32x32=64

gah@ugcs.caltech.edu (glen herrmannsfeldt)
20 Dec 2001 00:43:17 -0500

          From comp.compilers

Related articles
32x32=64 gah@ugcs.caltech.edu (2001-12-20)
Re: 32x32=64 RLWatkins@CompuServe.Com (R. L. Watkins) (2001-12-22)
| List of all articles for this month |

From: gah@ugcs.caltech.edu (glen herrmannsfeldt)
Newsgroups: comp.compilers
Date: 20 Dec 2001 00:43:17 -0500
Organization: California Institute of Technology, Pasadena
Summary: 64 bit multiplication
Keywords: arithmetic, comment
Posted-Date: 20 Dec 2001 00:43:17 EST

There is a discussion in comp.arch that wandered from its initial
discussion about typesetting to C code.


The point was made that mutliplying two 32 bit numbers, with one or
both cast to a 64 bit type, could be done using a 32 x 32 multiply
with 64 bit product instruction that many machines have. (Using the
"as if" rule)


The question, then, is do any compilers perform this optimization?


-- glen
[They sure did it for 16x16 -> 32 -John]


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.