Lyra

Tool

First seen 6/10/2026

Last seen 6/10/2026

Evidence 11 chunks

NEIGHBORHOOD

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

explore full graph →

RELATIONSHIPS

21 connections

Design Under Test uses → 100% 3e

Lyra executes and verifies the DUT on FPGA.

LyraGen uses → 100% 2e

Lyra uses LyraGen as its generative model for instruction generation.

Generative Model-Based Processor Fuzzing implements → 100% 2e

Lyra implements the generative model-based processor fuzzing approach.

FPGA-Accelerated Verification implements → 100% 2e

Lyra offloads verification to FPGA for hardware acceleration.

Differential Checking implements → 100% 2e

Lyra performs hardware-level differential checking between DUT and reference model.

Lyra uses the register coverage metric for coverage collection on FPGA.

Instruction Legality Checking implements → 100% 2e

Lyra includes an instruction legality checker to validate generated instructions.

Instruction Address Correction implements → 100% 2e

Lyra includes an address correction module to prevent invalid memory accesses.

ENCORE uses → 95% 2e

Lyra's differential checking module is based on the ENCORE structure.

DiFuzzRTL ← compares with 100% 2e

Lyra is empirically compared against DifuzzRTL in coverage and throughput.

Cascade ← compares with 100% 2e

Lyra is empirically compared against Cascade in coverage and throughput.

Lyra: A Hardware-Accelerated RISC-V Verification Framework with Generative Model-Based Processor Fuzzing ← introduces 100% 2e

The paper presents Lyra as its primary contribution.

coverage convergence uses → 95% 2e

Lyra targets accelerating coverage convergence as a key goal.

Stimulus Generation uses → 100% 2e

Lyra generates high-quality instruction stimuli using its generative model.

hardware fuzzing uses → 85% 2e

Lyra generates training data using hardware fuzzing as an initial stimulus.

Instruction Set Architecture uses → 95% 2e

Lyra exploits ISA semantics for instruction generation and verification.

Reference Model uses → 100% 2e

Lyra runs the reference model concurrently with the DUT for differential checking.

Rocket Core evaluates → 100% 1e

Lyra uses RocketCore as the DUT for all experiments.

RISC-V uses → 100% 1e

Lyra is a RISC-V verification framework.

RTL uses → 85% 1e

Lyra incorporates RTL-level coverage instrumentation methods.

Control and Status Register uses → 85% 1e

Lyra incorporates control register-based coverage instrumentation.