Developers should have an idea of the available budget for performance and understand the cost of the code that they write, not just in execution throughput but in readability, maintainability, and power efficiency. Throwing more cores at a unit of work is not nearly as good as refactoring it to be simpler.
Efficiency has become increasingly important, especially with the rise of mobile devices and cloud computing time-based usage billing. Parallelizing an inefficient algorithm may solve a performance problem in the time domain, but it's a crude brute-force approach, and altering the underlying implementation may be better.
Less is often more and sometimes doing nothing is the best approach. Software engineering is not only about knowing what to build but what not to build. Keeping things simple helps the others on your team use your work. You should aim to avoid surprising anyone with non-obvious behavior. For example, if you build an API then you should give it conventional...