Streamable Neural Video Compression: A Mixed Precision Approach for Cross-Platform Deployment

2026-08-01

AI Summary

This work addresses the practical deployment challenge of neural video codecs, whose cross-platform entropy decoding often desynchronizes due to floating-point non-determinism. To resolve this, the authors propose a streamable client-server neural video compression architecture featuring an innovative mixed-precision strategy: FP32 is employed at key frames and feature reset points to guarantee deterministic synchronization across diverse GPUs, while FP16 is used for all other P-frames to leverage efficient hardware acceleration. This approach maintains high compression efficiency and low latency while overcoming the performance limitations of integer-only quantization schemes. Experiments demonstrate seamless decoding across 12 GPUs spanning four generations, effectively eliminating intra-generation fragmentation; the 1080p compression efficiency loss is negligible, and end-to-end streaming feasibility is validated over real-world networks including Wi-Fi 6, 5G, and Starlink D2C.

Abstract

Neural Video Codecs (NVCs) offer unprecedented rate-distortion performance, making them highly attractive for bandwidth-constrained environments like 5G cellular networks and emerging satellite direct-to-cell (D2C) links. However, deploying NVCs in real-world streaming applications is severely hindered by cross-platform floating-point non-determinism, which causes arithmetic entropy coders to desynchronize and crash across different GPU architectures. While recent integer-based quantization methods address this, they incur either massive degradation in compression efficiency (INT8) or severe computational bottlenecks by bypassing hardware acceleration (INT16). In this paper, we propose a streamable, client-server NVC architecture featuring a novel Mixed Precision (FP16/FP32) strategy. By strategically executing P-frames in hardware-accelerated FP16 for real-time throughput, while forcing I-frames and periodic feature-adapter resets to IEEE-754 compliant FP32, we guarantee deterministic synchronization at critical boundaries. Through extensive cross-encode/decode evaluations across 12 GPUs spanning four architectural generations, we demonstrate that our approach successfully eliminates intra-generation fragmentation and substantially broadens cross-die interoperability, achieving seamless cross-generation decodability for recent architectures at 1080p. Crucially, this is achieved with a negligible impact on compression efficiency. Furthermore, we evaluate the system's end-to-end latency across diverse real-world networks, including Wi-Fi 6, 5G NR (FDD/TDD), and Starlink D2C, proving the practical viability of streamable learned video compression while highlighting unique challenges in Non-Terrestrial Networks.

Problem

• Neural Video Compression
• Cross-platform Deployment
• Floating-point Non-determinism
• Streamable Video Coding
• Entropy Coder Desynchronization

Innovation

• Mixed Precision
• Neural Video Compression
• Cross-Platform Determinism
• Streamable Codec
• Hardware-Accelerated FP16