There is always too much to write. And too much to read. And too much to make. And too much to feel.
Everything demands to be heard now. And felt now. And thought about now. And it feels as if everything that I can't fit into this hopelessly infinitesimal slice of time called now screams at me of injustice, of being ignored, of being stolen from their deserved chance at skewing ever so slightly who I am in the moment after now.
Benchmarking fib(n) in Oak
I'm in a benchmarking mood today. So I implemented the (naive, not-memoized) Fibonacci function in both Oak and Python (the language it's probably most comparable to in performance) and ran them through some measurements.
Here's the Oak version:
fn fib(n) if n {
0, 1 -> 1
_ -> fib(n - 1) + fib(n - 2)
}
print(string(fib(34)) + '\n')And here's the Python version, a direct port of the Oak one:
def fib(n):
if n == 0 or n == 1:
return 1
else:
return fib(n - 1) + fib(n - 2)
print(fib(34))I also made a Node.js version compiled from the Oak implementation, with oak build --entry fib.oak -o out.js --web.
With the magic of Hyperfine, we see ...
Benchmark 1: node out.js
Time (mean ± σ): 259.4 ms ± 1.9 ms [User: 252.6 ms, System: 11.7 ms]
Benchmark 2: python3 fib.py
Time (mean ± σ): 2.441 s ± 0.042 s [User: 2.421 s, System: 0.011 s]
Benchmark 3: oak fib.oak
Time (mean ± σ): 13.536 s ± 0.047 s [User: 14.767 s, System: 0.729 s]
Summary
'node out.js' ran
9.41 ± 0.18 times faster than 'python3 fib.py'
52.19 ± 0.43 times faster than 'oak fib.oak'The gap between Python and native Oak isn't too surprising -- Oak is generally 4-5 times slower than Python on numerical code, so this looks right. But I was quite surprised to see just how much faster Node.js runs. V8 is very, very good at optimizing simple number-crunching code.