Loading page
PDK Management
LiveThe PDK substrate - immutable, content-addressed process kits with lockfiles, so every run is reproducible.
Why it exists
PDKs are giant, versioned-by-folder-name, and quietly mutated. When a run can't be reproduced six months later, the kit is usually why - and nobody can prove what changed.
Silicrate manages process design kits the way modern software manages dependencies: content-addressed immutable storage, lockfiles that pin exactly what a flow consumed, and a clean API for tools to resolve kit assets - reproducibility as infrastructure, not discipline.
What it does
Every kit asset is stored by content hash - identical inputs are provably identical, and mutation is impossible by construction.
Flows record exactly which kit contents they consumed; reruns resolve the same bits, byte for byte.
Tools resolve PDK assets through a single API instead of hard-coded paths into vendor folder trees.
How it compares
Where this tool stands against the open-source path and the commercial incumbents - capability by capability.
| Capability | Silicrate | Folder trees + environment variables | Vendor-bundled kit management |
|---|---|---|---|
| Integrity | Content-hash guaranteed | Hope + naming conventions | Vendor-managed |
| Reproducibility | Lockfile-pinned | None | Partial |
| Tool interface | Uniform API | Hard-coded paths | Per-vendor |
| Deployment | Local-first | Local | Bundled with seats |
Qualitative capability comparison based on publicly documented behavior of the referenced tools. No performance figures are implied.
Keep exploring
Orchestrate your entire EDA flow as a graph - reproducible runs, convergence loops, and full visibility.
Explore FlowBit →A modern RTL synthesis suite - clean C++ core, built to be scripted, inspected, and trusted.
Explore Stella →A static timing engine with a real Python API - timing analysis you can script, query, and automate.
Explore SAC →Book a demo, or get early access to the platform.