Step 9 - Performance analysis Quexal includes an advanced Code Simulator that is used to benchmark resulting code. You can analyze in depth how the optimized code performs on the Pentium 4 / Pentium III / Athlon micro architectures and identify performance hot spots. Clicking on Run|Parallelism Chart (F5) opens the following dialog: Theoretical Analysis: this graph shows how the source code would run on a processor with infinite resources (such as infinite execution units and memory accesses). It is useful to visualize the amount of Instruction Level Parallelism (ILP) and it is an upper limit on achievable performance. The Variable Usage data helps you decide if a spilling policy is needed. Code Simulator: this graph shows how the optimized code performs on the Pentium III/Athlon micro architecture. You can monitor both the decoders (blue line) and the execution units (green and red lines). Clicking on a line opens up the following window: This dialog gives plenty of detail about what's going on into the CPU: you can see which instruction are decoded, dispatched and retired in each cycle. Dispatch and Retire IP: this table shows when the instructions, scheduled as in the optimized code, are dispatched and retired.