I am doing a project on speed up anaylysis of a parallel program and using MPI+C on LINUX. I have coded parallel programs on Matrix multiplication and pi calculation and now I am to analyse that. My guide asked me to use MPI_Wtime but each processor gives different times(as expected) for execution. Now how do I get a single execution time from first to last, which i can print using master.
Or lemme put it this way.....What do I take as the parallel execution time? The master time (as it does the final calculation), the max time (I mean max of all processors time) or anything else??
Thanks in advance, and please don't poke at my question, just answer if you can.