Accelerating on-device ML on Meta’s apps with ExecuTorch
WhatsApp/Messenger moved key models on-device; reduced model load & inference time and improved ANR metrics. (Engineering at Meta)
Explore 25 articles from Meta on mobile performance
Showing 20 of 25 articles (Page 1 of 2)
Accelerating on-device ML on Meta’s family of apps with ExecuTorch
ExecuTorch rollout across Instagram/WhatsApp/Messenger/Facebook improved latency and reliability; WhatsApp reports reduced model load/inference time and lower ANR metrics; several features moved on-device for performance/privacy. (Engineering at Meta)
How we think about Threads’ iOS performance
Meta fine-tunes Threads on iOS to keep it fast and frustration-free.
Translating Java to Kotlin at Scale
Meta shifted their giant Android codebase from Java to Kotlin over years.
Bringing HDR photo support to Instagram and Threads
Meta adds HDR photo support to make Instagram and Threads pop visually.
How Meta built Threads in 5 months
Meta shares how they cranked out Threads in just five wild months.
Bringing HDR video to Reels
Meta adds HDR video to Reels for sharper, richer playback vibes.
Deploying key transparency at WhatsApp
Meta rolls out key transparency to lock down WhatsApp’s security.
Why xHE-AAC is being embraced at Meta
Meta adopts xHE-AAC to pump up audio quality across their apps.
The evolution of Facebook’s iOS app architecture
Meta tells the 10-year tale of how their iOS app grew and changed.
Retrofitting null-safety onto Java at Meta
Meta’s Nullsafe tool catches Java null errors before they crash apps.
From zero to 10 million lines of Kotlin
Meta ditched Java for Kotlin across their massive Android codebase.
Improving App Startup: Facebook App
Discusses techniques Facebook implemented to improve the startup speed of its Android app.
Superpack: Pushing the limits of compression in Facebook’s mobile apps
Meta’s Superpack squeezes their apps smaller than ever before.
Hermes: An Open Source JavaScript Engine Optimized for Mobile Apps
Meta’s Hermes makes JavaScript run lean and mean on mobile devices.
MobileLab: Prevent Mobile Performance Regressions
Meta’s MobileLab stops slowdowns before they mess with users.
Profilo: Understanding App Performance in the Wild
Meta’s Profilo tracks how their apps behave in real-world chaos.
Multi-threaded rendering on Android (with Litho & Infer)
Meta’s team speeds up Android UI rendering with multi-threading magic.
Efficient and healthy background data prefetching at Instagram
Instagram grabs data early in the background to keep scrolling smooth.
Battery instrumentation at Facebook
Meta hunts down battery hogs in their apps to keep phones alive longer.