cvs.gedasymbols.org/archives/browse.cgi   search  
Mail Archives: djgpp-workers/1998/10/09/11:51:59

From: Kbwms AT aol DOT com
Message-ID: <c97dc42a.361e30e0@aol.com>
Date: Fri, 9 Oct 1998 11:50:56 EDT
To: rudd AT cyberoptics DOT com
Cc: djgpp-workers AT delorie DOT com
Mime-Version: 1.0
Subject: Re: libc math function upgrade work
X-Mailer: AOL 3.0 16-bit for Windows sub 38
Reply-To: djgpp-workers AT delorie DOT com

Subj:    Re: libc math function upgrade work
To:     rudd AT cyberoptics DOT com (Eric Rudd)

Dear Eric Rudd,

On 10-09-98 at 10:07:48 EST you wrote:
> CC:   djgpp-workers AT delorie DOT com
>
> Kbwms AT aol DOT com wrote:
>
> > Dear Eric Rudd,
> >
> > Appended are two reports in Cody-Waite table format that recap
> > the results of tests on 14 of your functions.  These are the
> > elementary math functions.
> >
> > In general, your functions perform as well or better on these
> > tests than the equivalent functions in the new libm.a library.
> > The exceptions are in the sine and cosine functions which do
> > not perform adequately on large arguments.
>
> Thank you for performing the tests on my library.

You are welcome.

> Where do we go from here?

As we saw from the initial trial, some of the source files cannot be
assembled under version 2.7 of the assembler.  Perhaps that ought to
be corrected.

You should consider replacing pow(x,y) with a version that does not
perform exponentiation of the error in argument y.

My advice is to ask Eli Zaretskii what the next steps should be.  For
sure, djgpp-workers should have a crack at reviewing them.

Appended is a table of execution times for the tests involves.  The
times are meant for comparison purposes only because America Online
- a well-known CPU hog - was running simultaneously.


K.B. Williams
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
                                        Elapsed Time in Seconds
Name of Test                             Rudd           libm.a
------------                            ------          ------
Test of acosh(x) vs. xacosh(x):          48.022          48.352
Test of asin(x) vs. xasin(x)
 & acos(x) vs. xacos(x):                118.407         118.516
Test of asinh(x) vs. xasinh(x):          38.132          38.242
Test of atan(x) vs. xatan(x)
 & atan2(x,y) vs. xatan2(x,y):           75.275          75.604
Test of atanh(x) vs. xatanh(x):          20.165          20.110
Test of exp(x) vs. xexp(x):              12.967          12.857
Test of log1p(x) vs xlog(1+x):            6.868           6.923
Test of log(x) vs xlog(x)
 & log10(x) vs xlog10(x)                 22.088          39.121
Test of pow(x,p) vs. xpow(x,p):          48.516          97.253
Test of sin(x) vs. xsin(x)
 & cos(x) vs. xcos(x):                   40.989          82.912
Test of sinh(x) vs. xsinh(x)
 & cosh(x) vs. xcosh(x):                 38.956          77.857
Test of sqrt(x) vs. xsqrt(x):            26.538          52.967
Test of tan(x) vs. xtan(x)
 & cot(x) vs. xcot(x):                   58.516         117.143
Test of tanh(x) vs. xtanh(x):            20.165          40.440


- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019