[PATCH] D106679: [WIP][X86] Move shl(x, 1) -> add(x, x) vector fold to ISEL (PR50468)
Simon Pilgrim via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Fri Jul 23 09:18:41 PDT 2021
RKSimon created this revision.
RKSimon added reviewers: craig.topper, spatel, pengfei.
Herald added a subscriber: hiraditya.
RKSimon requested review of this revision.
Herald added a project: LLVM.
Partial alternative to D106675 <https://reviews.llvm.org/D106675> to avoid the undef issues identified in PR50468 by using alternative isel patterns for vXi16/vXi32/vXi64 shl-by-one.
This affects many cases which weren't being picked up in time by combineShiftLeft() - which since this is a performance combine is probably a good thing - plus it avoids some HasOneUse issues that allows some additional optimization.
The vXi8 shl-by-one would still need to be performed using FREEZE - I don't know if anyone has concerns about having different methods?
Repository:
rG LLVM Github Monorepo
https://reviews.llvm.org/D106679
Files:
llvm/lib/Target/X86/X86ISelLowering.cpp
llvm/lib/Target/X86/X86InstrAVX512.td
llvm/lib/Target/X86/X86InstrSSE.td
llvm/test/CodeGen/X86/combine-mul.ll
llvm/test/CodeGen/X86/combine-sdiv.ll
llvm/test/CodeGen/X86/lower-vec-shift.ll
llvm/test/CodeGen/X86/oddsubvector.ll
llvm/test/CodeGen/X86/sdiv_fix_sat.ll
llvm/test/CodeGen/X86/sse2-intrinsics-fast-isel.ll
llvm/test/CodeGen/X86/stack-folding-int-avx512.ll
llvm/test/CodeGen/X86/udiv_fix_sat.ll
llvm/test/CodeGen/X86/vec_shift6.ll
llvm/test/CodeGen/X86/vector-fshr-128.ll
llvm/test/CodeGen/X86/vector-fshr-256.ll
llvm/test/CodeGen/X86/vector-fshr-512.ll
llvm/test/CodeGen/X86/vector-mulfix-legalize.ll
llvm/test/CodeGen/X86/vector-shift-shl-128.ll
llvm/test/CodeGen/X86/vector-shift-shl-256.ll
llvm/test/CodeGen/X86/widen_arith-4.ll
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D106679.361254.patch
Type: text/x-patch
Size: 126707 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210723/0cb4e6da/attachment.bin>
More information about the llvm-commits
mailing list