Invasive fractional flow reserve (FFR) is a numerical value indicating the severity of the blood flow decrease in the coronary artery due to stenosis and now is the gold standard technique used to diagnose the physiological significance of the coronary artery disease (CAD). In recent years, non-invasive FFR technology based on coronary computed tomography angiography (CTA) images and computational fluid dynamics (CFD) method has made significant progress. However, this technology tend to be computationally expensive, and usually require several hours for completing analysis. In this study, we apply a highly parallel finite element method to improve the timeliness of non-invasive FFR. 101 patients underwent invasive FFR measurement and the values were compared with clinical FFR. The comparison results show that the accuracy of our non-invasive FFR technology achieves 80.2%, and the parallel efficiency of the algorithm can exceed 70% when the number of processes is more than one thousand.