Speaker: Sandor Dargo

Audience level: [ Beginner | Intermediate ]

Clean code promises readability, maintainability, and clarity. But is it possible that the pursuit of clean code comes at a catastrophic cost to performance? Some sceptics argue that adhering to clean code practices means sacrificing years - if not decades - of hardware advancements. Is clean code really that expensive? Let’s explore the complex relationship between clean code and software performance. In this talk, we’ll delve into what makes software performant. We’ll demystify the effects of clean code on performance and discover what real bottlenecks are that you’ll encounter while working on enterprise software.

But what is performance, really? While many immediately think of raw runtime speed, the landscape is far more intricate. We’ll explore different facets of performance, including compile-time efficiency, memory usage, and the performance of developers themselves. However, our primary focus will be on runtime performance.

The pivotal question we aim to answer is whether sheer runtime performance always reigns supreme. Can you truly achieve a 10x scale-up by eliminating dynamic polymorphism and virtual function calls? The very idea might seem absurd, but, believe it or not, in some specialized niches, that’s part of the solution. Still, the vast majority of us don’t work in those niches. We face different challenges.

We’ll navigate through real-life examples where awful algorithms hardly mattered, where prudent database and network management became the true path to scalability. I don’t claim that writing code that’s easy to understand and modify always yields optimal performance. Yet, more often than not, performance bottlenecks and the road to performant software lie elsewhere.