Modern optical engineering, with the advent of novel optical concepts such as freeform optics and complex optical systems, requires an ever-increasing level of simulation and analysis complexity where the multiple reflections, refractions and scattering from all the optical surfaces are mandatory to faithfully describe the system in consideration. Traditional powerful optical simulation tools are bound to powerful yet fixed and hardly scalable hardware infrastructure and mostly rely on traditional CPU (or even multi-core CPU) based computing. Overcoming these barriers would enable fully comprehensive optical simulations and bring into reality for the optics industry concepts such as digital twin or companion, already available in the mechanical Computer Aided Design (CAD) realm. In recent years, the explosion of Graphical Processing Units (GPUs) uses driven by gaming and neural network applications, to name a few, have democratized extremely powerful computational capabilities particularly suitable for ray tracing. Although the use of GPUs towards optical ray tracing has started to be brought to the designer’s desktop, the massive cloud scalability of this kind of computing has, to date, remained untapped.
The 3DOptix platform overcomes these limitations by leveraging the power of cloud computing for optical simulations. By leveraging the parallel processing capabilities of GPUs, 3DOptix optical simulations can be run much faster than on traditional CPU-based systems. A benchmark of such a boost is presented in the Table below. 3DOptix shows over a 100-fold acceleration by running the same simulation on the 3DOptix cloud simulation (run on AWS-x5.xlarge GPU, Tensor Core, 24G GPU Memory, NVidia A10 Ampere) compared to a leading CPU-based simulation tool (that run on 12 th Gen Intel i7-12650H 2.3GHz 10cores, 16 threads, RAM 32G, GPU Memory 16GB). In addition to the performance benefits, 3DOptix cloud-based simulations are cost-effective as they eliminate the need to invest in expensive hardware, software, and infrastructure. By using cloud-based simulations, engineers and researchers can pay only for the resources they use, which can greatly reduce the costs of running simulations. Moreover, in an era of hybrid or remote work, cloud-based solutions naturally support access and work on simulations from anywhere, at any time, using any device with an internet connection.
