An Inline::C implementation runs about three times faster than the string interpolation version for integer comparison, and about 2.5 times faster for string comparisons.

use v5.16;
use strict;
use warnings;
use Inline 'C';

sub arrays_same_pp {
    "@{$_[0]}" eq "@{$_[1]}";
}

our @x = map { int(rand() * 1_000_000) } 1..10;
our @y = @x;

use Benchmark 'cmpthese';

cmpthese -1, {
    PP  => '::arrays_same_pp(\@::x, \@::y)',
    XS  => '::arrays_same_i(\@::x, \@::y)',
};

__END__
__C__

int arrays_same_i (SV* x, SV* y) {
    AV* arrx = (AV*)SvRV(x);
    AV* arry = (AV*)SvRV(y);
    
    if (arrx == arry) {
        return 2;
    }
    
    int len = av_len(arrx);
    if (len != av_len(arry)) {
        return 0;
    }
    
    int i;
    for (i=0; i<=len; i++) {
        SV** elemx = av_fetch(arrx, i, 0);
        SV** elemy = av_fetch(arry, i, 0);
        int ix = SvIV(*elemx);
        int iy = SvIV(*elemy);
        if (ix != iy) {
            return 0;
        }
    }
    
    return 1;
}

int arrays_same_s (SV* x, SV* y) {
    AV* arrx = (AV*)SvRV(x);
    AV* arry = (AV*)SvRV(y);
    
    if (arrx == arry) {
        return 2;
    }
    
    int len = av_len(arrx);
    if (len != av_len(arry)) {
        return 0;
    }
    
    int i;
    for (i=0; i<=len; i++) {
        SV** elemx = av_fetch(arrx, i, 0);
        SV** elemy = av_fetch(arry, i, 0);
        STRLEN dummy;
        char* strx = SvPV(*elemx, dummy);
        char* stry = SvPV(*elemy, dummy);
        if (strcmp(strx, stry) != 0) {
            return 0;
        }
    }
    
    return 1;
}
[download]

I might clean this up and release to CPAN if there's nothing similar already there.

Can you say a few words as to how this compiles?

The Inline::C module lets you include C code in your __END__ section, and compiles it into XS on-the-fly, so you should just be able to run that script like any other Perl script.

toby döt ink