Dragracing dotproducts

.gitignore

*
!*/
!*.*
![Mm]akefile
*.o
*.hi

drag.hs

-- We use Int (machine integer) to match Numba's speed.
-- If you want exact big math, keep 'Integer', but it will be slower.
main :: IO ()
main = do
    let n = 2^30 :: Integer
    -- print forces the calculation
    print $ sum $ zipWith (*) [1..n] [1..n]

endgame.c

// compile with: gcc -O3 -fopenmp endgame.c -o endgame
#include <stdio.h>
#include <omp.h>

int main() {
    long long n = 1L << 30; // 2^30
    double total = 0.0;

    // The Magic Pragma
    // "parallel": Spawn threads
    // "for": Split the loop iterations among them
    // "reduction(+:total)": Give each thread a local sum, then merge them
    #pragma omp parallel for reduction(+:total)
    for (long long i = 0; i < n; i++) {
        double val = (double)i;
        total += val * val;
    }

    printf("%e\n", total);
    return 0;
}

Makefile

CC = gcc  
HC = ghc
CFLAGS = -O3 -fopenmp   
HFLAGS = -O2


.PHONY = test all
all: drag endgame test

%: %.c 
	$(CC) $(CFLAGS) $< -o $@    

%: %.hs
	$(HC) $(HFLAGS) $< -o $@

test: drag endgame slow.py
	zsh -c "time ./slow.py"
	zsh -c "time ./drag"
	zsh -c "time ./endgame"

slow.py

#!/usr/bin/env python3
xs=range(0,2**30)
ys=range(0,2**30)
sum((x*y for x,y in zip(xs, ys)))