3 ms·
Indeed, the Fibonacci sequence is in the original benchmark and Fib(15) is not benchmarking big integer performance. It should have the same characteristics as
by bonzini 2y ago
Indeed, the Fibonacci sequence is in the original benchmark and Fib(15) is not benchmarking big integer performance. It should have the same characteristics as the always-zero function.
- spinningslate 2y agoI think gp's point is that bif[n_] == 0 for all n_. A 'smart' optimser would recognise this and so the time to compute would be (a) constant irrespective of the value of n_, and (b) instantaneous because the function call can be re-rewritten as the constant value 0.
- bonzini 2y agoUnderstood, and my point is that user Someone suggested "bif" as a way to avoid testing bigint performance; but fib[15] does not test bigint performance. So using bif is unnecessary and potentially harmful if it's optimized to zero.