Verilog Issue Tracking

Rambus supplies its customers with Verilog source code packages. When an internal engineer uncovers an issue in shipped code, Rambus opens a ticket, identifies every delivery bundle that contains the issue, checks whether each recipient still has active support, and compiles a customer-specific report so the engineer can craft and send a notification email.

The existing workflow relied on two databases: a real-time transactional system that was accurate but took roughly 20 seconds per report, and a reporting database that was faster but had stale information.

Sembit built a comparison harness that ran the full “issue to delivery to customer” matching algorithm against both databases, materializing complete result sets for side-by-side diffing. We then streamlined the core query path and indexing strategy inside the transactional database, shrinking end-to-end report generation from about 20 seconds to about 1 second and eliminating the need for a separate reporting store.

We also integrated the harness into the CI loop, so every revision automatically produced old-versus-new output comparisons, highlighting regressions or coverage gaps whenever the matching rules were adjusted.

Engineers now run a single, authoritative report in roughly one second, removing lengthy waits. The ability to report issues rapidly helps Rambus alert customers before they tape out, preventing respin costs and giving Rambus a clear competitive edge. The automated diff framework provides clear pass/fail evidence whenever the matching logic changes, boosting confidence that no customer will be overlooked.

The Rambus engagement shows how careful query optimization and an automated comparison framework can turn a slow workflow into a fast one with redundant checks. By consolidating reporting onto a single authoritative source and verifying every change in CI, Sembit helped Rambus disclose issues quickly and confidently, protecting both customers and the bottom line.