Project Topic: Cubic spline interpolation

Instructions to the GPU programming assignment.

The purpose of this assignment is to demonstrate your ability to design and develop a parallel GPU program performing an important computational task. You will have to submit a working program and a project documentation by the following deadline using the submission link at the end of this page.

$1.$ Introduction: A short description of the computational problem, the core sequential algorithm, application areas $($if applicable$).$

$2.$ Design: Analysis of and profiling the sequential algorithm to identify performance bottlenecks, typical execution times, dependence on problem size $($operation complexity$).$ review of the literature discussing previous approaches for creating parallel implementations $($CPU and$/$or GPU: what is relevant based on your literature search$).$ Outline the design strategy you chose for the problem, Describe how you map the problem to the GPU.

$3.$ Implementation: Provide details of the implementation. Not full source code but important details related to memory management, designing for performance, data communication at device and thread level, use of hierarchies, level of parallelism envisaged, etc. $$Describe the structure and operation of your program at the kernel level $($pseudo$-$code level$)$ focusing on the parallel execution steps.

$4.$ Results: Provide measurement results $($table and plots$)$ demonstrating the operation of the problem and the achieved performance $($execution times$).$ Compare the parallel times with that obtained with the CPU implementation.$$

$5.$ Conclusions: $$Summarize your work, describe the original goals and whether and to what extent you were able to achieve them. Highlight potential future improvement opportunities.