GPU Variable-Bitrate Stream Interleaving With Low Metadata Overhead

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data compression techniques for graphics processing units (GPUs) face challenges in efficiently compressing and decompressing large amounts of texture data, particularly due to the need for significant metadata to manage interleaved bitstreams, which increases data overhead and reduces compression efficiency.

Innovation Solution

The proposed solution involves interleaving variable bitrate streams produced by cascaded dictionary and entropy coders for GPU implementations, while minimizing metadata overhead and reducing GPU divergence and inefficient memory access patterns by fusing entropy and dictionary decoding stages.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If variable bitrate streams are interleaved for GPU compression, then compression efficiency is improved, but metadata overhead increases significantly

Engineering Contradiction:

Improvecompression efficiencyVSAvoidmetadata overhead

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The patent extracts and removes metadata from the interleaved bitstream processing pipeline entirely. Instead of managing variable bitrate streams with metadata headers, the invention uses fixed bitrate streams where each decoder receives predetermined, constant-rate data, eliminating the need for metadata that tracks variable bitrates and stream boundaries.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the bitrate parameter from variable to fixed. By converting variable bitrate streams into fixed bitrate streams, the system eliminates metadata overhead while maintaining compression efficiency through the interleaved structure of multiple decoders processing constant-rate data in parallel.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If separate entropy and dictionary decoding stages are used, then decoding flexibility is improved, but GPU divergence and memory access inefficiency increase

Engineering Contradiction:

Improvedecoding flexibilityVSAvoidGPU divergence

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges the separate entropy decoding and dictionary decoding stages into a unified, integrated decoding process. This fusion allows both decoding operations to occur simultaneously within a single GPU kernel, eliminating GPU divergence caused by sequential processing and improving memory access efficiency by reducing intermediate data transfers.

Inventive Principle:

Principle #5Merging (Combining)

3Adaptability or versatility

If separate entropy and dictionary decoding stages are used, then decoding flexibility is improved, but processing speed decreases

Engineering Contradiction:

Improvedecoding flexibilityVSAvoidprocessing speed

Core Design Contradiction:

Adaptability or versatilityVSSpeed

Solution Approach 1:

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent ensures continuous useful action by allowing entropy decoding and dictionary decoding to proceed simultaneously without interruption. The fused architecture enables both decoders to operate in parallel on the same data stream, maintaining continuous processing flow and eliminating idle time between decoding stages.

Inventive Principle:

Principle #20Continuity of useful action

Data Source

PatentUS12299940B2Interleaving of variable bitrate streams for GPU implementations

Publication Date: 2025.05.13 INTEL CORP

AI summary

Interleaving of variable bitrate streams for GPU implementations is described. An example of an apparatus includes one or more processors including a graphic processor, the graphics processor including a super-compression encoder pipeline to provide variable width interleaved coding; and memory for storage of data, wherein the graphics processor is to perform parallel dictionary encoding on a bitstream of symbols one of multiple workgroups, the workgroup to employ a plurality of encoders to generate a plurality of token-streams of variable lengths; create a histogram including at least tokens from the plurality of token-streams for the workgroup to generate an optimized entropy code; entropy code each of the plurality of token-streams for the workgroup into an encoded bitstream; and variably interleave the encoded bitstreams to generate an interleaved bitstream and bookkeep a size of the interleaved bitstream.