The calculation of electronic structure is crucial for understanding the macroscopic properties of matter, but it faces significant challenges in terms of computational complexity. With the development of computing technology, scientists have adopted parallel computing strategies, especially GPU parallel computing, to address this challenge. This study optimized Hartree Fock (HF) and density functional theory (DFT) computational programs using GPU and heterogeneous computing platforms in the domestic quantum chemistry program BDF. OpenCL was used to implement cross-platform compilation, and parallel algorithms OpenMP + OpenCL and MPI + OpenMP + OpenCL were proposed. The test results show that the program is suitable for different hardware and exhibits good acceleration effects. In addition, BDF also supports time-domain density functional theory (TDDFT) calculations and has been successfully applied to predict the UV visible absorption spectra of acetic acid molecules. In future, we plan to expand the Fock matrix construction algorithm, develop more complete parallel HF and DFT algorithms, integrate BDF visualization software, and accelerate TDDFT calculations to further expand the application scope of BDF programs.