240
edits
Changes
→Pi
With regards to the hotspots, both programs contain one area that holds over 99% of their total execution time; these areas will be the focus for parallelization. The programs were revised to remove all data dependencies, thus there should be no conflicts between threads when the parallel codes are applied.
Note: The compiler on the test system appears to be using the C++11 standard; notice the accuracy compared to the samples taken from Matrix.