AFFORDABLE COMPUTATION FOR ARCHITECTURE

Abstract

Current architectural requirements prioritize the need to minimize the ecological footprint. By taking advantage of computational design approaches like Algorithmic Design (AD), architects can enhance their design processes with analysis, optimization, and visualization mechanisms, which are critical to explore design solutions that meet this need. However, these mechanisms are also highly time- and resource-consuming, often implying a quality tradeoff or the acquisition of High-Performance Computing (HPC) machines. The latter are not yet affordable for most design studios but, fortunately, they can be contracted as a service. This paper evaluates the impact of computation as a service in architecture and, more specifically, the remote use of HPC for AD, with the aim of reducing the time and costs associated with computationally expensive processes. A set of experiments were made involving analysis, optimization, and rendering of a selected case study. Results indicate that HPC services are advantageous, particularly when performing embarrassingly parallelizable tasks such as rendering. However, some challenges remain, namely the required expertise.