I suppose we should support that. However, I think this should be done in the memory system. Because of the choices I've made here, I think what we'll end up with is TWO RI type instructions for enqueueing addresses (load and store) and then a total of six R1 type instructions for (en|de)queueing words of different types, those being 32-bit word (int or float), ARGB to 4 float conversion, and conversion of two 16-bit words to floats. We'd never otherwise directly manipulate 16-bit floats.
I don't know how to teach the compilers about this, though. :) On Sat, Mar 23, 2013 at 6:56 AM, Chris Matrakidis <[email protected]>wrote: > > http://www.cs.binghamton.edu/~millerti/ins-formats2.pdf > > What about adding instructions to convert between float and half float > (IEEE binary16)? > > > Best Regards, > > Chris Matrakidis > -- Timothy Normand Miller, PhD Assistant Professor of Computer Science, Binghamton University http://www.cs.binghamton.edu/~millerti/ Open Graphics Project
_______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
