Re: [lmi] use boost::math:expm1() and log1p()

[Greg Chicares]

On 2009-04-02 23:12Z, Vaclav Slavik wrote:
> 
> as promised, here's a patch to use expm1() and log1p() instead of C99
> expm1l() and log1pl() (removal of expm1.{c,h} is not included in the
> patch for simplicity). 

Before we consider switching to a different boost version or using
boost's build system, let's measure the speed and accuracy of these
functions. Since you've already built this, could I ask you to run
the math_functors_test$(EXEEXT) unit test and post the results?

If they're too inaccurate, then this is a dead end.

If they're accurate but too slow, then this is useful for toolchains
that lack these C99 transcendental functions, but we would probably
continue using 'expm1.c' in production. I wouldn't expect boost's
implementation to be as fast as the F2XM1 hardware instruction, and
I believe we perform these calculations often enough to affect run-
time performance.