White dwarf main-sequence star (WDMS) is a detached binary star system consisting of a white dwarf primary (WD) and a low-mass main-sequence (MS) companion. Despite the relatively smaller number of WDMS spectra, WDMS has positive significance to the study of evolution of close binaries, especially the evolution of common envelope (CE). The Sloan Digital Sky Survey (SDSS) released the latest data (DR12) which was the final data release of the SDSS-III, containing all SDSS observations. Based on the WDMS spectra identified in previous work, in this paper, we proposed an automatic and efficient method for measuring parameters of WDMS spectra. Least square method was used to decompose the WDMS spectra and the basic parameters including effective temperature、surface gravity and metallicity derived by template matching. Graphical processing units (GPU) was carried out to solve the problem of large computation in parallel mode. Experiment result showed that our code gave excellent performance and accuracy for spectra decomposition and had a 32.128X speedup compared with the serial CPU program.