> Well,
> 
> I did some tests using Povray on various platforms, for my own interest
> (yes I have an interesting life, honest). I was ray tracing the 'Chess'
> textured chessboard scene.

Hmm a good test I think since these things are being done in the real world
at the moment with VRML etc....
 
> 1) The 700+FPA11 at 34MHz does this three times faster than a 250MHz
>    StrongARM.

A ha a fault in A.R.M.'s floating point architecture (is the FPA11 at 34MHz
too)
 
> 2) The 200MHz Pentium pro box in the office manages to ray trace this scene at
>    least thirty (30) times faster than the StrongARM.

That's not something A.R.M. will want to hear
 
> 3) The Alpha 2100 here eclipses all of these by a high margin.

DEC seem to have the right idea...

> I've not tried this on any other systems, we have some SGs, VAXen, etc. but I
> think the above says it all about the ARM FP performance.
> 
> 	Yours aye,
> 
> 	Neil
> 

Well then I think I know where I would put the money if I had it! But
remember in the REAL WORLD there are those who do have the money (By the way
I think it's A.R.M.'s fear of microcode that is preventing the FPU being put
into the chip, and the extra power that would be  used by this unit, since
it could well be more complex than the rest of the chip...?)

As for ART saying NO to FPU(A)'s are they nuts or is Acorn going to make the
same sort of mistakes they are well known to do (like put their head's in
the sand and hope it goes away....)

RATTY!