Step 7 - Compiling code The Quexal development environment is not only about easing MMX/SSE programming: it will turn every programmer into an expert code optimizer, without requiring any previous experience. Scheduling instructions by hand is a boring and tiresome job, and it requires a deep knowledge of the micro architectures. Optimized code is also difficult to read and maintain, since it is common to join different execution paths together to maximize the level of parallelism that can be achieved: the resulting code does not show a clear logical flow. Even worse, modifying an optimized piece code often turns out to be impossible, since any change would likely break the carefully planned scheduling strategy: it is often necessary to dump the optimized code, modify the original code then repeat the scheduling phase from scratch. All these tasks greatly reduce the productivity of MMX/SSE programmers. Quexal provides an optimizing compiler that can quickly turn the original code into highly tuned assembly instructions, ready to be pasted into any development environment, even those that do not support MMX and SSE opcodes thanks to the bytecode compiler. The benefits of removing the optimization phase from the development process cannot be underestimated: the learning curve is dramatically flattened: you can ignore most of highly technical micro architectural details, because Quexal does the job for you; the development process is shorter: aside from the gain of time obtained by avoiding the optimization phase, you can easily maintain the source code in its original form using the comfortable Quexal development environment, and optimize it each time you need to; the resulting code is less likely to have hard-to-catch bugs, due to incorrect dependency analysis or register misallocation. By following a few guidelines, you can be sure that the optimized code behaves exactly as the original one does. Performing the optimization of the code built in the previous section is as easy as clicking on the red light button on the toolbar, or choosing Run|Compile (F9) from the menu. The compiler will keep running until you stop it by clicking again on the same button. While compiling, you can see some meaningful statistics on the status bar: the most important one is the Best Benchmark that shows how many solutions provided a performance improvement over the first found one, and how does the best solution perform on the Pentium 4 / Pentium III / Athlon micro architecture. When the source code has been optimized, you can examine it by clicking on the finger button on the toolbar (or Run|Show Optimized Code from the menu, F10). A new window, named Compiled optimized code, pops up showing the assembly code ready to be pasted into any development environment: clicking on the Copy button copies all the optimized code into the Windows clipboard.