Upon consultation with the customer, we first measure the performance of the existing software and identify issues (bottlenecks) for software acceleration. We then estimate the cost of the acceleration work, implement the software acceleration, and deliver the complete set.
We will reduce the calculation volume and accelerate the calculation speed by implementing the accelerator using hardware-specific instructions to extract the performance of the accelerator.
・Acceleration of arithmetic and rendering operations in image processing and display applications using SIMD instructions
・Further acceleration of the Semi Global Matching algorithm in GPU implementations
|Development Languages||C language (C99) / C++ (11, 14, 17, 20) / C# / Python3 / CUDA, etc.|
|Frameworks & Libraries||OpenCV(3,4) / OpenGL(ES) / Windows Thread / pthread / STL, etc.|