The method comprises: connecting to a plurality of AI computing boards in an AI processing resource pool and a plurality of video encoding and decoding boards in a video processing resource pool by means of a unified high-speed interface; respectively allocating a specified number of AI computing boards and video encoding and decoding boards on account of resources and bandwidths required for completing a processing task to form a temporary cooperation relationship based on the processing task; in response to resource overflow or insufficiency in the AI processing resource pool or the video processing resource pool caused by a processing task change, accessing more AI computing boards or video encoding and decoding boards or stopping using redundant AI computing boards or video encoding and decoding boards; performing the processing task on account of the allocated AI computing boards or video encoding and decoding boards, and releasing the temporary cooperation relationship.

The method comprises: connecting to a plurality of AI computing boards in an AI processing resource pool and a plurality of video encoding and decoding boards in a video processing resource pool by means of a unified high-speed interface; respectively allocating a specified number of AI computing boards and video encoding and decoding boards on account of resources and bandwidths required for completing a processing task to form a temporary cooperation relationship based on the processing task; in response to resource overflow or insufficiency in the AI processing resource pool or the video processing resource pool caused by a processing task change, accessing more AI computing boards or video encoding and decoding boards or stopping using redundant AI computing boards or video encoding and decoding boards; performing the processing task on account of the allocated AI computing boards or video encoding and decoding boards, and releasing the temporary cooperation relationship.

A method of processing video data of a picture is described, the method including: allocating memory for a decoded picture in a decoded picture buffer, DPB, the decoded picture comprising pixels representing video data; receiving a bitstream comprising decoding units, DUs, and storing the DUs in a coded picture buffer, CRB, the DUs representing a coded picture that needs to be decoded into the decoded picture, each of the DUs representing a coded block of pixels; determining if, during decoding of the coded picture, at least one partial output can be performed, the at least one partial output including copying the one or more decoded DUs from the DPB to a data sink, while one or more DUs of the coded picture are not yet decoded and removed the CPB, the one or more decoded DUs representing a part of the decoded picture; and, or performing the at least one partial output if the processor determines that the at least one partial output can be performed, the performing including marking the one or more decoded DUs stored in the DPB as being ready for partial output, the marking signaling the decoder apparatus not to remove the one or more decoded DUs from the DPB; and, copying the one or more marked decoded DUs from the DPB to the data sink without removing the one or more decoded DU from the DPB.

In various examples, machine learning of encoding parameter values for a network is performed using a video encoder. Feedback associated with streaming video encoded by a video encoder over a network may be applied to an MLM(s). Using such feedback, the MLM(s) may predict a value(s) of an encoding parameter(s). The video encoder may then use the value to encode subsequent video data for the streaming. By using the video encoder in training, the MLM(s) may learn based on actual encoded parameter values of the video encoder. The MLM(s) may be trained via reinforcement learning based on video encoded by the video encoder. A rewards metric(s) may be used to train the MLM(s) using data generated or applied to the physical network in which the MLM(s) is to be deployed and/or a simulation thereof. Penalty metric(s) (e.g., the quantity of dropped frames) may also be used to train the MLM(s).

In various examples, machine learning of encoding parameter values for a network is performed using a video encoder. Feedback associated with streaming video encoded by a video encoder over a network may be applied to an MLM(s). Using such feedback, the MLM(s) may predict a value(s) of an encoding parameter(s). The video encoder may then use the value to encode subsequent video data for the streaming. By using the video encoder in training, the MLM(s) may learn based on actual encoded parameter values of the video encoder. The MLM(s) may be trained via reinforcement learning based on video encoded by the video encoder. A rewards metric(s) may be used to train the MLM(s) using data generated or applied to the physical network in which the MLM(s) is to be deployed and/or a simulation thereof. Penalty metric(s) (e.g., the quantity of dropped frames) may also be used to train the MLM(s).

An image encoding apparatus based on a system on chip (SoC) which encodes a residual block of a current block including a first value calculator configured to calculate a first value for the residual block in a space domain, a comparator configured to decide whether to transform the residual block into a transform domain according to a size of the first value, and a transform unit configured to transform the residual block into the transform domain according to a decision on whether to transform.

An image encoding apparatus based on a system on chip (SoC) which encodes a residual block of a current block including a first value calculator configured to calculate a first value for the residual block in a space domain, a comparator configured to decide whether to transform the residual block into a transform domain according to a size of the first value, and a transform unit configured to transform the residual block into the transform domain according to a decision on whether to transform.

Various innovations in media encoding are presented herein. In particular, the innovations can reduce the computational complexity of encoding by selectively skipping certain evaluation stages during encoding. For example, based on analysis of decisions made earlier in encoding or based on analysis of media to be encoded, an encoder can selectively skip evaluation of certain coding tools (such as residual coding or rate-distortion-optimized quantization), skip evaluation of certain values for parameters or settings (such as candidate unit sizes or transform sizes, or candidate partition patterns for motion compensation), and/or skip evaluation of certain coding modes (such as frequency transform skip mode) that are not expected to improve rate-distortion performance during encoding.

Systems and methods for improving determination of encoded image data using a video encoding pipeline, which includes a first transcode engine that entropy encodes a first portion of a bin stream to determine a first bit stream including first encoded image data that indicates a first coding group row and that determines first characteristic data corresponding to the first bit stream to facilitate communicating a combined bit stream; and a second transcode engine that entropy encodes a second portion of the bin stream to determine a second bit stream including second encoded image data that indicates a second coding group row while the first transcode engine entropy encodes the first portion of the bin stream and that determines second characteristic data corresponding to the second bit stream to facilitate communicating the combined bit stream, which includes the first bit stream and the second bit stream, to a decoding device.

Systems and methods for improving determination of encoded image data using a video encoding pipeline, which includes a first transcode engine that entropy encodes a first portion of a bin stream to determine a first bit stream including first encoded image data that indicates a first coding group row and that determines first characteristic data corresponding to the first bit stream to facilitate communicating a combined bit stream; and a second transcode engine that entropy encodes a second portion of the bin stream to determine a second bit stream including second encoded image data that indicates a second coding group row while the first transcode engine entropy encodes the first portion of the bin stream and that determines second characteristic data corresponding to the second bit stream to facilitate communicating the combined bit stream, which includes the first bit stream and the second bit stream, to a decoding device.