Abstract: The deterministic calculation method based on multi-group cross section has always been an important approach in the design of nuclear reactors. The accuracy of multi-group cross section directly affects the precision of nuclear reactor physics calculations. To generate high-precision cross section data for fast reactors, North China Electric Power University developed the high-precision cross section processing code MGGC2.0. This paper conducts benchmark verification and validation of the code. The infinite homogeneous mixed media UO₂, MOX, and U-TRU-Zr fuels are calculated based on the ENDF/B-VII.1 library, and the macroscopic cross sections generated by MGGC2.0 are compared with those produced by MCNP to verify the accuracy of the program in generating multi-group cross sections. The relative deviation of the macroscopic multi-group total cross section from the reference solution of MCNP is generally within 5%. Subsequently, calculations are performed for the Russian fast reactor BFS97-1 experiment, and a homogenization method of the fuel few-group cross section for various fuel arrangements is proposed. The collision probability method in MGGC2.0 is used to calculate the few-group cross section data for the fuel, and the DIF3D program is employed for core calculations. Additionally, this study compares the results obtained using different cross section homogenization methods. The research findings indicate that for BFS97-1, if the cross sections generated directly by critical search are used, the absolute deviation of the keff calculated by DIF3D from that calculated by MCNP is 2.541×10⁻². This paper improves the calculation method of axial fuel inhomogeneity, reducing the deviation to below 5.0×10⁻⁴. The deviations between the calculated results for BFS97-1, BFS97-2, BFS97-5, and BFS97-6 and the MCNP results are all within 3.0×10⁻³, validating the high accuracy of the code in generating multi-group and few-group cross section, which meets the requirements of engineering design.