Re: [hatari-devel] FPU update

[ Thread Index | Date Index | More Archives ]

Andreas Grabher schrieb:

> I didn't read through the standard. So this formula: (-1) ^ s x 2 ^ (1 - b) x 0.f for denormals might be only true for single and double precision.

Indeed. The formula above is mentioned in the standard only for single
and double format numbers.

> I agree that this implementation would make sense. But x87 obviously has the other implementation, which is also supported by SoftFloat, which claims to be IEEE compliant. That confuses me.

Again, the standard says very little about extended formats. There's
just a table of minimum requirements, which doesn't specify the bias or
the encoding of de-normalized numbers, though. So I think both the x87
implementation and the MC6888x implementation can claim that they follow
IEEE 754-1985, even though their extended formats differ.

>> Wouldn't the easiest way be to check this on actual hardware?

> Absolutely. But i don't have any 68k hardware. So if anyone wants to test this, that would be very helpful.

I don't have a 68k system with a coprocessor and could thus only test
against Motorola's software implementation of the floating point
instructions. But surely someone on the Hatari list will have an FPU?
Otherwise, if you provide a compiled test program, I can ask in an Atari
forum. There are several people there with MC6888x.

Christian Zietz  -  CHZ-Soft  -  czietz@xxxxxxx
PGP/GnuPG-Key-ID: 0x52CB97F66DA025CA / 0x6DA025CA

Mail converted by MHonArc 2.6.19+