Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.milw.wisc.edu!lll-winken!uunet!crdgw1!sungod.crd.ge.com!davidsen
From: davidsen@sungod.crd.ge.com (William Davidsen)
Newsgroups: comp.lang.c
Subject: Davidsen's device
Message-ID: <1430@crdgw1.crd.ge.com>
Date: 1 Aug 89 18:07:50 GMT
Sender: news@crdgw1.crd.ge.com
Reply-To: davidsen@crdos1.UUCP (bill davidsen)
Organization: General Electric Corp. R&D, Schenectady, NY
Lines: 27


  I recently discovered the following technique. While I can't swear
that it's new and unique to me, I can claim independent discovery. I
make no claims that it is suited to any problem you have, will have, or
would admit to having.

  I was calculating a function and had two algorithms, one of which was
13 times faster than the other but unreliable for input values less than
1.0, the other very accurate but slow. I had to evaluate the function
6E8 times on a slow machine.

Here's the code:

	double hacs1(), hacs2();
	double asect;

	(*(asect < 1.0 ? hacs2 : hacs1)(asect);

  Since the name of a function is typed "pointer to function
returning..." this was treated as "(*ptr)()" and generated the desired
code. It was slightly faster than the obvious:

	if (asect < 1.0) hasc2(asect);
	else hasc1(asect);
	bill davidsen		(davidsen@crdos1.crd.GE.COM)
  {uunet | philabs}!crdgw1!crdos1!davidsen
"Stupidity, like virtue, is its own reward" -me