Databases
StackOverflow2013
Postsdbo
POFast multiplication/division by 2 for floats and doubles (C/C++)
Body
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POFast multiplication/division by 2 for floats and doubles (C/C++)
    text
    copied!<p>In the software I'm writing, I'm doing millions of multiplication or division by 2 (or powers of 2) of my values. I would really like these values to be <code>int</code> so that I could access the bitshift operators </p> <pre><code>int a = 1; int b = a&lt;&lt;24 </code></pre> <p>However, I cannot, and I have to stick with doubles. </p> <p>My question is : <strong>as there is a standard representation of doubles (sign, exponent, mantissa), is there a way to play with the exponent to get fast multiplications/divisions by a power of 2</strong>?</p> <p>I can even assume that the number of bits is going to be fixed (the software will work on machines that will always have 64 bits long doubles)</p> <p>P.S : And yes, the algorithm mostly does these operations only. This is the bottleneck (it's already multithreaded).</p> <p>Edit : Or am I completely mistaken and clever compilers already optimize things for me?</p> <hr> <p>Temporary results (with Qt to measure time, overkill, but I don't care): </p> <pre><code>#include &lt;QtCore/QCoreApplication&gt; #include &lt;QtCore/QElapsedTimer&gt; #include &lt;QtCore/QDebug&gt; #include &lt;iostream&gt; #include &lt;math.h&gt; using namespace std; int main(int argc, char *argv[]) { QCoreApplication a(argc, argv); while(true) { QElapsedTimer timer; timer.start(); int n=100000000; volatile double d=12.4; volatile double D; for(unsigned int i=0; i&lt;n; ++i) { //D = d*32; // 200 ms //D = d*(1&lt;&lt;5); // 200 ms D = ldexp (d,5); // 6000 ms } qDebug() &lt;&lt; "The operation took" &lt;&lt; timer.elapsed() &lt;&lt; "milliseconds"; } return a.exec(); } </code></pre> <p>Runs suggest that <code>D = d*(1&lt;&lt;5);</code> and <code>D = d*32;</code> run in the same time (200 ms) whereas <code>D = ldexp (d,5);</code> is much slower (6000 ms). I <strong>know</strong> that this is a micro benchmark, and that suddenly, my RAM has exploded because Chrome has suddenly asked to compute Pi in my back every single time I run <code>ldexp()</code>, so this benchmark is worth nothing. But I'll keep it nevertheless.</p> <p>On the other had, I'm having trouble doing <code>reinterpret_cast&lt;uint64_t *&gt;</code> because there's a <code>const</code> violation (seems the <code>volatile</code> keyword interferes)</p>
 

Querying!

 
Guidance
An individual column

Larger individual text columns get their own page to allow for proper reading.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload