In this project, you will implement three parallel versions of sparse matrix vec-
tor multiplication (spmv) on GPUs. SPMV is arguably the most important oper-
ation in sparse matrix computations [1]. You will exploit ne-grained parallelism
in SPMV and write a GPU program with the parallel programming interface called
CUDA.
You need to submit both code and report. The code comprises 60% and the
report comprises 40% of project 1 grade. In your report, you will need to present
the evaluation results of three different implementations as well as the strengths
and weaknesses of each implementation. You can also report any new ndings that
you have made