Hardware Accelerator for Video Quality Metrics

Solution Overview

Problem

Existing video transcoding systems face challenges in efficiently computing complex video quality metrics at low power, leading to resource overheads and suboptimal video quality measurements, as they rely on software implementations of computationally intensive algorithms.

Innovation Solution

A hardware accelerator architecture is developed to efficiently compute objective video quality metrics, supporting both simpler and more complex algorithms, capable of parallel processing and optimized memory usage, including local caches for efficient pixel data reuse.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If complex video quality metrics algorithms are implemented in software, then video quality measurement accuracy is improved, but system resource consumption and power usage increase

Engineering Contradiction:

Improvevideo quality measurement accuracyVSAvoidpower consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent replaces software-based computation with a dedicated hardware accelerator that computes video quality metrics using specialized circuitry. This substitution of general-purpose software processing with purpose-built hardware significantly reduces power consumption while maintaining measurement accuracy, as the hardware accelerator can perform the same computational tasks with much lower energy requirements.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent extracts the computationally intensive video quality metric computation tasks from the general-purpose software system and isolates them into a separate hardware accelerator module. This extraction allows the complex algorithms to be implemented in dedicated hardware that operates independently from the main processor, thereby reducing overall system resource consumption and power usage.

Inventive Principle:

Principle #2Taking out (Extraction)

2Measurement precision

If complex video quality metrics algorithms are implemented in software, then video quality measurement accuracy is improved, but system resource overhead increases

Engineering Contradiction:

Improvevideo quality measurement accuracyVSAvoidsystem resource overhead

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces software-based computation with a dedicated hardware accelerator that computes video quality metrics using specialized circuitry. This substitution of general-purpose software processing with purpose-built hardware significantly reduces power consumption while maintaining measurement accuracy, as the hardware accelerator can perform the same computational tasks with much lower energy requirements.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent extracts the computationally intensive video quality metric computation tasks from the general-purpose software system and isolates them into a separate hardware accelerator module. This extraction allows the complex algorithms to be implemented in dedicated hardware that operates independently from the main processor, thereby reducing overall system resource consumption and power usage.

Inventive Principle:

Principle #2Taking out (Extraction)

3Measurement precision

If pixel data is repeatedly read from memory during computation, then computation accuracy is maintained, but memory access time and power consumption increase

Engineering Contradiction:

Improvecomputation accuracyVSAvoidmemory access time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements local caches within the hardware accelerator that pre-load and store pixel data before it is needed for computation. By performing preliminary data loading and caching operations, the system maintains computation accuracy while significantly reducing the time and power required for repeated memory accesses during the actual quality metric calculations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces local caches with higher-speed memory located close to the computation units within the hardware accelerator. This local quality improvement provides fast access to frequently used pixel data, reducing dependency on slower external memory while maintaining the accuracy required for precise video quality measurements.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12086972B2Optimizing memory reads when computing video quality metrics

Publication Date: 2024.09.10 META PLATFORMS INC

AI summary

Techniques to optimize memory reads when computing a video quality metric are disclosed. In some embodiments, an application-specific integrated circuit for computing video quality metrics includes a set of caches configured to store neighbor pixel data for edge width searches of pixels comprising a frame of a video being analyzed for a video quality metric and a kernel configured to receive corresponding neighbor pixel data for pixels comprising a current processing block of the frame from a subset of the set of caches and simultaneously perform edge width searches for pixels comprising the current processing block to determine corresponding pixel edge width values used for computing the video quality metric.