PrototypeCommon
giannisanni/pulsar
SSD-streaming inference engine for giant MoE models (Rust + CUDA). GLM 5.2 743B at 2 tok/s and Hy3 295B at 7 tok/s on two consumer 16GB GPUs. Zero-config multi-GPU: measures PCIe bandwidth, places attention and hot experts where they fit.
- Owner
- giannisanni (personal account)
- Stars
- Stars: 234
- Forks
- Forks: 28
- Language
- Rust
- License
- None
- Created
- 2026
- Topics
- cudaggufglminference-enginellmlocal-llmmixture-of-expertsmoemulti-gpuquantization
- Tier
- Prototype (100 to 249 ★) What this tier means
Nearby in Rust
All Rust repos- dreamlonglll/mini-termStars: 234RustPrototype
- aeroxy/ast-broStars: 234RustPrototype
- anomalyco/hokumStars: 234RustPrototype
- tari/llvm-sys.rsStars: 234RustPrototype
- bheisler/raytracerStars: 234RustPrototype
- ritiek/piano-rsStars: 234RustPrototype