bcmul

I made this to multiply an unlimited size of integers together (meaning no decimals)..
This could be useful for those without the BCMath extension.

<?php

function Mul($Num1='0',$Num2='0') {
  // check if they're both plain numbers
  if(!preg_match("/^\d+$/",$Num1)||!preg_match("/^\d+$/",$Num2)) return(0);

  // remove zeroes from beginning of numbers
  for($i=0;$i<strlen($Num1);$i++) if(@$Num1{$i}!='0') {$Num1=substr($Num1,$i);break;}
  for($i=0;$i<strlen($Num2);$i++) if(@$Num2{$i}!='0') {$Num2=substr($Num2,$i);break;}

  // get both number lengths
  $Len1=strlen($Num1);
  $Len2=strlen($Num2);

  // $Rema is for storing the calculated numbers and $Rema2 is for carrying the remainders
  $Rema=$Rema2=array();

  // we start by making a $Len1 by $Len2 table (array)
  for($y=$i=0;$y<$Len1;$y++)
    for($x=0;$x<$Len2;$x++)
      // we use the classic lattice method for calculating the multiplication..
      // this will multiply each number in $Num1 with each number in $Num2 and store it accordingly
      @$Rema[$i++%$Len2].=sprintf('%02d',(int)$Num1{$y}*(int)$Num2{$x});

  // cycle through each stored number
  for($y=0;$y<$Len2;$y++)
    for($x=0;$x<$Len1*2;$x++)
      // add up the numbers in the diagonal fashion the lattice method uses
      @$Rema2[Floor(($x-1)/2)+1+$y]+=(int)$Rema[$y]{$x};

  // reverse the results around
  $Rema2=array_reverse($Rema2);

  // cycle through all the results again
  for($i=0;$i<count($Rema2);$i++) {
    // reverse this item, split, keep the first digit, spread the other digits down the array
    $Rema3=str_split(strrev($Rema2[$i]));
    for($o=0;$o<count($Rema3);$o++)
      if($o==0) @$Rema2[$i+$o]=$Rema3[$o];
      else @$Rema2[$i+$o]+=$Rema3[$o];
  }
  // implode $Rema2 so it's a string and reverse it, this is the result!
  $Rema2=strrev(implode($Rema2));

  // just to make sure, we delete the zeros from the beginning of the result and return
  while(strlen($Rema2)>1&&$Rema2{0}=='0') $Rema2=substr($Rema2,1);

  return($Rema2);
}

$A='5650175242508133742';
$B='2361030539975818701734615584174625';

printf("  Mul(%s,%s); // %s\r\n",$A,$B,  Mul($A,$B));
printf("BCMul(%s,%s); // %s\r\n",$A,$B,BCMul($A,$B)); // build-in function

/*
  This will print something similar to this..
    Mul(5650175242508133742,2361030539975818701734615584174625);
  BCMul(5650175242508133742,2361030539975818701734615584174625);

  both of which should be followed by the answer:
  13340236303776981390475700774516825287352418182696750
*/

?>

It was a fun experience making.. even though this took me longer than the BCAdd alternative I did..
Memory allocation might be an issue for rediculously larger numbers though.. if someone wants to benchmark the performance of my function; feel free.
Enjoy,
Nitrogen.

$float = 0.31234144143341;
$float1 = 0.00000000000000000000000000000005;
echo $float, "\n";
//0.31234144143341
echo $float1, "\n";
//5.0E-32
echo $float*$float1, "\n";
//1.5617072071671E-32

<?php
/*bcmul read float as string*/
echo bcmul($float, $float1, 32),"\n";
//0
echo bcmul($float, sprint('%.32f',$float1), 32);
//0.000000000000000000000000000000015617072071671;
?>

Well, I have a little problem implementing Blake Hash in my server because it is not a x64 server machine. I made a little function that use the powerfull of BC library to do the bitwise operation Shift.

<?php
echo 'Left Shift test<br />';
bprint('1', decbin(1));
bprint('1 << 32 (Fail)', decbin(1 << 32)); //Fail, operation not succesfull in 32-bit machine
bprint('shiftleft(1, 32) (Success)', dec2bin(shiftleft('1', '32'))); //decbin fails, so we use personalized function, success

echo '<br />';
echo 'Right Shift test<br />';
bprint('9223372036854775808', dec2bin('9223372036854775808'));
bprint('9223372036854775808 >> 63 (Fail)', decbin(9223372036854775808 >> 63));
bprint('rightshift(9223372036854775808, 63) (Success)', decbin(rightshift('9223372036854775808', '63')));

function shiftleft($num, $bits) {
    return bcmul($num, bcpow('2', $bits));
}

function rightshift($num, $bits) {
    return bcdiv($num, bcpow('2', $bits));
}

function bprint($title, $content) {
    echo $title . '<br />' . str_pad($content, 64, '0', STR_PAD_LEFT) . '<br />' . PHP_EOL;
}

//http://www.php.net/manual/en/function.decbin.php#99533
function dec2bin($dec) {
    // Better function for dec to bin. Support much bigger values, but doesn’t support signs
    for ($b = '', $r = $dec; $r >1;) {
        $n = floor($r / 2);
        $b = ($r - $n * 2) . $b;
        $r = $n; // $r%2 is inaccurate when using bigger values (like 11.435.168.214)!
    }
    return ($r % 2) . $b;
}
?>

When using printf to print the results of bcmath operations, use string format, i.e. '%s', not numeric formats such as '%d' or '%f'.  For example, the output of factorial (23) will be incorrect if using %d or %f:

Result using %f:
factorial (22) = 1124000727777607680000 (correct)
factorial (23) = 25852016738884978212864 (incorrect)

Result using %s:
factorial (22) = 1124000727777607680000
factorial (23) = 25852016738884976640000

Using echo, this is not a problem - PHP will output the bcmath string type correctly.