Page MenuHomePhabricator

[CodeGen][X86] Don't scalarize vector saturating add/sub

Authored by nikic on Jan 28 2019, 11:48 AM.



This matches what the cost model already does.

This relies on vector SADDO/SSUBO though. I think this works fine because those are expanded on all archs (for vectors), but is possibly not entirely reliable.

Diff Detail


Event Timeline

nikic created this revision.Jan 28 2019, 11:48 AM

This might be a little premature as we haven't got PR40442 yet.

5325 ↗(On Diff #183928)

You can keep this but just change to isInteger()?

nikic marked an inline comment as done.Feb 7 2019, 1:30 PM

The relevant part of PR40442 has been resolved now.

5325 ↗(On Diff #183928)

I removed those because I thought these asserts are now redundant with the specification in, or is that not enforced?

Herald added a project: Restricted Project. · View Herald TranscriptFeb 7 2019, 1:30 PM
nikic updated this revision to Diff 186155.Feb 10 2019, 9:25 AM

Restore assertions (and move to towards top). Regenerate tests (changes in register allocation).

This revision is now accepted and ready to land.Feb 10 2019, 10:49 AM
This revision was automatically updated to reflect the committed changes.