Skip to main content
. 2013 Oct 29;4(11):2667–2672. doi: 10.1364/BOE.4.002667

Fig. 1.

Fig. 1

Diagram illustrating the method employed to interface MATLAB with the CUDA Monte Carlo forward model kernels. CUDA host code containing the CUDA kernels and calls to the CUBLAS library were compiled using NVIDIA CUDA Compiler (nvcc). The MATLAB MEX C/C++ code, wherein the CUDA binaries are called, were compiled subsequently with the MATLAB MEX compiler and utilized by the main MATLAB script.