>> What made you choose D over going straight to C to work >> directly with cuda? > `faster_lmm_d` was written in D. Since, I needed to use CUDA, I had to create the D bindings. I found using CUDA for GPU computing in D as easy as doing it in C. Regards, Prasun