Re: math coprocessor

[not@available.ru]

----------
In article <3690f26f.3803810@news2.lightlink.com>,
edhel@LOSETHECAPITALWORDSapple2.com (Edhel Iaur, Esq.) wrote:


>On 4 Jan 1999 08:35:55 GMT, sagochangi@aol.com (Sagochangi) wrote:
>
>>is/was there a math coprocessor unit available for the iigs, in hardware or
>>software?
>
>Yeah, AE made one called the FastMath, and other companies made 'em as
>well.  However, if you got an accelerator (ZipGS, TransWarpGS, etc.),
>pretty much everything went faster, not just math functions that went
>through GS/OS's SANE toolbox, so the math co-processors didn't catch
>on much.

The FastMath was about the worst product ever released by AE. The math chip,
a national Semiconductor, was so stupid it was hardly acting as a valuable
addition to the 65816.

The only capable math coprocessor were the Floating Point Engine and its
substitute Number Cruncher by, I think, Andreas Schmid. The Number Cruncher
had an edge over the FPE because it actually worked.

It is not correct that math cards were an useless add-on. Orca can compile
source code specifically for FPE/NC enhanced systems, the SANE patches
routed some math calls through the coprocessor, and lastly there are
programs that take advantage  of the FPE. Ok, I can only think of one at
this time. Think tenfold speed increase with math routines.

- henrik