X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FX86%2FX86InstrFragmentsSIMD.td;h=b75a9f4b2d029fca343704e5eef544df52b5cb37;hb=0973b7ddb8f8267132147c8b24dae7b2dfa1fd02;hp=09ab995166981a8def6d597ced2a34a1f156a350;hpb=226e0e6264dc15ea8f26261a813eae3c17987b3b;p=oota-llvm.git diff --git a/lib/Target/X86/X86InstrFragmentsSIMD.td b/lib/Target/X86/X86InstrFragmentsSIMD.td index 09ab9951669..b75a9f4b2d0 100644 --- a/lib/Target/X86/X86InstrFragmentsSIMD.td +++ b/lib/Target/X86/X86InstrFragmentsSIMD.td @@ -11,11 +11,24 @@ // //===----------------------------------------------------------------------===// +//===----------------------------------------------------------------------===// +// MMX specific DAG Nodes. +//===----------------------------------------------------------------------===// + +// Low word of MMX to GPR. +def MMX_X86movd2w : SDNode<"X86ISD::MMX_MOVD2W", SDTypeProfile<1, 1, + [SDTCisVT<0, i32>, SDTCisVT<1, x86mmx>]>>; +// GPR to low word of MMX. +def MMX_X86movw2d : SDNode<"X86ISD::MMX_MOVW2D", SDTypeProfile<1, 1, + [SDTCisVT<0, x86mmx>, SDTCisVT<1, i32>]>>; + //===----------------------------------------------------------------------===// // MMX Pattern Fragments //===----------------------------------------------------------------------===// def load_mmx : PatFrag<(ops node:$ptr), (x86mmx (load node:$ptr))>; +def load_mvmmx : PatFrag<(ops node:$ptr), + (x86mmx (MMX_X86movw2d (load node:$ptr)))>; def bc_mmx : PatFrag<(ops node:$in), (x86mmx (bitconvert node:$in))>; //===----------------------------------------------------------------------===// @@ -25,7 +38,13 @@ def bc_mmx : PatFrag<(ops node:$in), (x86mmx (bitconvert node:$in))>; def SDTX86FPShiftOp : SDTypeProfile<1, 2, [ SDTCisSameAs<0, 1>, SDTCisFP<0>, SDTCisInt<2> ]>; def SDTX86VFCMP : SDTypeProfile<1, 3, [SDTCisInt<0>, SDTCisSameAs<1, 2>, - SDTCisFP<1>, SDTCisVT<3, i8>]>; + SDTCisFP<1>, SDTCisVT<3, i8>, + SDTCisVec<1>]>; + +def X86umin : SDNode<"X86ISD::UMIN", SDTIntBinOp>; +def X86umax : SDNode<"X86ISD::UMAX", SDTIntBinOp>; +def X86smin : SDNode<"X86ISD::SMIN", SDTIntBinOp>; +def X86smax : SDNode<"X86ISD::SMAX", SDTIntBinOp>; def X86fmin : SDNode<"X86ISD::FMIN", SDTFPBinOp>; def X86fmax : SDNode<"X86ISD::FMAX", SDTFPBinOp>; @@ -42,6 +61,8 @@ def X86for : SDNode<"X86ISD::FOR", SDTFPBinOp, [SDNPCommutative, SDNPAssociative]>; def X86fxor : SDNode<"X86ISD::FXOR", SDTFPBinOp, [SDNPCommutative, SDNPAssociative]>; +def X86fandn : SDNode<"X86ISD::FANDN", SDTFPBinOp, + [SDNPCommutative, SDNPAssociative]>; def X86frsqrt : SDNode<"X86ISD::FRSQRT", SDTFPUnaryOp>; def X86frcp : SDNode<"X86ISD::FRCP", SDTFPUnaryOp>; def X86fsrl : SDNode<"X86ISD::FSRL", SDTX86FPShiftOp>; @@ -52,8 +73,8 @@ def X86hadd : SDNode<"X86ISD::HADD", SDTIntBinOp>; def X86hsub : SDNode<"X86ISD::HSUB", SDTIntBinOp>; def