Re: gfortran speed benchmarks

[Richard E Maine <Richard dot Maine at nasa dot gov>]

On Nov 8, 2005, at 10:56 AM, Thomas Koenig wrote:

> Sorry to appear dense, but what is TESTFPU.f90?

You are almost better off not knowing. :-( It is a "benchmark" written 
by a comp.lang.fortran denizen notorious for, well... many things, 
pretty much none of them positive. These include apparently deliberate 
refusal to actually read material written by folk who are experts in 
well... pretty much anything... with the result that he ignores all the 
lessons to be learned from prior mistakes of others. Combine this with 
refusal to learn from his own mistakes, and you'll begin to get part of 
the picture. Specific instances of the above-described patterns are 
regularly observed in him in the fields of benchmarking, numerical 
analysis, and language features.

Oh, and he intentionally distorts and misrepresents even simple factual 
material whenever it suits his point and excessively indulges in 
continual flame wars.

I killfiled him some time ago.

Of course, all the above remarks being ad hominems, this doesn't mean 
that one shouldn't look at testfpu to see if it can be of use, as long 
as you don't treat it as the one and only, be-all measure of all 
performance issues (which is the kind of attitude that its author 
expresses).

--
Richard Maine                |  Good judgment comes from experience;
Richard.Maine@nasa.gov       |  experience comes from bad judgment.
                            |        -- Mark Twain