Re: [Flightgear-devel] GCC 3.0 Floating Point

[C. H. <cho...@ho...>]

Even though values are stored as floats, it is entirely possible that the
hardware is playing conversion games. Between compiler tricks and
hardware behavior we might see some counter intuitive results. I'd like
to see some profiling before believing that using floats throughout saves
any time.

Given that so much of OpenGL is written with float in mind I would bet
that only the JBSim code would be the only place that choosing between
doubles and floats might make a difference.

Regards,

Charlie H.

David Megginson wrote:

> C. Hotchkiss writes:
>
>  > Some of compilers do all floating point in double and the compiler
>  > converts back and forth between float and double to satisfy the
>  > programmer.
>
> Interesting.  We need to think a little more about what's going on.
>
>  > Given that floating point calcs are done on any large processor by
>  > dedicated hardware, I would expect no difference in speed of
>  > execution.
>
> Not even the constant conversion back-and-forth between the four-byte
> value stored in memory and the eight-byte value used by hardware?
>
> All the best,
>
> David
>
> --
> David Megginson
> da...@me...
>
> _______________________________________________
> Flightgear-devel mailing list
> Fli...@li...
> http://lists.sourceforge.net/lists/listinfo/flightgear-devel

--
"There are two major products that come out
of Berkeley: LSD and UNIX. We don't believe
this to be a coincidence."
                          -     J. Anderson