X86comi : SDNode<"X86ISD::COMI", SDTX86CmpTest>; def X86ucomi : SDNode<"X86ISD::UCOMI", SDTX86CmpTest>; -def X86cmpss : SDNode<"X86ISD::FSETCCss", SDTX86Cmpss>; -def X86cmpsd : SDNode<"X86ISD::FSETCCsd", SDTX86Cmpsd>; +def X86cmps : SDNode<"X86ISD::FSETCC", SDTX86Cmps>; +//def X86cmpsd : SDNode<"X86ISD::FSETCCsd", SDTX86Cmpsd>; def X86pshufb : SDNode<"X86ISD::PSHUFB", SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>, SDTCisSameAs<0,2>]>>; @@ -73,37 +94,46 @@ def X86pinsrb : SDNode<"X86ISD::PINSRB", def X86pinsrw : SDNode<"X86ISD::PINSRW", SDTypeProfile<1, 3, [SDTCisVT<0, v8i16>, SDTCisSameAs<0,1>, SDTCisVT<2, i32>, SDTCisPtrTy<3>]>>; -def X86insrtps : SDNode<"X86ISD::INSERTPS", +def X86insertps : SDNode<"X86ISD::INSERTPS", SDTypeProfile<1, 3, [SDTCisVT<0, v4f32>, SDTCisSameAs<0,1>, - SDTCisVT<2, v4f32>, SDTCisPtrTy<3>]>>; + SDTCisVT<2, v4f32>, SDTCisVT<3, i8>]>>; def X86vzmovl : SDNode<"X86ISD::VZEXT_MOVL", SDTypeProfile<1, 1, [SDTCisSameAs<0,1>]>>; -def X86vzmovly : SDNode<"X86ISD::VZEXT_MOVL", - SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>, - SDTCisOpSmallerThanOp<1, 0> ]>>; - -def X86vsmovl : SDNode<"X86ISD::VSEXT_MOVL", - SDTypeProfile<1, 1, - [SDTCisVec<0>, SDTCisInt<1>, SDTCisInt<0>]>>; - def X86vzload : SDNode<"X86ISD::VZEXT_LOAD", SDTLoad, [SDNPHasChain, SDNPMayLoad, SDNPMemOperand]>; def X86vzext : SDNode<"X86ISD::VZEXT", SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>, - SDTCisInt<0>, SDTCisInt<1>]>>; + SDTCisInt<0>, SDTCisInt<1>, + SDTCisOpSmallerThanOp<1, 0>]>>; def X86vsext : SDNode<"X86ISD::VSEXT", SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>, - SDTCisInt<0>, SDTCisInt<1>]>>; + SDTCisInt<0>, SDTCisInt<1>, + SDTCisOpSmallerThanOp<1, 0>]>>; +def X86vtrunc : SDNode<"X86ISD::VTRUNC", + SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>, + SDTCisInt<0>, SDTCisInt<1>, + SDTCisOpSmallerThanOp<0, 1>]>>; +def X86trunc : SDNode<"X86ISD::TRUNC", + SDTypeProfile<1, 1, [SDTCisInt<0>, SDTCisInt<1>, + SDTCisOpSmallerThanOp<0, 1>]>>; + +def X86vtruncm : SDNode<"X86ISD::VTRUNCM", + SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>, + SDTCisInt<0>, SDTCisInt<1>, + SDTCisVec<2>, SDTCisInt<2>, + SDTCisOpSmallerThanOp<0, 2>]>>; def X86vfpext : SDNode<"X86ISD::VFPEXT", SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>, - SDTCisFP<0>, SDTCisFP<1>]>>; + SDTCisFP<0>, SDTCisFP<1>, + SDTCisOpSmallerThanOp<1, 0>]>>; def X86vfpround: SDNode<"X86ISD::VFPROUND", SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>, - SDTCisFP<0>, SDTCisFP<1>]>>; + SDTCisFP<0>, SDTCisFP<1>, + SDTCisOpSmallerThanOp<0, 1>]>>; def X86vshldq : SDNode<"X86ISD::VSHLDQ", SDTIntShiftOp>; def X86vshrdq : SDNode<"X86ISD::VSRLDQ", SDTIntShiftOp>; @@ -111,6 +141,21 @@ def X86cmpp : SDNode<"X86ISD::CMPP", SDTX86VFCMP>; def X86pcmpeq : SDNode<"X86ISD::PCMPEQ", SDTIntBinOp, [SDNPCommutative]>; def X86pcmpgt : SDNode<"X86ISD::PCMPGT", SDTIntBinOp>; +def X86IntCmpMask : SDTypeProfile<1, 2, + [SDTCisVec<0>, SDTCisSameAs<1, 2>, SDTCisInt<1>]>; +def X86pcmpeqm : SDNode<"X86ISD::PCMPEQM", X86IntCmpMask, [SDNPCommutative]>; +def X86pcmpgtm : SDNode<"X86ISD::PCMPGTM", X86IntCmpMask>; + +def X86CmpMaskCC : + SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisInt<0>, SDTCisVec<1>, + SDTCisSameAs<1, 2>, SDTCisVT<3, i8>]>; +def X86CmpMaskCCScalar : + SDTypeProfile<1, 3, [SDTCisInt<0>, SDTCisSameAs<1, 2>, SDTCisVT<3, i8>]>; + +def X86cmpm : SDNode<"X86ISD::CMPM", X86CmpMaskCC>; +def X86cmpmu : SDNode<"X86ISD::CMPMU", X86CmpMaskCC>; +def X86cmpms : SDNode<"X86ISD::FSETCC", X86CmpMaskCCScalar>; + def X86vshl : SDNode<"X86ISD::VSHL", SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>, SDTCisVec<2>]>>; @@ -128,12 +173,24 @@ def X86vsrai : SDNode<"X86ISD::VSRAI", SDTIntShiftOp>; def SDTX86CmpPTest : SDTypeProfile<1, 2, [SDTCisVT<0, i32>, SDTCisVec<1>, SDTCisSameAs<2, 1>]>; +def X86subus : SDNode<"X86ISD::SUBUS", SDTIntBinOp>; def X86ptest : SDNode<"X86ISD::PTEST", SDTX86CmpPTest>; def X86testp : SDNode<"X86ISD::TESTP", SDTX86CmpPTest>; +def X86kortest : SDNode<"X86ISD::KORTEST", SDTX86CmpPTest>; +def X86testm : SDNode<"X86ISD::TESTM", SDTypeProfile<1, 2, [SDTCisVec<0>, + SDTCisVec<1>, + SDTCisSameAs<2, 1>]>>; +def X86testnm : SDNode<"X86ISD::TESTNM", SDTypeProfile<1, 2, [SDTCisVec<0>, + SDTCisVec<1>, + SDTCisSameAs<2, 1>]>>; +def X86select : SDNode<"X86ISD::SELECT" , SDTSelect>; def X86pmuludq : SDNode<"X86ISD::PMULUDQ", SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>, SDTCisSameAs<1,2>]>>; +def X86pmuldq : SDNode<"X86ISD::PMULDQ", + SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>, + SDTCisSameAs<1,2>]>>; // Specific shuffle nodes - At some point ISD::VECTOR_SHUFFLE will always get // translated into one of the target nodes below during lowering. @@ -141,20 +198,38 @@ def X86pmuludq : SDNode<"X86ISD::PMULUDQ", def SDTShuff1Op : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisSameAs<0,1>]>; def SDTShuff2Op : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>, SDTCisSameAs<0,2>]>; +def SDTShuff3Op : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>, + SDTCisSameAs<0,2>, SDTCisSameAs<0,3>]>; +def SDTShuff2OpM : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>, + SDTCisVec<2>]>; def SDTShuff2OpI : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>, SDTCisInt<2>]>; def SDTShuff3OpI : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>, SDTCisSameAs<0,2>, SDTCisInt<3>]>; -def SDTVBroadcast : SDTypeProfile<1, 1, [SDTCisVec<0>]>; +def SDTVBroadcast : SDTypeProfile<1, 1, [SDTCisVec<0>]>; +def SDTVBroadcastm : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>]>; + def