gnu@sun.uucp (John Gilmore) (06/28/85)
John Levine, ima!johnl, said: > Sounds pretty smart to me. Why waste chip real estate with locking circuits > that'll be used .0001% of the time? I expect that the WE 32100 chip > special-cases the interlocking between branches and the tests they depend > on. For that matter, the 360/91 did that 15 years ago. Of course, the /91 had to do it while running object code that ran on all the other models. It therefore needed the wasteful locking circuits. I agree that on a new machine, the frequency of "read condition codes" is so small that effectively making it a double-size opcode (NOP,READCONDITIONCODES) is a win. Of course, when they build a chip that runs two instructions simultaneously, they'll need those ubiquitous locking circuits again for object code compatability with today's chips.