Drexel dragonThe Math ForumDonate to the Math Forum



Search All of the Math Forum:

Views expressed in these public forums are not endorsed by Drexel University or The Math Forum.


Math Forum » Discussions » sci.math.* » sci.math.independent

Topic: Fast exponent and logarithm, given initial estimate
Replies: 29   Last Post: Nov 8, 2004 2:31 AM

Advanced Search

Back to Topic List Back to Topic List Jump to Tree View Jump to Tree View   Messages: [ Previous | Next ]
Glen Low

Posts: 41
Registered: 12/13/04
Re: Fast exponent and logarithm, given initial estimate
Posted: Oct 21, 2004 8:03 PM
  Click to see the message monospaced in plain text Plain Text   Click to reply to this topic Reply


> A method that improves on Horner's method in this aspect
> is Estrin's method. It rewrites the polynomial as a (mostly
> balanced) binary tree using x^2 and A*x+B terms (where A
> and B are different subexpressions).

I googled for the technique and applied it, simple and
straightforward. Instead of 6 deep multiplies using Horner's method,
it ended being 8 multiplies only 3 deep. Unfortunately profiling my
code tells me that it spends the lion's share in writing it out still,
and I also don't do enough work to fill the pipeline. So in practice,
there was no speed difference between the two methods.

I also got the compiler to unroll the loops automatically. With this
option, Horner's method comes out slightly ahead (3% ?), since the
pipelines can be filled in parallel from each unrolled iteration.

Cheers,
Glen Low, Pixelglow Software
www.pixelglow.com



Date Subject Author
10/18/04
Read Fast exponent and logarithm, given initial estimate
Glen Low
10/18/04
Read Re: Fast exponent and logarithm, given initial estimate
Jeremy Watts
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Peter Spellucci
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/18/04
Read Re: Fast exponent and logarithm, given initial estimate
bv
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
George Russell
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/20/04
Read Re: Fast exponent and logarithm, given initial estimate
George Russell
10/20/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/21/04
Read Re: Fast exponent and logarithm, given initial estimate
Christer Ericson
10/21/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/22/04
Read Re: Fast exponent and logarithm, given initial estimate
Christer Ericson
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Martin Brown
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Richard Mathar
10/19/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/20/04
Read Re: Fast exponent and logarithm, given initial estimate
Gert Van den Eynde
10/20/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/20/04
Read Re: Fast exponent and logarithm, given initial estimate
Richard Mathar
10/21/04
Read Re: Fast exponent and logarithm, given initial estimate
Gert Van den Eynde
10/21/04
Read Re: Fast exponent and logarithm, given initial estimate
bv
10/22/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/22/04
Read Re: Fast exponent and logarithm, given initial estimate
Peter Spellucci
10/22/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low
10/23/04
Read Re: Fast exponent and logarithm, given initial estimate
bv
10/24/04
Read Re: Fast exponent and logarithm, given initial estimate
Gert Van den Eynde
10/25/04
Read Re: Fast exponent and logarithm, given initial estimate
Peter Spellucci
10/20/04
Read Re: Fast exponent and logarithm, given initial estimate
Gert Van den Eynde
11/8/04
Read Re: Fast exponent and logarithm, given initial estimate
Glen Low

Point your RSS reader here for a feed of the latest messages in this topic.

[Privacy Policy] [Terms of Use]

© Drexel University 1994-2014. All Rights Reserved.
The Math Forum is a research and educational enterprise of the Drexel University School of Education.