Initial Performance Comparison

[KieranP]

I realise comparing performance of Boa to Node, Deno, or Bun is unfair at such an early stage of the project. No doubt focus has been on feature completeness before performance. But it might also be useful (?) to have/track a test that is outside of specific benchmarks, something more real to life?

I have a repo consisting of multiple near-identical implementations of Conways Game of Life (https://github.com/KieranP/Game-Of-Life-Implementations) which I find serves as a good comparison of a languages current speed.

Here are the stats of the main 3 JS runtimes (measured in milliseconds):

Iterations	Combined Lowest	Combined Avg	Tick Lowest	Tick Avg	Render Lowest	Render Avg	Interpreter / Runtime
83775	0.311	0.356	0.100	0.113	0.211	0.243	Bun 1.2.22
75948	0.355	0.392	0.097	0.110	0.258	0.282	Node 24.8.0
74808	0.365	0.399	0.100	0.114	0.265	0.285	Deno 2.4.5

By comparison, here is where Boa currently sites:

Iterations	Combined Lowest	Combined Avg	Tick Lowest	Tick Avg	Render Lowest	Render Avg	Interpreter / Runtime
???	19.000	20.819	14.000	15.130	5.000	5.689	Boa 0.21.0

(unable to get total iterations because of #4493)

Compared to Node averages, that puts Boa ~138x slower on tick operations (loop over map values, calculate next state, set current state to next state), and ~20x slower on render operations (loop over x*y, fetch cell from map, render character, concat into large string). Combined, Boa is currently 53x slower overall.

If it would be helpful, as each new major release of Boa comes out, I'll report back on what change any performance improvements have made, if any.