Mon, 3 Jan 2011 02:29:59 +1100

Related articles |
---|

Integer division drizzle76@gmail.com (dz) (2010-12-27) |

Re: Integer division derek@_NOSPAM_knosof.co.uk (Derek M. Jones) (2010-12-29) |

Re: Integer division gah@ugcs.caltech.edu (glen herrmannsfeldt) (2010-12-30) |

Re: Integer division robin51@dodo.com.au (robin) (2011-01-03) |

From: | "robin" <robin51@dodo.com.au> |

Newsgroups: | comp.compilers |

Date: | Mon, 3 Jan 2011 02:29:59 +1100 |

Organization: | Compilers Central |

References: | 10-12-050 10-12-060 |

Keywords: | arithmetic |

Posted-Date: | 02 Jan 2011 20:41:33 EST |

From: "glen herrmannsfeldt" <gah@ugcs.caltech.edu>

Sent: Thursday, 30 December 2010 9:03 PM

*> dz <drizzle76@gmail.com> wrote:*

*>*

*>> I am looking for references on how to support general 32 bit*

*>> integer division (not just invariants) on hardware without dedicated*

*>> integer-division support and 64 bit floating point support.*

*>*

*> Well, there is always the shift/subtract loop generating*

*> one bit at a time. Either the restoring or non-restoring*

*> division algorithm should be easy to find. Many processors without*

*> hardware divide have instruction to make this easier, such as*

*> rotate and rotate through carry.*

Whether restoring or not, rotate and rotate thru carry are not

required. All that is needed is shift and add/subtract.

Both methods have been used from the early days when computers did not

have an integer divide instruction. Even when the computer did have a

hardware instruction, one of the above methods was used to check that

the hardware divide instruction was working correctly.

Post a followup to this message

Return to the
comp.compilers page.

Search the
comp.compilers archives again.