Skip to content
STIMSMITH

RISCSmith

Tool
First seen 7/1/2026
Last seen 7/1/2026
Evidence 16 chunks

NEIGHBORHOOD

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

explore full graph →

RELATIONSHIPS

31 connections
riscv-dv compares with → 100% 3e
RISCSmith is directly compared with RISCV-DV in experiments, outperforming it in bugs found and coverage.
Source Instrumentation implements → 100% 3e
RISCSmith instruments RISC-V implementation source with lightweight logging hooks into the RTL.
Per-Instruction Logging implements → 100% 3e
RISCSmith records per-instruction register updates, memory writes, and exceptions with minimal overhead during simulation.
Instruction Modeling implements → 100% 3e
RISCSmith automatically builds rich instruction models by parsing the RISC-V UnifiedDB.
Rich Instruction Construction implements → 100% 3e
RISCSmith implements rich instruction construction as a core technique for generating high-quality test programs.
Register and Memory Initialization implements → 100% 3e
RISCSmith automatically constructs programs that initialize register and memory contexts.
RISC-V UnifiedDB uses → 100% 3e
RISCSmith uses the RISC-V UnifiedDB as its primary data source for automated instruction model construction.
On-the-fly Differential Analysis implements → 100% 3e
RISCSmith implements on-the-fly differential analysis to detect behavioral divergences between CPU implementations.
Cascade ← compares with 100% 3e
RISCSmith is directly compared with Cascade in experiments, outperforming it in bugs found and coverage.
Strongly Typed Instruction Model implements → 100% 3e
RISCSmith constructs strongly typed instruction models that capture operand roles, type constraints, and runtime semantics.
Rocket evaluates → 100% 2e
RISCSmith was evaluated on Rocket and discovered bugs in it.
Differential Testing implements → 100% 2e
RISCSmith conducts differential testing through a three-stage pipeline comparing target implementations against reference models.
XiangShan evaluates → 100% 2e
RISCSmith was evaluated on XiangShan and discovered bugs in it.
Verilator uses → 100% 2e
RISCSmith uses Verilator to compile and simulate instrumented RISC-V implementations.
spike uses → 100% 2e
RISCSmith uses Spike as the golden reference model for differential comparison.
Kronos evaluates → 100% 2e
RISCSmith was evaluated on Kronos and discovered bugs in it.
RISC-V uses → 100% 2e
RISCSmith targets RISC-V CPUs and uses RISC-V instruction specifications for modeling.
RTL uses → 100% 2e
RISCSmith instruments RTL implementations and simulates them to detect bugs.
pre-silicon verification uses → 90% 2e
RISCSmith is used in the context of pre-silicon verification to detect bugs before tape-out.
PicoRV32 evaluates → 100% 2e
RISCSmith was evaluated on PicoRV32 and discovered bugs in it.
The paper proposes and introduces RISCSmith as a new fuzzing framework for detecting RISC-V CPU bugs.
CVA6 evaluates → 100% 2e
RISCSmith was evaluated on CVA6 and discovered multiple bugs in it.
VexRiscv evaluates → 100% 2e
RISCSmith was evaluated on VexRiscv and discovered bugs in it.
Inter-Instruction Dependency implements → 100% 2e
RISCSmith models inter-instruction data dependencies in generated programs.
CPU uses → 90% 1e
RISCSmith targets CPU implementations as the subject of its fuzzing.
Binary Search-Based Minimization implements → 100% 1e
RISCSmith applies a binary search-based minimization algorithm to isolate the shortest instruction sequence that triggers a bug.
Exception Carry-on Template implements → 90% 1e
RISCSmith inserts exception carry-on templates as part of its test program generation.
GNU RISC-V Toolchain uses → 95% 1e
RISCSmith uses the GNU RISC-V toolchain to assemble and compile generated test programs.
Hardware fuzzing uses → 100% 1e
RISCSmith is a hardware fuzzing framework targeting RISC-V CPUs.
Coverage-guided Fuzzing uses → 80% 1e
RISCSmith collects coverage data to measure and improve test effectiveness.
formal verification implements → 50% 1e
The paper positions RISCSmith as an alternative to formal verification, which is costly.