How to answer these questions for a perflab.

Call Optimization for Filter Can estimate total # calls/pixel in original Can explain why eliminating calls reduces CPE Used inline minimal code change: 15 pts 10 pts . 10 pts Data Structure Size Can explain why char would improve over int for array . Modified code to compact data . 5 pts 10 pts Loop Traversal Order Explain order impact on stride Can identify and compute stride for inner 10 pts . . Loop Unrolling Explain why unroll hel ps: 5 pts . Explain how unroll works for different unroll amounts (including cleanup of remainder iterations) 10 pts Call Optimization for Filter Can estimate total # calls/pixel in original Can explain why eliminating calls reduces CPE Used inline minimal code change: 15 pts 10 pts . 10 pts Data Structure Size Can explain why char would improve over int for array . Modified code to compact data . 5 pts 10 pts Loop Traversal Order Explain order impact on stride Can identify and compute stride for inner 10 pts . . Loop Unrolling Explain why unroll hel ps: 5 pts . Explain how unroll works for different unroll amounts (including cleanup of remainder iterations) 10 pts