Dynamic Algorithm Implementaion and Tutorial - CUDA

I need a very detailed tutorial how to implement a dynamic programming algorithm (of my choose) first normally on CPU and than how to make it work faster as a praller application on a GPU. I need it to be done in CUDA. Plus comparison of the speed to the CPU implemented version. Tutorial must be very datailed, written so that person with no knowledge on programming on CUDA could make both the programs workable by him/her self. So I need a tutorial that will let me implement the algorithm normally that will work on CPU and than how to optimalize it and implement in CUDA. Step by step from the very scratch like installation of CUDA or compiler etc. So not only codes to copy and pasete should be in tutorial but also description of algorithms and optimalizations and what thay do etc. There will be a lot of comparisons in the algorithm as far as i know. The final app should first do the algorithm on CPU than on GPU and show results and compare computing times. In points: 1> create a detailed tutorial about programming CUDA in our Project. 2> analyze the Algorithm Implementation and how to program it and optimize for GPGPU 3> implement the algorytmik on CPU. 4> implement the algorithm on GPU. 5> compare and create performance function for both the implementations and also mention optimized solutions for future.