This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: What is acceptable for -ffast-math? A numerical viewpoint wrote:
> <<Stop telling us what we're allowed to want.
> We are just trying to understand what you want, it was you who said
> that you do not want optimizations that change program results, not me!

No it wasn't. You seem to be confusing me with somebody else.

> Now you change it to "change program results enough to matter". And
> that of course is the point, we have to agree on what is enough to matter.

No we don't. Only the people who actually use -fextreme-maths-optimise
(or whatever it gets called) need to agree. People who don't like what
it does don't have to use it.

Nobody's suggested that it should be the default. In fact, there seems
to be unanimous agreement that it shouldn't be enabled by any -O level.
Personally, I think -mieee should be the default (at least on machines
like IA32 where native FP is close enough to IEEE that it doesn't have
to be emulated). (But then, I also think -ansi -pedantic should be the
default, and for that matter that -ansi and -mieee should be renamed
-iso and -miec559, so what do I know?)

This is why I can't understand the position you and Gabriel are taking.
You're acting as though _you_ were going to be forced to use this
option. Since you obviously want last-decimal-place precision,
presumably you're going to use -mieee instead. So why do you care what
the extreme maths option does?

Speaking as someone with an interest in games programming, I would
_love_ to have an option that tells GCC "apply any transform that would
make sense for true real numbers, and to hell with the last few decimal

Ross Smith <> The Internet Group, Auckland, New Zealand
"Unix has always lurked provocatively in the background of the operating
system wars, like the Russian Army."                  -- Neal Stephenson

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]