[Oberon] FP

Jan Verhoeven jan at verhoeven272.nl
Tue Apr 15 10:56:18 CEST 2014


Holger Goetz wrote:
> Hi,
>
> my 2cts:


Appreciated!


> This sounds like best thinking for the emulator, but it makes a
> difference for the FPGA, where it is like it's with most embedded
> systems: "size matters". That was done to keep memory requirements low
> (well, even lower).


We are talking about an FPGA board with 1.000.000 bytes of RAM. Not a 
PIC32 SBC with 256 bytewide registers. Or a ZX81 with a whopping 1024 
bytes of 250 ns SRAM.
And the extra cost is just one 32 bit word....

And, let's face it, who ever needs floating point numbers when you can 
use scaled integers?

NW says: make it as simple as possible. An FPnum format consisting of 
1/4 word exponent and 3/4 word mantissa is *not* simple. We should act 
on his words and improve this, now that things are still in the infant 
phase.

Have you seen the FP routines in the emulator? It's complex math. And 
it's unnecessary.


-- 
Groetjes

Jan Verhoeven




More information about the Oberon mailing list