Tagged articles
8 articles
Page 1 of 1
Machine Heart
Machine Heart
Apr 25, 2026 · Artificial Intelligence

Jeff Dean’s New Paper Shows Elastic Large‑Scale Distributed Pre‑Training Is Now Feasible

Decoupled DiLoCo, a new distributed training framework introduced by Jeff Dean and colleagues, enables resilient large‑scale AI pre‑training across heterogeneous hardware by decoupling learners, using lightweight syncers, adaptive quorum, and balanced tensor fragmentation, dramatically improving goodput and reducing bandwidth while preserving model quality.

Bandwidth ReductionDecoupled DiLoCoDistributed Training
0 likes · 10 min read
Jeff Dean’s New Paper Shows Elastic Large‑Scale Distributed Pre‑Training Is Now Feasible
JD Cloud Developers
JD Cloud Developers
Dec 21, 2021 · Artificial Intelligence

How JD Cloud’s Mobile Super‑Resolution SDK Boosts Video Quality and Cuts Bandwidth by 30%

JD Cloud’s new mobile super‑resolution SDK leverages deep‑learning ESPCN algorithms with ROI‑based processing to upscale video streams in real time, delivering up to 80% longer playback, 30% lower bandwidth costs, and measurable quality gains demonstrated through PSNR, VMAF, and SSIM metrics.

Bandwidth ReductionDeep LearningESPCN
0 likes · 6 min read
How JD Cloud’s Mobile Super‑Resolution SDK Boosts Video Quality and Cuts Bandwidth by 30%
iQIYI Technical Product Team
iQIYI Technical Product Team
Nov 5, 2021 · Artificial Intelligence

iQIYI’s QAV1 Encoder Achieves High Compression and Bandwidth Savings Using AV1 and Deep Learning

iQIYI’s QAV1 encoder, which combines the next‑generation AV1 codec with deep‑learning techniques, delivers 20‑42% bandwidth savings and up to 36% higher compression efficiency than x265 while maintaining ultrafast 60 fps encoding speeds, enabling high‑quality 4K/8K streaming and live broadcast across devices.

AV1Bandwidth ReductionDeep Learning
0 likes · 6 min read
iQIYI’s QAV1 Encoder Achieves High Compression and Bandwidth Savings Using AV1 and Deep Learning
Alibaba Cloud Developer
Alibaba Cloud Developer
Mar 10, 2020 · Artificial Intelligence

Can Frequency‑Domain Learning Boost Image Inference Efficiency?

This article presents a system‑level approach that performs deep‑learning inference directly on JPEG frequency components, uses a gating mechanism to select important DCT coefficients, and demonstrates higher accuracy with far lower bandwidth for image classification and instance segmentation tasks.

Bandwidth ReductionComputer VisionDeep Learning
0 likes · 22 min read
Can Frequency‑Domain Learning Boost Image Inference Efficiency?
Tencent Cloud Developer
Tencent Cloud Developer
Dec 16, 2019 · Mobile Development

Optimizing GIF Usage in Information Flow: Converting GIF to MP4 and SharpP Evaluation

To address the slow loading and high bandwidth costs of GIFs in short‑content feeds, the team evaluated APNG, WebP, SharpP, and MP4, found MP4 video conversion offers the best compression, universal support and comparable CPU usage, achieving a 62% size reduction, 90% instant‑open rate and a 5.6% exposure boost.

Bandwidth ReductionGIF optimizationMP4 conversion
0 likes · 10 min read
Optimizing GIF Usage in Information Flow: Converting GIF to MP4 and SharpP Evaluation
21CTO
21CTO
Jan 29, 2018 · Fundamentals

How Tencent Cut Hundreds of Gigabytes of Bandwidth with Advanced Image Compression

This article reviews the evolution of image formats such as JPEG, WebP, HEVC, and Tencent's proprietary WXAM and SHARP, explains psychovisual JPEG optimization with Guetzli, details GPU‑accelerated performance tweaks, and shows how these techniques saved terabytes of bandwidth and reduced user download latency across Tencent's massive image platform.

Bandwidth ReductionGPU AccelerationGuetzli
0 likes · 14 min read
How Tencent Cut Hundreds of Gigabytes of Bandwidth with Advanced Image Compression
ITFLY8 Architecture Home
ITFLY8 Architecture Home
Jan 23, 2018 · Backend Development

How Tencent Scaled Video Playback to Billions: Architecture & Optimization Secrets

This article details how Tencent's QQ Space grew daily video plays from 50 million to over a billion, improving playback success to 99.92%, cutting first‑buffer time to 0.70 s and second‑buffer probability to 0.48% through a series of backend architecture, bandwidth, codec, monitoring, and client‑side optimizations.

Bandwidth ReductionPerformance OptimizationVideo Streaming
0 likes · 22 min read
How Tencent Scaled Video Playback to Billions: Architecture & Optimization Secrets