Recently, many researchers have developed algorithms for automatic segmentation of intravascular stent struts in Optical Coherence Tomography (OCT) images. While such algorithms are quite accurate, it takes much time to run them. Due to their long execution time, they have been unusable in hospital operating rooms for clinicians to decide whether additional balloon inflation is required. In this paper, we introduce a high-speed automatic stent segmentation algorithm using the Intel® IPP library and the NVIDIA CUDA technology. We evaluated our algorithm with 3 pullback patient data, and the average precision of the algorithm is 86.4% and the average execution time is 0.279 seconds per image. Unlike the previous algorithms starting which take OCT images as inputs, our algorithm takes binary raw data as an input and generates OCT images from it. Also, we provide 3D information with en face projection images to help medical doctors to inspect the patient data more easily and clearly.