Running LLM on smartwatch – found llama.cpp loading model twice in RAM
Found llama.cpp loading models twice in RAM — fixed with host_ptr, 74% reduction.
clone of hyperupnp client for android tv
This is a pragmatic fork: the author pared down features and redesigned the UI to work with Fire TV remotes, which actually matters for couch-to-TV UX. It’s not reinventing UPnP — it’s useful glue for people who want an easier media-browsing experience on a stick — but documentation, release builds, and a clearer changelog would make adoption frictionless.
Fire TV / Android TV users who want a simple UPnP/DLNA client, hobbyist developers and tinkerers
Found llama.cpp loading models twice in RAM — fixed with host_ptr, 74% reduction.
Bookmarks song timestamps by reading Android media notifications.
Shared inference runtime at the OS level saves RAM compared to per-app model bundling.
Necessary life support for ingress-nginx, but just a fork keeping the lights on.
Navidrome-first Android client that prioritizes smooth playback over feature bloat.
First tool bridging Apple Watch to Android without jailbreak or workarounds.