One platform. Every layer of the yield stack.
Three functional tiers — data ingest, intelligence engine, action layer — connected on your hardware. No cloud, no change control required for basic integration.
How the three tiers connect.
Each tier is independently deployable. Start with the ingest layer and pattern recognition; add the full intelligence engine and action layer when your yield team is ready.
Data Ingest Layer
Parses KLARF files, SECS/GEM streams, SPC database exports, and FDC event logs. Normalizes coordinate systems and lot metadata into a unified schema. Typical setup: under 4 hours from first KLARF to first correlation result.
Intelligence Engine
The core analytical layer: defect pattern classification, cross-correlation of inspection coordinates with process layer history, SPC Cpk/Ppk computation, FDC event correlation, and commonality analysis across tool and chamber IDs.
Action Layer
Delivers ranked evidence packages to yield engineers via dashboard, email notification, or MES callback. Optionally writes corrective action signals back to APC via SECS/GEM. Integrates with your existing shift-change workflow.
Six capability modules, one connected platform.
Yield Intelligence Engine
Core correlation engine: yield learning curve, Pareto excursion, kill ratio, commonality analysis.
View module →Wafer Map Analysis
Spatial pattern recognition: edge ring, center spot, arc, scratch — systematic vs random classification.
View module →FDC & SPC Integration
Fault detection and classification, Cpk/Ppk live computation, EWMA filter, RtR control.
View module →Commonality Analysis
Signature analysis across chamber ID, slot position, recipe version, and operator shift.
View module →Equipment OEE Tracking
Equipment availability, performance, quality. CMP slurry, etch uniformity, photolithography overlay CDU.
View module →MES / SECS-GEM Connectors
SECS/GEM, GEM300, e-Diagnostics. Supports Workstream, Camstar, Synopsys Yield Explorer, PDF Exensio.
View module →On-premise, air-gap capable. No exceptions.
Wafercadence does not offer a cloud-hosted SaaS version of the platform. There is no multi-tenant deployment, no shared inference infrastructure, and no path by which your process data reaches an external endpoint. On-premise is the product, not an option tier.
On-premise — Fab Server
Installed on a dedicated server inside the fab's secure network. Connects to inspection tools, SPC server, and MES via SECS/GEM or file-path integration.
- Minimum hardware: 32-core server, 64GB RAM, 2TB NVMe
- GPU optional: NVIDIA A30 or equivalent for inference acceleration
- OS: RHEL 8 / Rocky Linux 8 / Ubuntu 22.04 LTS
- No outbound internet connection required
Air-gapped — Isolated Network
Full platform functionality in a completely isolated network environment. Software delivered via secured physical media. Updates via controlled transfer process.
- Same hardware requirements as standard deployment
- Software licensed offline via hardware security token
- Model updates delivered as validated release packages
- No data exfiltration path by design
Ready to see the platform in your environment?
Schedule a technical evaluation. Bring your KLARF files — we'll show you what Wafercadence finds in 30 minutes.