Many different scales of experiments in neuroscience research attempt to clarify the functions of the brain. From the genetics of single molecules to the behavioral research of cognitive neuroscience, studies lead to a better understanding of nervous systems. When in vivo and in vitro experiments are impossible due to the complexity of neuronal structures or the lack of equipment, computational neuroscience provides new opportunities. Its unique access to any brain regions as well as its different levels of abstraction allows biologically-realistic neural simulations, and thus faster neuroscience advances. However, neural simulations have always involved a trade-off between execution time and biophysical realism. Even as neuron models are simplified and approximated, the neural regions of interest may require an unreasonable amount of running time. To further drive computational neuroscience research, computer scientists and engineers have created optimized simulation programs and advanced hardware architecture, respectively. We present a novel CPU/GPU simulation environment for large-scale neural modeling, called the Neocortical Simulator (NCS) version 6. At the cellular level NCS implements several built-in neuron models (e.g. Izhikevich, leaky integrate-and-fire). Computationally, shared-memory multiprocessor architectures and recent experiments with clustered GPUs indicate that we are close to simulate a million cells in real time without sacrificing biological detail. Our demonstration will be based on the design of large-scale brain models using NCS, and real-time simulations on a single or multiple machine(s).