00

FuzzGPU

Tool ☆ WATCH ⎘ SHARE

First seen 9/29/2026

Last seen 9/29/2026

Evidence 10 chunks

01

NEIGHBORHOOD

12 nodes · 19 edges

graph · FuzzGPU · depth=1

02

RELATIONSHIPS

11 connections

Ported-Cascade ← compares with 93% 2e evidence

The study compares FuzzGPU against Ported-Cascade as a baseline.

Top-Level Generation Loop implements → 88% 2e evidence

FuzzGPU includes a top-level loop to coordinate multi-warp execution and barriers.

Instruction-Stream Fuzzing implements → 66% 2e evidence

FuzzGPU generates per-warp instruction streams as its fuzzing target.

Ported-DiveFuzz ← compares with 93% 2e evidence

The study compares FuzzGPU against Ported-DiveFuzz as a baseline.

SIMT Program Generator implements → 94% 2e evidence

FuzzGPU’s core is a SIMT program generator that creates valid SIMT tests.

Barrier Planner implements → 95% 2e evidence

FuzzGPU includes a deadlock-free barrier planner to coordinate synchronization.

Interleaved Warp Generation implements → 90% 2e evidence

FuzzGPU interleaves per-warp generation and pauses at memory ops to reason about conflicts.

Generation Stack implements → 95% 1e evidence

FuzzGPU uses a SIMT-inspired generation stack to manage divergence and reconvergence.

Control Flow Graph (CFG) uses → 84% 1e evidence

FuzzGPU’s generator organizes code into basic blocks connected into a CFG.

ISS-guided Generation implements → 82% 1e evidence

During generation, FuzzGPU leverages an ISS to resolve outcomes and obtain context.

Per-Warp Generation Loop implements → 88% 1e evidence

FuzzGPU implements the per-warp loop that constructs basic blocks and instruction streams.