Ron Parker wrote:
> 
> On Mon, 02 Nov 1998 11:46:46 +0100, Markus Becker
>         <bec### [at] zessuni-siegende> wrote:
> >That's not quite right. It's a SIMD architecture (Single Instruction,
> >Multiple, Data) that is similar to the MMX by Intel. It's right
> >that you can handle several floats at once, but with a limited
> >precision.
> 
> So what about what I said is inaccurate?  ISTR that at least the

Not what you said is inaccurate. It's just that I thought the
results of the SIMD calculations would be more inaccurate.

> 2-operand form allows you to work with two 32-bit floats, which
> are equivalent in precision to the floats that are used with, for
> example, colours.  And overriding the colour functions, such as
> adding, scaling, and inverse-scaling colours, might not be such
> a difficult task.  However, again, I don't think one would gain
> much for one's labour.

Right.

Markus

Post a reply to this message