Zettabolt's static verification family. Four sign-off quality checkers, LINT, CDC, DFT and TXM, find RTL coding bugs, clock-domain bugs, testability problems and wrong timing exceptions without a single test vector. Catch them early, when they are cheap to fix, not after synthesis or tape-out.
Modern SoCs have hundreds of IP blocks, dozens of asynchronous clock domains, millions of scan flops and thousands of timing constraints. Simulation only finds what your testbench exercises, and STA trusts every timing exception you give it, even the wrong ones. Static verification analyses the whole design structurally and formally, so whole classes of bugs are found exhaustively and early.
LINT is the first line of defence for RTL quality. It checks every line of your Verilog and SystemVerilog against a comprehensive ruleset, so the code that reaches simulation, synthesis and your IP customers is clean, consistent and reusable.
RTL is parsed; language and coding-rule violations are flagged.
Declarations, usage, widths and connections are checked for meaning, not just syntax.
A fast built-in synthesis step finds constructs that won't synthesize, or will synthesize differently than they simulate.
The elaborated netlist is checked for structural issues such as combinational loops and undriven or floating nets.
Signals crossing between asynchronous clocks can go metastable, lose data or arrive out of order. These bugs are intermittent, hard to reproduce in simulation and often found only in silicon. CDC finds every crossing in your design and confirms each one is correctly synchronized.
Issues found are analysed and fixed in RTL, then CDC is re-run until the design is clean and ready for sign-off.
Testability problems such as blocked scan chains, uncontrolled clocks and resets, or black boxes, usually surface only after scan insertion or ATPG. Fixing them then means going back to RTL, re-synthesizing and re-inserting. DFT checks testability at RTL, estimates the fault coverage you will reach, and tells you which fixes will raise it, before DFT insertion starts.
Read libraries and RTL or netlist, then define clocks, test clocks, test signals, scan-enable and scan chains in a short Tcl constraint file.
About 300 rules check clocks, resets, scan, topology, latches, RAMs, tristates and connections. Results are graded ERROR, WARN or INFO.
Stuck-at and transition fault coverage and test coverage are calculated, with controllability and observability analysis showing exactly where and why faults can't be tested.
A coverage audit lists each problem with the fault and test coverage you would reach after fixing it, so you fix the biggest wins first.
STA assumes every path must complete in a single clock cycle. Timing exceptions, meaning false paths, multicycle paths and max/min delays, tell STA where that assumption doesn't apply. But STA can't check whether an exception is correct. It skips every path it is told to skip. TXM is a formal verification tool built for timing exceptions. It proves each one is correct, or shows you the counterexample.
The first gate. It flags non-standard SDC, wildcard and path-syntax errors, undefined clocks, and overlaps where the same path gets conflicting exceptions.
Confirms every constrained path actually exists. It flags dangling constraints and paths blocked by black boxes, combinational loops, multi-driven nets or set_disable_timing, and identifies CDC and constant-value paths.
The RTL and SDC are turned into mathematical assertions, and a satisfiability solver proves each false path is truly false and each multicycle path truly meets its cycle count. Any failure comes with a counterexample waveform.
Proves that relaxing the path can't let a glitch propagate in practice. This applies to false paths and to the extended hold window of multicycle paths.