X-ray diffraction tomography (XDT) is used to probe the material composition of objects providing improved contrast between materials compared with conventional transmission-based computed tomography (CT). Current challenges presented with XDT include long image acquisition and simulation time. To accelerate the simulation speed, our approach is to adopt NVIDIA’s OptiX ray-tracing engine, a parallelized pipeline for graphics processing units (GPUs), to perform XDT simulations on objects by making use of the innovative transformation from conventional 3-D physical space into a 2-D quasi-reciprocal space. The advantage is that ray tracing in this domain requires only 3-D mesh objects, yielding calculations without the need for voxels. The simulated XDT projections demonstrate high consistency with voxel models, with a normalized mean square difference less than 0.66%, and the ray-tracing time is two orders of magnitude less than the previously reported voxel-based GPU ray-tracing results. Due to an accelerated simulation time, XDT projections of objects with three spatial dimensions (4-D tensor) have also been reported, demonstrating the feasibility for large-scale high-dimensional tensor tomography simulations.