Optimization of dot product computation of two vectors using vector instructions

The objective of this is to optimize compiler generated code to compute dot product using vector instructions. Tasks to perform: 1. Use CPUID instruction to determine your processor vector processing capabilities. 2. Write C++ function to compute dot product in Visual Studio environment. Place the function in a separate file from main() that calls this function. Vector sizes should be powers of 2 (e.g. 16, 32, 64, .....512, ..216 etc.). Disable Automatic Parallelization, /Qpar, and Automatic Vectorization, /arch. Use QueryPerformanceCounter function to measure execution time. Plot graph: time versus vector size. 3. Compile code in §2. Enable Automatic Parallelization, /Qpar, and Automatic Vectorization, /arch. Use QueryPerformanceCounter function to measure execution time. Plot graph: time versus vector size. Inspect compiler generated assembly code. Observe if compiler vectorized code for very large vector sizes. Try to optimize compiler generated code. Based on compiler generated assembly code (or your optimized code) create an assembly code for dot product computation function (in the same way as shown in the text book for “clear-array example for MIPS”). Please refer to Tutorial in the Appendix. Use QueryPerformanceCounter function to measure execution time. Plot graph time versus vector size. 4. To optimize the code further use vector instruction DPPS in function that computes dot product. instruction to Improve performance of vector assembly code in §2. Use QueryPerformanceCounter function to measure execution time. Plot graph: time versus vector size. 5. Compare all plots in one figure. 6. Perform this in LINUX using gcc.