Optimization of WinRAR Password Cracking Algorithm Based on Heterogeneous Computing

Different versions of WinRAR compression software use SHA-1, SHA-256 as the core and use AES-128, AES-256, and CRC32 as verification encryption. WinRAR passwords are difficult to crack, and high-efficiency hardware accelerators are usually utilized for passwords recovery. However, the unrar decompression function cannot distinguish between damaged files and incorrect passwords. The SHA-1 and SHA-256 hash algorithms have a considerable number of cycles. This paper proposes a method of password recovery based on heterogeneous computing and uses three optimization methods to solve the collision problem and accelerate the password recovery algorithm. Our experiments show that the password collision problem can be solved without reducing the password cracking speed. The optimized implementation WinRAR3 increases by 50%, while winRAR5 increases by two times.