Developer Tools●●●Banger
Clangd for CUDA Device Code
Compile CUDA for AMD GPUs with zero code changes—breaks NVIDIA lock-in.
WizardryZero to One
JonChesterfield
802mo ago
Back to browse
Dust – Device Unified Serving Toolkit (CUDA for Phones)
by ruizprogelio·Mar 2, 2026·3 points·1 comment
AI Analysis
●●●BangerShip ItSolve My ProblemNiche Gem
CUDA for phones: native runtimes, thin bridges, real demos shipping GGUF and ONNX inference.
Strengths
- •Clean architectural separation (bridge/runtime/contract) scales across iOS and Android without code duplication
- •15 open-source modules + working demos (Qwen LLM, YOLO detection) prove the infrastructure actually works today
- •Solves genuine friction in mobile ML: lifecycle management, model downloads, session pooling all first-class
Weaknesses
- •v0.1.0 signals early stage; no evidence of production adoption or handling edge cases at scale
- •Capacitor dependency narrows audience to hybrid app developers, not native-first teams
Category
Target Audience
Mobile developers, on-device ML engineers, infrastructure builders
Similar To
TensorFlow Lite · MediaPipe · ONNX Runtime
Similar Projects
AI/ML●●Solid
Smile-Serve – Inference Server for ML, ONNX, and LLM
Unifies ONNX, classic ML, and Llama 3 in one JVM server when others force separate stacks.
Niche GemShip It
haifeng
501mo ago
Infrastructure●●Solid
ZSE – Single-file LLM engine with dual INT4 kernels
INT4 inference engine beats llama.cpp on VRAM, but competing against established tools.
WizardryShip It
zyoralabs
103mo ago
Education●●Solid
Tiny-vLLM – high performance LLM inference engine in C++ and CUDA
Build vLLM from scratch with PagedAttention kernels when llama.cpp already exists.
Big BrainNiche Gem
yu3zhou4
203185d ago
AI/ML●●Solid
Mamba SSM in Rust – training and inference with custom CUDA kernels
Custom CUDA kernels for SSM recurrence with zero framework dependencies.
WizardryNiche Gem
silvermpx
102mo ago
AI/ML●●●Banger
MemFactory: Unified Inference and Training Framework for Agent Memory
LLaMA-Factory for agent memory with native GRPO and 14% performance gains.
Big BrainWizardryNiche Gem
MemTensor
801mo ago