julia> using BenchmarkTools julia> function sum1(num) total = 0 for i in num total += i end return total end sum1 (generic function with 1 method) julia> a = collect(1:1000); julia> @btime b = sum(a) 99.192 ns (1 allocation: 16 bytes) 5005…

You seem to be interpolating a at some points and not at others, this could probably affect timing. When the input is a range type I guess the standard sum dispatches to a method using some summation formula, e.g. S(n)=n*(n+1)/2, or similar. Not sure why sum1 also improves that much, 2ns seems like…

Why the function sum1 is faster than builtin sum

General Usage Performance

GunnarFarneback September 20, 2023, 10:54am 12

No, I don’t.

Topic		Replies	Views
A fast sum. Any downsides? Performance sum	18	970	December 16, 2024
Compare julia sum to a cpp implementation - julia is extremely slow?! Performance question	35	1795	October 7, 2019
How to improve performance of sum() Performance question , fast-math , tullio , loopvectorization , sum	19	4861	July 19, 2021
How to speed up this simple code? Multithreading, simd, inbounds Performance	39	7042	January 29, 2019
Summing matrix elements is >1000X slower than summing vector elements General Usage performance	8	1330	April 17, 2017

Why the function sum1 is faster than builtin sum

Related topics