Development and implementation of methods and tools that adequately model fundamental physics and allow credible physics-based optimization for future operational hypersonic vehicle systems are becoming more important due to requirements of ensuring their flight safety. The methods of computational fluid dynamics (CFD) are extensively applied in design and optimization of hypersonic vehicles to get more insight into complex flowfields. Computer simulation is particularly attractive due to its relatively low cost and its ability to deliver data that cannot be measured or observed. Flow discontinuities, high gradients of flow quantities, turbulence effects, flow separation and other flow features impose great demands on the underlying numerical methods. The use of Graphics Processor Units (GPUs) is a cost effective way of improving substantially the performance in CFD applications. GPU platforms make it possible to achieve speedups of an order of magnitude over a standard CPU in many CFD applications. The parallel capabilities of in-house compressible CFD code for hypersonic flow simulations are assessed and successful design of a highly parallel computation system based on GPUs is demonstrated. Possibilities of the use of GPUs for the simulation of high-speed and high-temperature flows are discussed. The results obtained are generally in a reasonable agreement with the available experimental and computational data, although some important sensitivities are identified.