A pointless, but epic C vs Haskell battle!

ccalc.c

/*
 * Byte calculator. Sums bytes of a given file and displays to screen.
 *
 * Compile:
 *  gcc ccalc.c -o ccalc -O2
 *
 */
#include <stdlib.h>
#include <stdio.h>
#include <stdint.h>
#define BUFFER (1 << 20)

unsigned long long do_calc(FILE *f) {
    uint8_t buf[BUFFER];
    unsigned long long sum = 0;
    size_t nbytes, i;

    while ((nbytes = fread(buf, 1, BUFFER, f)) > 0)
        for (i = 0; i < nbytes; i++)
            sum += buf[i];

    return sum;
}

int main(int argc, const char* argv[]) {
    FILE *f;
    unsigned long long res;

    if (argc != 2)
        return 1;
   
    if ((f = fopen(argv[1], "rb")) == NULL) {
        perror("fopen");
        return 1;
    };

    res = do_calc(f);
    fclose(f);

    printf("%llu\n", res);
    return 0;
}

hcalc.hs

{- Compile: ghc --make -O2 hcalc.hs -}

{-# LANGUAGE BangPatterns #-}
module Main where

import System.IO
import Data.Word
import System.Environment
import qualified Data.ByteString as B

buffer = 1024

toInt :: Word8 -> Int
toInt = fromIntegral

sum_bytes :: Handle -> IO Int
sum_bytes fh = sum_bytes' fh 0

sum_bytes' :: Handle -> Int -> IO Int
sum_bytes' fh !acc = do
    res <- B.hGet fh buffer
    if B.null res
        then return acc
        else sum_bytes' fh (B.foldl' (\acc e -> (toInt e) + acc) acc res)

main = do
    args  <- getArgs
    let filepath = head args
    fh <- openBinaryFile filepath ReadMode
    val <- sum_bytes fh
    hClose fh
    print val

@joukewitteveen

You can always intercept the C code GHC generates and optimize there to have a C program that is guaranteed to be faster than the Haskell version.

Haha, well, sort of true, that's why it says "pointless" in the title. Though, the compiled Haskell C code will depend on the GHC RTS. The GHC RTS is written in C and C--. So you would have to re-do the C-- part in C too!