Fig. 1. Comparison of CPU and GPU architectures with the same silicon area

Fig. 2. GPU kernel runtime method

Fig. 3. Heterogeneous accelerated computing flow

Fig. 4. Calculating 5 × 5 Zernike polynomials

Fig. 5. Performance comparison of heterogeneous linear solvers

Fig. 6. Heterogeneous computational approach for fitting face acceleration ratios

Fig. 7. Parallel peak/valley solving algorithm

Fig. 8. Performance of different algorithms for finding peak/valley value

Fig. 9. Example of the surface to be measured and the interferometer

Fig. 10. Picture of 1 024 × 1 024 resolution flatness

Fig. 11. Acceleration ratios for different steps with different resolution plane to be measured

Fig. 12. Standard deviation for different steps with different resolution plane to be measured