OK, this is why provocation and trolling is useful at times. People start posting cool stuff like this code above... ;) Thanks!
Andreas Koenig
Posts 9 09 Dec 2016 13:07
Wrong! Trolling always sucks. You could have just asked nicely.
Richard Gatineau
Posts 60 09 Dec 2016 16:14
Hi Gunnar,
Can we imagine to use the AMMX power to emulate a 68882 FPU through a library?
Samuel Crow
Posts 424 09 Dec 2016 17:16
Richard Gatineau wrote:
Can we imagine to use the AMMX power to emulate a 68882 FPU through a library?
Knowing Gunnar, AMMX2 will likely support floating point operations but probably not all the fancy rounding modes of a 68882. Probably more like the floating point ops of the 68040 or 68060 internally.
Presently AMMX has two data types supported for the base unit of a vector: 8 byte values per 64-bit vector register or 4 16-bit word values per vector register.