Résumé
We present a fair CPU and GPU benchmark comparison of the discrete dipole approximation (DDA) codes DDSCAT, ADDA, and IFDDA by strictly aligning all numerical parameters to achieve machine-precision agreement between implementations. The results presented here are derived from our recent study and identify the practical strengths and limitations of each implementation, provide concrete best practices for users and developers, and highlight directions for improving the performance and usability of modern DDA codes.