This is an archive of the discontinued LLVM Phabricator instance.

[CompilerRT] Fix 128-bit floating point division
Needs ReviewPublic

Authored by kpdev42 on Apr 14 2022, 1:41 AM.

Details

Summary

Patch fixes number of iterations which should be 5 for given reciprocal precision, not 4.

OS Laboratory. Huawei Russian Research Institute. Saint-Petersburg

Diff Detail

Event Timeline

kpdev42 created this revision.Apr 14 2022, 1:41 AM
Herald added a project: Restricted Project. · View Herald TranscriptApr 14 2022, 1:41 AM
kpdev42 requested review of this revision.Apr 14 2022, 1:41 AM
Herald added a project: Restricted Project. · View Herald TranscriptApr 14 2022, 1:41 AM
Herald added a subscriber: Restricted Project. · View Herald Transcript

Can you give some context here? How did we end up with "4" in the first place, and how do you know "5" is sufficient for all 128-bit floats?