Pars (means "leopard" in Turkish) is a parallelization library that can be used in real-time / near real-time applications to increase the performance in all the programming languages supported by .NET. It parallelizes order-dependent periodic workloads on regular multi-core CPUs, without the need for a GPU or a hardware support such as NVIDIA's CUDA.
A short demonstration: SIFT based real-time object recognition with and without Pars (Parallelizer & Stabilizer)
Atasoy, H., Yildirim, E., Yildirim, S., & Kutlu, Y. "A Real-time Parallel Image Processing Approach on Regular PCs with Multi-core CPUs", Elektronika Ir Elektrotechnika (Accepted Paper)
real-time image processing, parallel processing, parallel image processing