Discussion 2

• Take the time, speedup, and efficiency plots you created in Lab 1 - Mandelbrot and analyse these. Does the speedup make sense to you? Most likely your speedup curves are non linear. What characteristics (load balance, memory, IO) in the program contribute? Each group member brings this analysis to the group. Then the group discusses these. Are there differences? If so, what could cause these? Come to a group conclusion.
• Now take the reruns of Lab 2 with dynamic schedules and (see Quinn 17.7.3) experiments with chunk sizes into account. Plot the speedup for the same range of threads. How is this different from the default schedule in Lab 1? Can you explain it? Again, come to a group conclusion.
• One of the group members posts the result in main discussion board in RamCT.