List append and sum benchmark

[certik]

This benchmark is from https://lemire.me/blog/2012/06/20/do-not-waste-time-with-stl-vectors/

Python:

from ltypes import i32, i64

def test_list(n: i32) -> i64:
    a: list[i32] = [0]
    i: i32
    for i in range(n):
        a.append(i)
    s: i64 = 0
    for i in range(n):
        s += a[i]
    return s

print(test_list(100000002))

and C++:

#include <vector>
#include <iostream>

int64_t test_list(int32_t n) {
    std::vector<int32_t> a = {0};
    for (int32_t i = 0; i < n; i++) {
        a.push_back(i);
    }
    int64_t s=0;
    for (int32_t i = 0; i < n; i++) {
        s += a[i];
    }
    return s;
}

int main() {
    std::cout << test_list(100000002) << std::endl;
    return 0;
}

Results on Apple M1 Max:

$ clang++ -std=c++17 -O3 a.cpp
$ time ./a.out
5000000050000000
./a.out  0.06s user 0.07s system 91% cpu 0.145 total
$ lpython --fast a.py
$ time ./a.out
5000000050000000
./a.out  0.08s user 0.04s system 97% cpu 0.126 total
$ time PYTHONPATH=src/runtime/ltypes python a.py
5000000050000000
PYTHONPATH=src/runtime/ltypes python a.py  6.63s user 0.47s system 99% cpu 7.107 total

[Thirumalai-Shaktivel]

Results on Intel® Core™ i5-8250U CPU @ 1.60GHz × 8 (OS: Ubuntu 22.04 LTS)

$ clang++ -std=c++17 -O3 -march=native a.cpp
$ time ./a.out
5000000050000000
./a.out  0.28s user 0.25s system 99% cpu 0.532 total
$ lpython --fast a.py
$ time ./a.out
5000000050000000
./a.out  0.14s user 0.11s system 99% cpu 0.246 total
$ time python a.py
5000000050000000
python a.py  13.77s user 1.40s system 99% cpu 15.178 total

[Smit-create]

Apple Air Mac M1 (2020)

% lpython expr.py
% time ./a.out 
5000000050000000
./a.out  0.50s user 0.14s system 41% cpu 1.557 total

% lpython --fast expr.py
% time ./a.out          
5000000050000000
./a.out  0.13s user 0.13s system 40% cpu 0.617 total

% clang++ -std=c++17 -O3 test_cpp.cpp -o b.out
% time ./b.out
5000000050000000
./b.out  0.09s user 0.10s system 20% cpu 0.922 total

% time PYTHONPATH=$PWD/src/runtime/ltypes python expr.py
5000000050000000
PYTHONPATH=$PWD/src/runtime/ltypes python expr.py  12.13s user 2.63s system 90% cpu 16.368 total

[czgdp1807]

Comparison with codon,

Compiler	Time [s]	Relative
LPython (lpython --fast)	0.156	1.0
codon 0.15.5 (codon build -release -exe and time ./executable)	0.654	4.19

Codes

from ltypes import i32, i64

def test_list(n: i32) -> i64:
    a: list[i32] = [i32(0)]
    i: i32
    for i in range(n):
        a.append(i)
    s: i64 = i64(0)
    for i in range(n):
        s += i64(a[i])
    return s

print(test_list(100000002))

def test_list(n):
    a = [0]
    for i in range(n):
        a.append(i)
    s = 0
    for i in range(n):
        s += a[i]
    return s

print(test_list(100000002))

[certik]

Here is the updated benchmark:

from lpython import i32, i64

def test_list(n: i32) -> i64:
    a: list[i32] = [0]
    i: i32
    for i in range(n):
        a.append(i)
    s: i64 = i64(0)
    for i in range(n):
        s += i64(a[i])
    return s

print(test_list(100000002))

It still runs at the same speed for me.