Why Computational Complexity Matters and Optimization Does Not

When developers hear word "optimization" they think of:

  • assembler and C/C++
  • ultra-optimizing compilers
  • JIT, native code generation
  • high-price hardware

Meanwhile:

  • Google is run by semi-interpreted language, on commodity hardware
  • Swiftfox lags behind Google Chrome, despite of MMX/SSE/3Dnow support
  • Saxon outperforms MSXML and XslCompiledTransform
  • IE9 will be slow despite graphics card acceleration

Optimization and speed is not about assembler and other tricks, it is about reducing computational complexity. Reducing computational complexity is about improving algorithms by using hash-functions, indexes, quick sort, etc.

Bubble-sort implemented in utterly optimized compiled language will never be as fast as quick-sort implemented in puppy interpreted language.

Comments

Post new comment

The content of this field is kept private and will not be shown publicly.