The Research Of The Block Cipher Mode Based On CPU/GPU

Currently, since NVIDIA released the first GPU (Graphic Processing Unit), the development of GPU has maintained a high speed. With the popularization of general computing API (Application Programming Interface) of GPU, as represented by CUDA (Compute Unified Device Architecture), the role of GPU in the computer will be more important, and the meaning of GPU has extended from the graphics processor to general purpose Unit. By utilizing the powerful computing capability of GPU, GPU can provide new solutions for the processing of large data.During the development of the information age, the data security plays an important role, and one of the important ways to protect data is to encrypt. But in the process of converting plaintext and ciphertext, massive computation is required. When encrypting and decrypting a large amount of data, it raises a higher demand on CPU. Therefore, the process of encryption and decryption which is the most time consuming part can be assigned to the GPU so as to effectively improve the encryption efficiency. There are a lot of studies on application of GPU to accelerating the cryptographic algorithm which are done at home and abroad, and they have shown that CPU is effective in improving the encryption efficiency, so it is important to study the block-cipher mode of operation of CPU/GPU co processing.At present, block-cipher mode of operation is the crypography scheme that uses a block-cipher to solve practical problems. A good model can make up the defects of the block cipher, while a bad work mold may bring trouble. Most of the traditional block-cipher mode can not be efficiently used on the GPU. In this paper, it focuses on the design and realization of encryption working mode under CPU/GPU, based on CUDA parallel technology. The research mainly includes the following aspects. First, it studies on the processing flow of traditional GPU computing, analyzes the reasons of applications limit of the traditional GPU computing, and makes research on the system architecture of CUDA computing architecture and experimental platform, which lays a foundation for the follow-up work of architecture. Second, it studies and analyzes the basic block-cipher mode of operation, as well as its parallel mode under CPU/GPU co processing, and makes conclusions through contrast experiments. Third, in CPU/GPU co processing, it uses the CBC as the experimental object, adopts the chaos algorithm of RCBC mode, and designs this mold to mapping scheme of CUDA based on CUDA computing architecture. From the perspective of high performance of algorithm implementation, it discusses the adoption of technique in the optimization model, and analyzes the safety and performance of the RCBC mode through the experimental comparison, as well as the advantages and disadvantages of this mode.