Currently we expand and scalarize these operations, but I think we should be able to implement ADD/SUB with KXOR and MUL with KAND.
We already do this for scalar i1 operations so I just extended it to vectors of i1.
Paths
| Differential D28888
[AVX-512] Support ADD/SUB/MUL of mask vectors ClosedPublic Authored by craig.topper on Jan 18 2017, 8:05 PM.
Details Summary Currently we expand and scalarize these operations, but I think we should be able to implement ADD/SUB with KXOR and MUL with KAND. We already do this for scalar i1 operations so I just extended it to vectors of i1.
Diff Detail Event TimelineThis revision is now accepted and ready to land.Jan 18 2017, 11:06 PM Closed by commit rL292474: [AVX-512] Support ADD/SUB/MUL of mask vectors (authored by ctopper). · Explain WhyJan 18 2017, 11:23 PM This revision was automatically updated to reflect the committed changes.
Revision Contents
Diff 84925 lib/Target/X86/X86ISelLowering.cpp
test/CodeGen/X86/avx512-mask-op.ll
test/CodeGen/X86/avx512bw-mask-op.ll
|