[PATCH] D66050: Improve division estimation of floating points.

[Chaofan Qiu via Phabricator via llvm-commits]

qiucf added a comment.

In D66050#1624955 <https://reviews.llvm.org/D66050#1624955>, @nemanjai wrote:

> 1. The patch needs full context to be reviewable
> 2. The formatting needs to be fixed up (some lines too long, etc.)
> 3. We need testing (I imagine this makes a bunch of LIT tests fail which need to be updated)


I'm updating the patch :) Broken LIT tests aren't many.

> 3. It might not be a bad idea to add a test to `test-suite` that will do some fast division vs. non-fast division to make sure the accuracy isn't too bad

I'll add a portable case can be compiled and run at all platforms. That can be merged into our `test-suite`.

> 4. We should do some thorough analysis of the accuracy of the algorithm vs. the HW implementation on a wide range of values on a couple of targets (as was suggested above). I know this was done on PPC, so please share those results and try to replicate on another easily available target (such as X86).

Yes. But actually only a few platforms have support for such estimation. Clang doesn't do it on X86 but GCC does. I'll add the author to reviewer list.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66050/new/

https://reviews.llvm.org/D66050