Computational Storage Video Decoding Parallel Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video decoding processes in computing and data storage systems are inefficient, particularly when performed on general-purpose processing circuits, which can burden the host's processing and memory resources.
Innovation Solution
A computational storage device with a processing circuit configured to process multiple data units of an image in parallel, utilizing techniques such as entropy decoding, inverse discrete cosine transform, and deblocking filters, to enhance video decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If video decoding is performed on general-purpose processing circuits, then the host system can handle video processing tasks, but the processing and memory resources of the host are burdened
Solution Approach 1:
The patent extracts video decoding functionality from the host system and relocates it to a storage device. The storage device includes a processing circuit that can perform video decoding operations (entropy decoding, inverse discrete cosine transform, deblocking filtering) directly on compressed video data stored in its non-volatile memory, thereby removing the processing burden from the host while maintaining video processing capability
Solution Approach 2:
The storage device acts as an intermediary between the video data storage and the host system. Instead of the host directly processing video data, the storage device's processing circuit serves as an intermediate layer that performs decoding operations and returns decoded video to the host, reducing the host's resource consumption while enabling video processing
2Productivity
If multiple data units are processed in parallel, then video decoding throughput is improved, but the processing circuit complexity increases
Solution Approach 1:
The patent segments the video decoding process into multiple independent data unit processing operations that can be executed in parallel. Each data unit (containing image portions and their components) is processed independently through entropy decoding, inverse discrete cosine transform, and deblocking filtering stages, allowing simultaneous processing of multiple segments without requiring complex inter-dependent circuitry
Solution Approach 2:
The patent introduces parallelism as a new dimension for video decoding by processing multiple data units simultaneously rather than sequentially. The processing circuit is designed with multiple processing paths that operate concurrently on different data units, transforming the decoding process from a single-threaded sequential operation to a multi-threaded parallel operation, thereby increasing throughput without proportionally increasing sequential complexity
Data Source
AI summary
A system and method for in-storage video processing. In some embodiments, the system includes a computational storage device, the computational storage device including non-volatile storage and a processing circuit. The processing circuit may be configured to process a first data unit and to process a second data unit, in parallel with the first data unit.