeline, in electronic communication with the integrated cache memory, and configured to execute repetitive image-to-image processing instructions on visual data fetched from the integrated cache memory; wherein a graphics processing unit scheduler is configured to provide a job control function for the visual data processing engine; and wherein the visual data processing engine is configured responsively to the graphics processing unit scheduler to operate in parallel with at least one of the texture unit, the execution unit, or the machine-learning neural network engine using a separate dataflow.

Abstract: Provided is a data stream processor comprising: a configurable compute unit comprising plural processing units each configured to receive at least one portion of input data and process the at least one portion of a repetitive arithmetical/logical operation on the data; an input memory unit in electronic communication with the configurable compute unit and configured to supply at least one portion of the input data to at least one of the plural processing units in the configurable compute unit; and at least one accumulator unit in electronic communication with the configurable compute unit and configured to receive at least two portions of processed data from the configurable compute unit and to output accumulated data; wherein each of the plural processing units is further configured to forward its processed data to a next processing unit and/or to an accumulator unit.

Abstract: Provided is a data stream processor comprising a streamed data transceiver interface, a structure of processing units configurable to transform data received from a data source over the streamed data transceiver interface according to a specified output requirement, and a configuration unit operable in electronic communication with a data consumer to receive an output requirement and to configure the operation and linkage of a processing unit in the structure of processing units to transform input data to output data according to the specified output requirement; wherein the structure of processing units is further operable to provide the output data for output over the streamed data transceiver interface.

Abstract: In a data processing system, when displaying a foveated image, a producer processing unit generates plural different resolution versions of the frame to be displayed. A display processor then generates a view orientation transformed output version of the frame to be displayed using data from the plural different resolution versions of the frame to be displayed generated by the producer processing unit based on data indicative of which resolution version of the frame is to be used for respective regions of the view orientation transformed output version of the frame to be displayed provided to the display processor.

Abstract: A display controller 93 in a data processing system includes a timewarp module (transformation circuitry) 100 that is operable to perform timewarp processing of a rendered frame 92 generated by a graphics processor (GPU) 91 for provision to a display panel 94. The timewarp module (transformation circuitry) 100 operates to transform an input surface 92 read by the display controller 93 based on received view orientation data to provide an appropriately “timewarped” transformed version of the input surface as an output transformed surface for display on the display 94.

Abstract: A method comprising, in an image processing operation, identifying location data indicative of a read path for the image processing operation, the read path at least partly traversing a block of pixels of an image. Parameter data relating to a characteristic of the read path in the context of the block is generated from the location. Storage prioritization data is associated with the block at least partly on the basis of the parameter data. The storage prioritization data is for determining whether block data representative of the block is to be evicted from storage.

Abstract: When displaying a frame that is to be transformed based on a predicted view orientation, the frame is transformed based on a predicted view orientation by, for each of plural different regions of the frame, determining a predicted view orientation transformation matrix to use to transform the region of the frame by interpolating between a first predicted view orientation matrix that corresponds to a transformation for a predicted view orientation for a first time during the display of the frame and a second predicted view orientation matrix that corresponds to a predicted view orientation for a second, later time during the display of the frame; and then using the interpolated predicted view orientation transformation matrix determined for the region of the frame to transform the region of the frame. In this way each different region of the frame is subjected to a different predicted view orientation transformation.

Abstract: In a data processing system, when displaying a foveated image, a producer processing unit generates plural different resolution versions of the frame to be displayed. A display processor then generates a view orientation transformed output version of the frame to be displayed using data from the plural different resolution versions of the frame to be displayed generated by the producer processing unit based on data indicative of which resolution version of the frame is to be used for respective regions of the view orientation transformed output version of the frame to be displayed provided to the display processor.

Abstract: When displaying a frame that is to be transformed based on a predicted view orientation, the frame is transformed based on a predicted view orientation by, for each of plural different regions of the frame, determining a predicted view orientation transformation matrix to use to transform the region of the frame by interpolating between a first predicted view orientation matrix that corresponds to a transformation for a predicted view orientation for a first time during the display of the frame and a second predicted view orientation matrix that corresponds to a predicted view orientation for a second, later time during the display of the frame; and then using the interpolated predicted view orientation transformation matrix determined for the region of the frame to transform the region of the frame. In this way each different region of the frame is subjected to a different predicted view orientation transformation.

Abstract: A method comprising, in an image processing operation, identifying location data indicative of a read path for the image processing operation, the read path at least partly traversing a block of pixels of an image. Parameter data relating to a characteristic of the read path in the context of the block is generated from the location. Storage prioritization data is associated with the block at least partly on the basis of the parameter data. The storage prioritization data is for determining whether block data representative of the block is to be evicted from storage.

Abstract: A method including receiving input data representing an input frame of a video. Transformation data representing at least one transformation for applying to an image to adjust a geometric distortion of the image is received. A first and second plurality of input tiles, each including a respective portion of the input frame, is processed with at least part of the transformation data, thereby generating first and second sub-window data representing a first and second sub-window of an output frame of the video, respectively.

Abstract: A storage system comprising storage including a plurality of storage banks and a storage controller to control writing of received pixel data to the storage banks in a distribution pattern. The distribution pattern includes pattern sections, each of which corresponds to pixels from a row of input pixels, and pattern blocks, each of which corresponds to pixels from a plurality of adjacent rows and columns of the input pixels. Pixel data in a pattern section and a pattern block are each written to different ones of the storage banks. The pattern section includes a first section part, which overlaps the pattern block and a second section part, which does not overlap the pattern block. Pixel data of the first section part is written to a first set of storage banks and pixel data of the second section part is written to a second, different, set of storage banks.

Abstract: A storage system comprising storage including a plurality of storage banks and a storage controller to control writing of received pixel data to the storage banks in a distribution pattern. The distribution pattern includes pattern sections, each of which corresponds to pixels from a row of input pixels, and pattern blocks, each of which corresponds to pixels from a plurality of adjacent rows and columns of the input pixels. Pixel data in a pattern section and a pattern block are each written to different ones of the storage banks. The pattern section includes a first section part, which overlaps the pattern block and a second section part, which does not overlap the pattern block. Pixel data of the first section part is written to a first set of storage banks and pixel data of the second section part is written to a second, different, set of storage banks.

Abstract: A method including receiving input data representing an input frame of a video. Transformation data representing at least one transformation for applying to an image to adjust a geometric distortion of the image is received. A first and second plurality of input tiles, each including a respective portion of the input frame, is processed with at least part of the transformation data, thereby generating first and second sub-window data representing a first and second sub-window of an output frame of the video, respectively.

Abstract: A display controller 93 in a data processing system includes a timewarp module (transformation circuitry) 100 that is operable to perform timewarp processing of a rendered frame 92 generated by a graphics processor (GPU) 91 for provision to a display panel 94. The timewarp module (transformation circuitry) 100 operates to transform an input surface 92 read by the display controller 93 based on received view orientation data to provide an appropriately “timewarped” transformed version of the input surface as an output transformed surface for display on the display 94.