Cloud-related fast processes occur over a wide range of time-space scales, and interact with one another in a complex way. Understanding and upscaling these processes occurring at the level of individual droplets to cloud scales and to the scale of climate model grid size presents a grand challenge for improving related parameterizations in climate models. The challenge becomes especially acute for those processes occurring at the current sub-LES scales (e.g., less than 100 m horizontally) due to extremely limited physical understanding and measurements, e.g., entrainment-mixing processes and turbulence-microphysics interactions. To address this challenge, the FAst-physics System TEstbed and Research (FASTER) Project has been developing a direct numerical simulation (DNS) model and explore its applications. This presentation will introduce the new DNS model, and demonstrate its potentials for developing and improving parameterizations in future high-resolution climate models. Also explored is the way to integrate the DNS with LES/CRM models to best inform parameterizations necessary at sub-LES scales.