Skip to content
STIMSMITH

Lockstep Co-Simulation Methodology

Technique
First seen 7/10/2026
Last seen 7/10/2026
Evidence 3 chunks

NEIGHBORHOOD

No graph connections found for this entity yet. It may appear in future ingestion runs.

explore full graph →

RELATIONSHIPS

28 connections
Scoreboard uses → 90% 2e
A UVM component feeds results to scoreboards as part of the lockstep co-simulation integration.
Functional Coverage uses → 88% 2e
The lockstep methodology feeds results to coverage collectors as part of the UVM integration.
UVM uses → 95% 2e
A UVM component synchronizes RTL and reference model at retirement and integrates the lockstep flow.
Configurable State Mask uses → 93% 2e
A configurable state mask is used in the lockstep methodology to control which architectural state elements are compared.
RVVI uses → 97% 2e
RVVI is used as the interface to standardize exchange of retire, state, and event information between DUT, testbench, and reference model in the lockstep methodology.
ImperasDV ← implements 97% 2e
ImperasDV provides the reference model and lockstep co-simulation methodology integrating with SV/UVM flows and RVVI.
Retirement-Level Step-and-Compare uses → 97% 2e
The lockstep co-simulation methodology uses retirement-level step-and-compare to synchronize and compare RTL and reference model states at each retire event.
Checkpoint-Based Regression ← part of 88% 1e
Checkpoints and consistent seed logging ensure repeatability for long regressions in the lockstep flow.
This paper is cited as a reference for lockstep co-simulation methodology with asynchronous events and stimulus.
This paper is cited as a reference for retirement-level lockstep co-simulation methodology and mismatch reporting.
RTL uses → 97% 1e
The lockstep co-simulation methodology runs the RTL alongside a reference model and synchronizes at instruction retirement.
Constrained-Random Instruction Streams uses → 92% 1e
A comprehensive testbench stimulus combines constrained-random instruction streams to expose real bugs in the lockstep methodology.
Directed ISA Tests uses → 92% 1e
Directed ISA tests are combined with constrained-random streams as stimulus in the lockstep methodology.
Asynchronous Events uses → 93% 1e
Asynchronous events are injected at random retire points to stress the design in the lockstep methodology.
Virtual Memory Operations evaluates → 90% 1e
The lockstep methodology stresses virtual memory operations to trigger corner-case bugs.
CSR Side Effects evaluates → 90% 1e
The lockstep methodology explores CSR side effects to trigger corner-case bugs.
Weak Memory Model evaluates → 90% 1e
The lockstep methodology tests weak versus strong memory models to trigger corner-case bugs.
SystemVerilog uses → 92% 1e
The lockstep co-simulation methodology is implemented using SystemVerilog-based co-simulation flows.
RISC-V evaluates → 97% 1e
The lockstep co-simulation methodology is applied to verify RISC-V processor designs.
Synopsys ← mentions 92% 1e
Synopsys documents the lockstep co-simulation methodology for ImperasDV with retirement-level compare and integration with VCS and Verdi.
Lockstep Co-Simulation implements → 98% 1e
The lockstep co-simulation methodology implements the lockstep co-simulation concept by running RTL alongside a reference model and comparing architectural state at each instruction retirement.
Multi-Hart Design evaluates → 85% 1e
The lockstep methodology addresses multi-hart RISC-V designs as part of its verification scope.
Instruction Retirement uses → 97% 1e
The methodology synchronizes RTL and reference model at instruction retirement points.
Architectural State uses → 97% 1e
Architectural state is captured and compared at each retire event in the lockstep methodology.
Co-Simulation Harness uses → 95% 1e
The RTL and reference model run the same programme under a co-simulation harness in the lockstep methodology.
Reference Model uses → 97% 1e
A validated reference model runs alongside the RTL in the lockstep co-simulation methodology.
Non-Deterministic Source Masking ← part of 92% 1e
Non-deterministic sources are masked or stubbed to avoid spurious mismatches in the lockstep methodology.
Seed-Based Deterministic Replay ← part of 92% 1e
Seed-based deterministic replay ensures that randomized injections remain reproducible for debugging in the lockstep methodology.