Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:51433
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain prohost.org designates 209.85.161.42 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <op.vqj8xup0idpuyk@damnation>
References: <op.vqj8xup0idpuyk@damnation>
Date: Tue, 8 Feb 2011 07:55:41 -0500
Message-ID: <AANLkTinss8tJOhVqFkP3+m6BPNb6_8Tm8Fq6ufgaV-ia@mail.gmail.com>
To: Gustavo Lopes <glopes@nebm.ist.utl.pt>
Cc: "internals@lists.php.net" <internals@lists.php.net>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] Change default serialize precision from 100 to 17
From: ilia@prohost.org (Ilia Alshanetsky)

+1

On Mon, Feb 7, 2011 at 8:26 PM, Gustavo Lopes <glopes@nebm.ist.utl.pt> wrot=
e:
> The default serialize precision is currently [1] set at 100. A little cod=
e
> inspection shows precision, in this case, takes the usual meaning of numb=
er
> of significant digits.
>
> Given that the implicit precision of a (normal) IEEE 754 double precision
> number is slightly less than 16 digits [2], this is a serious overkill. P=
ut
> another way, while the mantissa is composed of 52 bits plus 1 implicit bi=
t,
> 100 decimal digits can carry up to 100*log2(10) =3D~ 332 bits of informat=
ion,
> around 6 times more.
>
> Given this, I propose changing the default precision to 17 (while the
> precision is slightly less than 16, a 17th digit is necessary because the
> first decimal digit carries little information when it is low).
>
> From my tests, this makes serialization and unserialization of doubles
> around 3 times faster (counting the function calls to serialize/unseriali=
ze,
> plus a loop variable increment and stop condition check). It also makes t=
he
> serialization data.
>
> Crucially, from my tests, the condition that the variable stays the same
> before and after serialization+unserialization still holds. The test
> include, for little endian machines, verifies this.
>
> If no one objects, I'll change the default precision to 17.
>
> //run with php -d serialize_precision=3D17
> $numbers =3D array(
> =A0 =A0 =A0 =A0"0000000000000000", //0
> =A0 =A0 =A0 =A0"2d431cebe2362a3f", //.0002
> =A0 =A0 =A0 =A0"2e431cebe2362a3f", //.0002 + 10^-Accuracy[.0002]*1.01
> =A0 =A0 =A0 =A0"0000000000001000", //2^-1022. (minimum normal double)
> =A0 =A0 =A0 =A0"0100000000001000", //2^-1022. + 10^-Accuracy[2^-1022.]*1.=
01
> =A0 =A0 =A0 =A0"ffffffffffffef7f", //2^1024. (maximum normal double)
> =A0 =A0 =A0 =A0"feffffffffffef7f", //2^1024. - 10^-Accuracy[2^1024.]
> =A0 =A0 =A0 =A0"0100000000000000", //minumum subnormal double
> =A0 =A0 =A0 =A0"0200000000000000", //2nd minumum subnormal double
> =A0 =A0 =A0 =A0"fffffffffffff000", //maximum subnormal double
> =A0 =A0 =A0 =A0"fefffffffffff000", //2nd maximum subnormal double
> =A0 =A0 =A0 =A0"0000000000000f7f", //+inf
> =A0 =A0 =A0 =A0"0000000000000fff", //-inf
> );
>
> foreach ($numbers as $ns) {
> =A0 =A0 =A0 =A0$num =3D unpack("d", pack("H*", $ns)); $num =3D reset($num=
);
> =A0 =A0 =A0 =A0echo "number: ", sprintf("%.17e", $num), "... ";
> =A0 =A0 =A0 =A0$num2 =3D unserialize(serialize($num));
> =A0 =A0 =A0 =A0$repr =3D unpack("H*", pack("d", $num2)); $repr =3D reset(=
$repr);
> =A0 =A0 =A0 =A0if ($repr =3D=3D $ns)
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0echo "OK\n";
> =A0 =A0 =A0 =A0else
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0echo "mismatch\n\twas: =A0 =A0$ns\n\tbecam=
e: $repr\n";
> }
>
>
> =A0 =A0[1]: http://lxr.php.net/opengrok/xref/PHP_TRUNK/main/main.c#451
> =A0 =A0[2]: http://en.wikipedia.org/wiki/Double_precision_floating-point_=
format
> --
> Gustavo Lopes
>
> --
> PHP Internals - PHP Runtime Development Mailing List
> To unsubscribe, visit: http://www.php.net/unsub.php
>
>