Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.arch    |    Apparently more than just beeps & boops    |    131,241 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 130,296 of 131,241    |
|    Terje Mathisen to MitchAlsup    |
|    Re: Multi-precision addition and archite    |
|    15 Nov 25 15:36:22    |
      From: terje.mathisen@tmsw.no              MitchAlsup wrote:       > CARRY in My 66000 essentially provides an accumulator for a few instructions       > that supply more operands to and receives another result from a calculation.       > Most multiprecision calculation sequences are perfectly happy with another       > register used as an accumulator.              I think I've said so before, but it bears repeating:              I _really_ love CARRY!              It provides a lot of "missing link" operations, while adding zero extra       bits to all the instructions that don't need it.              That said, if I had infinite resources (in this case infinity == 4       sources), I would like to have an unsigned integer MulAddAdd like this:               (hi, lo) = a*b+c+d              simply because this is the largest possible building block that cannot       overflow, the result range covers the full 128 bit space.               From what you've taught us about multipliers, adding one (or in this       case two) extra inputs to the adder that aggregates all the partial       multiplication products will be close to free in time, but the routing       of the extra set of inputs might require an extra cycle?              Terje       --       - |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca