AnsweredAssumed Answered

suggestion about 40bit float

Question asked by bookevg on Nov 22, 2011
Latest reply on Dec 17, 2011 by bookevg

I have the discussion about 40 bit float and problem solving. But one of the problem ("Rx= PASS Ry;") demand to overwork for AD. I offer to create new builtin function which do float point operation using the instruction "Fx= PASS Fy;" instead of using the instruction "Rx= PASS Ry;" and without to set RND40, because it's to decrease the perfomance to switch on/off RND40/RND32. The user define himself RND40/RND32 mode and the memory width (40 or 32 bits).