diff options
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | 2016-01-25 21:51:34 +0000 |
|---|---|---|
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | 2016-01-25 21:51:34 +0000 |
| commit | d1d118097d32b0e58aea5cdf1f413e9df2e9fc67 (patch) | |
| tree | 62392baed25520269acf10719e254e9ca4d90dc9 /llvm/lib | |
| parent | e420dd4d844ad3358bbaeed3b322a61838a6d986 (diff) | |
| download | bcm5719-llvm-d1d118097d32b0e58aea5cdf1f413e9df2e9fc67.tar.gz bcm5719-llvm-d1d118097d32b0e58aea5cdf1f413e9df2e9fc67.zip | |
[X86][AVX] Add commutation support for VPERM2X128 instructions
Its main use is to allow memory folding of the 1st operand
Differential Revision: http://reviews.llvm.org/D16521
llvm-svn: 258726
Diffstat (limited to 'llvm/lib')
| -rw-r--r-- | llvm/lib/Target/X86/X86InstrInfo.cpp | 14 | ||||
| -rw-r--r-- | llvm/lib/Target/X86/X86InstrSSE.td | 2 |
2 files changed, 16 insertions, 0 deletions
diff --git a/llvm/lib/Target/X86/X86InstrInfo.cpp b/llvm/lib/Target/X86/X86InstrInfo.cpp index 246804e3428..0c7ce7c2490 100644 --- a/llvm/lib/Target/X86/X86InstrInfo.cpp +++ b/llvm/lib/Target/X86/X86InstrInfo.cpp @@ -3275,6 +3275,20 @@ MachineInstr *X86InstrInfo::commuteInstructionImpl(MachineInstr *MI, MI->getOperand(3).setImm(Imm); return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2); } + case X86::VPERM2F128rr: + case X86::VPERM2I128rr: { + // Flip permute source immediate. + // Imm & 0x02: lo = if set, select Op1.lo/hi else Op0.lo/hi. + // Imm & 0x20: hi = if set, select Op1.lo/hi else Op0.lo/hi. + unsigned Imm = MI->getOperand(3).getImm() & 0xFF; + if (NewMI) { + MachineFunction &MF = *MI->getParent()->getParent(); + MI = MF.CloneMachineInstr(MI); + NewMI = false; + } + MI->getOperand(3).setImm(Imm ^ 0x22); + return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2); + } case X86::CMOVB16rr: case X86::CMOVB32rr: case X86::CMOVB64rr: case X86::CMOVAE16rr: case X86::CMOVAE32rr: case X86::CMOVAE64rr: case X86::CMOVE16rr: case X86::CMOVE32rr: case X86::CMOVE64rr: diff --git a/llvm/lib/Target/X86/X86InstrSSE.td b/llvm/lib/Target/X86/X86InstrSSE.td index 345e8d0ba21..5f02ea9e145 100644 --- a/llvm/lib/Target/X86/X86InstrSSE.td +++ b/llvm/lib/Target/X86/X86InstrSSE.td @@ -8158,6 +8158,7 @@ def : Pat<(v2i64 (X86VPermilpi (loadv2i64 addr:$src1), (i8 imm:$imm))), // VPERM2F128 - Permute Floating-Point Values in 128-bit chunks // let ExeDomain = SSEPackedSingle in { +let isCommutable = 1 in def VPERM2F128rr : AVXAIi8<0x06, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src1, VR256:$src2, u8imm:$src3), "vperm2f128\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}", @@ -8531,6 +8532,7 @@ defm VPERMPD : avx2_perm_imm<0x01, "vpermpd", loadv4f64, v4f64, //===----------------------------------------------------------------------===// // VPERM2I128 - Permute Floating-Point Values in 128-bit chunks // +let isCommutable = 1 in def VPERM2I128rr : AVX2AIi8<0x46, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src1, VR256:$src2, u8imm:$src3), "vperm2i128\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}", |
