DeepDi

A fast and accurate x86/ARMv8 disassembler

DeepDi is built to fully utilize CPU/GPU parallelism to provide disassembly results accurately and efficiently, and currently supports x86/x64/ARMv8 ELF and PE files. It is 2x faster than IDA Pro on CPU and 350x faster on GPU*.

Features

Fast and accurate x86/x64/ARMv8 disassembler
Obfuscation resilient
GPU acceleration
Good scalability
Python interface
Produce disassembly instructions and function boundaries

Online Demo

We provide an online disassembler powered by DeepDi. It visualizes disassembly texts and control flow graphs.

Binary Release

We offer a community edition free for non-commercial use only. More details can be found in our GitHub repository.

Publication

[USENIX Security'22] Sheng Yu, Yu Qu, Xunchao Hu, and Heng Yin, DeepDi: Learning a Relational Graph Convolutional Network Model on Instructions for Fast and Accurate Disassembly, in the 31st USENIX Security Symposium, August 2022.

* Evaluated on AMD Ryzen 3900x, single-threaded, and on NVIDA RTX 2080Ti.