vcd2df-spark

Spark-native parser for VCD (Value Change Dump) files. Converts RTL simulation traces into structured DataFrames for scalable analysis.

Features

• Parses VCDs in parallel using Spark
• Emits (file, signal, tick, value) rows
• Writes output as partitioned Parquet
• Supports distributed filtering and SQL queries

Output Schema

• file: source VCD filename
• signal: register or wire name
• tick: simulation time as int
• value: signal value ('0', '1', etc.)

Assumptions

• Signals with the same encoded name across modules are treated as equivalent if their values do not diverge
• No reconstruction of module hierarchy; signal names are treated as flat strings (To simplify cross-file analysis, all signal names are treated as flat strings without reintroducing module hierarchy.)
• Non-numeric values (e.g. 'x', 'z') are mapped to -1 for compatibility
• Only value changes are recorded; unchanged signals between timestamps are omitted