4 Commits

Author SHA1 Message Date
Nikita Popov
9a9421a461 Reapply [InstCombine] Fold multiuse shr eq zero
This was reverted due to performance regressions in ARM benchmarks,
which have since been addressed by D101196 (SCEV analysis improvement)
and D101778 (CGP reverse transform).

-----

The single-use case is handled implicity by converting the icmp
into a mask check first. When comparing with zero in particular,
we don't need the one-use restriction, as we only produce a single
icmp.

https://alive2.llvm.org/ce/z/MSixcm
https://alive2.llvm.org/ce/z/GwpG0M
2021-05-22 14:46:50 +02:00
Nikita Popov
24e9fbc1a3 Revert "[InstCombine] Fold multiuse shr eq zero"
This reverts commit 9423f78240a216e3f38b394a41fe3427dee22c26.

A performance regression with this patch has been reported at
https://reviews.llvm.org/rG9423f78240a2#990953. Reverting for now.
2021-04-21 21:40:52 +02:00
Nikita Popov
9423f78240 [InstCombine] Fold multiuse shr eq zero
The single-use case is handled implicity by converting the icmp
into a mask check first. When comparing with zero in particular,
we don't need the one-use restriction, as we only produce a single
icmp.

https://alive2.llvm.org/ce/z/MSixcm
https://alive2.llvm.org/ce/z/GwpG0M
2021-04-19 22:13:11 +02:00
Craig Topper
36b5d09b07 [X86] Add phase ordering test for the problem D99427 is trying to solve. NFC 2021-03-28 12:14:30 -07:00