Why Does Julia Work So Well?

Why Does Julia Work So Well?

There is an obvious reason to choose Julia:

it's faster than other scripting languages, allowing you to have the rapid development of Python/MATLAB/R while producing code that is as fast as C/Fortran

Newcomers to Julia might be a little wary of that statement.

1. Why not just make other scripting languages faster? If Julia can do it, why can't others?
2. How do you interpert Julia benchmarks to confirm this? (This is surprisingly difficult for many!)
3. That sounds like it violates the No-Free-Lunch heuristic. Is there really nothing lost?

Many people believe Julia is fast because it is Just-In-Time (JIT) compiled (i.e. every statement is run using compiled functions which are either compiled right before they are used, or cached compilations from before). This leads to questions about what Julia gives over JIT'd implementations of Python/R (and MATLAB by default uses a JIT). These JIT compilers have been optimized for far longer than Julia, so why should we be crazy and believe that somehow Julia quickly out-optimized all of them? However, that is a complete misunderstanding of Julia. What I want show, in a very visual way, is that Julia is fast because of its design decisions. The core design decision, type-stability through specialization via multiple-dispatch is what allows Julia to be very easy for a compiler to make into efficient code, but also allow the code to be very concise and "look like a scripting language". This will lead to some very clear performance gains.

But what we will see in this example is that Julia does not always act like other scripting languages. There are some "lunches lost" that we will have to understand. Understanding how this design decision effects the way you must code is crucial to producing efficient Julia code.

To see the difference, we only need to go as far as basic math.