gistfile1.php

<?php
ini_set('memory_limit','4G');

$arr_args = getopt('', Array('file:', 'port:', 'step:', 'runs', 'name:'));

if(!isset($arr_args['file'])) {
    die("Please pass an input filename of sets!\n");
}
if(!isset($arr_args['port'])) {
    die("Pass a port where redis-server is running!\n");
}

$str_file = $arr_args['file'];
$i_port   = (int)$arr_args['port'];
$i_step   = isset($arr_args['step']) ? $arr_args['step'] : 10000;
$i_runs   = isset($arr_args['runs']) ? $arr_args['runs'] : 5;
$str_name = isset($arr_args['name']) ? $arr_args['name'] : NULL;

$obj_r = new Redis();

if(!$obj_r->connect('127.0.0.1', $i_port)) {
    echo "Can't communicate with Redis on port $i_port!\n";
    exit(1);
}
if(!isset($arr_args['name'])) {
    echo "Give this run a name!\n";
    exit(1);
}
if(!file_exists($str_file)) {
    echo "Sets file '$str_file' not found!\n";
    exit(1);
}

// Headers
fputcsv(STDOUT,Array('sets','incard','outcard',$str_name));

// The sets in question
$arr_sets = explode("\n", file_get_contents($str_file));

// Start with stepping, increase the set count each time
for($i=$i_step; $i<count($arr_sets);$i+=$i_step) {
    // Grab a slice of N sets
    $arr_run_sets = array_slice($arr_sets, 0, $i);

    // Determine input cardinality
    $obj_r->pipeline();
    foreach($arr_run_sets as $str_set) {
        $obj_r->zcard($str_set);
    }
    $i_in_card = array_sum($obj_r->exec());

    // Total timing for this run
    $i_tot = 0;

    // Run "runs" time
    for($j=0;$j<$i_runs;$j++) {
        $obj_r->del('dest_set');

        $st = microtime(true);
        $i_out_card = $obj_r->zunionstore('dest_set', $arr_run_sets);
        $et = microtime(true);

        $i_tot += $et-$st;
    } 

    // Average the runs
    $f_timing = round($i_tot/$i_runs, 4);

    fputcsv(STDOUT, Array($i, $i_in_card, $i_out_card, $f_timing));
}