SDTBlend : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>, - SDTCisSameAs<1,2>, SDTCisVT<3, i32>]>; + SDTCisSameAs<1,2>, SDTCisVT<3, i8>]>; + +def SDTFPBinOpRound : SDTypeProfile<1, 3, [ // fadd_round, fmul_round, etc. + SDTCisSameAs<0, 1>, SDTCisSameAs<0, 2>, SDTCisFP<0>, SDTCisInt<3>]>; def SDTFma : SDTypeProfile<1, 3, [SDTCisSameAs<0,1>, SDTCisSameAs<1,2>, SDTCisSameAs<1,3>]>; - -def X86PAlign : SDNode<"X86ISD::PALIGN", SDTShuff3OpI>; +def SDTFmaRound : SDTypeProfile<1, 4, [SDTCisSameAs<0,1>, + SDTCisSameAs<1,2>, SDTCisSameAs<1,3>, SDTCisInt<4>]>; +def STDFp1SrcRm : SDTypeProfile<1, 2, [SDTCisSameAs<0,1>, + SDTCisVec<0>, SDTCisInt<2>]>; +def STDFp2SrcRm : SDTypeProfile<1, 3, [SDTCisSameAs<0,1>, + SDTCisVec<0>, SDTCisInt<3>]>; +def STDFp3SrcRm : SDTypeProfile<1, 4, [SDTCisSameAs<0,1>, + SDTCisVec<0>, SDTCisInt<3>, SDTCisInt<4>]>; + +def X86PAlignr : SDNode<"X86ISD::PALIGNR", SDTShuff3OpI>; +def X86VAlign : SDNode<"X86ISD::VALIGN", SDTShuff3OpI>; def X86PShufd : SDNode<"X86ISD::PSHUFD", SDTShuff2OpI>; def X86PShufhw : SDNode<"X86ISD::PSHUFHW", SDTShuff2OpI>; @@ -176,18 +251,40 @@ def X86Movhlps : SDNode<"X86ISD::MOVHLPS", SDTShuff2Op>; def X86Movlps : SDNode<"X86ISD::MOVLPS", SDTShuff2Op>; def X86Movlpd : SDNode<"X86ISD::MOVLPD", SDTShuff2Op>; +def SDTPack : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>, SDTCisSameAs<2, 1>]>; +def X86Packss : SDNode<"X86ISD::PACKSS", SDTPack>; +def X86Packus : SDNode<"X86ISD::PACKUS", SDTPack>; + def X86Unpckl : SDNode<"X86ISD::UNPCKL", SDTShuff2Op>; def X86Unpckh : SDNode<"X86ISD::UNPCKH", SDTShuff2Op>; -def X86VPermilp : SDNode<"X86ISD::VPERMILP", SDTShuff2OpI>; -def X86VPermv : SDNode<"X86ISD::VPERMV", SDTShuff2Op>; -def X86VPermi : SDNode<"X86ISD::VPERMI", SDTShuff2OpI>; +def X86VPermilpv : SDNode<"X86ISD::VPERMILPV", SDTShuff2OpM>; +def X86VPermilpi : SDNode<"X86ISD::VPERMILPI", SDTShuff2OpI>; +def X86VPermv : SDNode<"X86ISD::VPERMV", SDTShuff2Op>; +def X86VPermi : SDNode<"X86ISD::VPERMI", SDTShuff2OpI>; +def X86VPermv3 : SDNode<"X86ISD::VPERMV3", SDTShuff3Op>; +def X86VPermiv3 : SDNode<"X86ISD::VPERMIV3", SDTShuff3Op>; def X86VPerm2x128 : SDNode<"X86ISD::VPERM2X128", SDTShuff3OpI>; def X86VBroadcast : SDNode<"X86ISD::VBROADCAST", SDTVBroadcast>; +def X86VBroadcastm : SDNode<"X86ISD::VBROADCASTM", SDTVBroadcastm>; +def X86Vinsert : SDNode<"X86ISD::VINSERT", SDTypeProfile<1, 3, + [SDTCisSameAs<0, 1>, SDTCisPtrTy<3>]>, []>; +def X86Vextract : SDNode<"X86ISD::VEXTRACT", SDTypeProfile<1, 2, + [SDTCisVec<1>, SDTCisPtrTy<2>]>, []>; def X86Blendi : SDNode<"X86ISD::BLENDI", SDTBlend>; + +def X86Addsub : SDNode<"X86ISD::ADDSUB", SDTFPBinOp>; + +def X86faddRnd : SDNode<"X86ISD::FADD_RND", SDTFPBinOpRound>; +def X86fsubRnd : SDNode<"X86ISD::FSUB_RND", SDTFPBinOpRound>; +def X86fmulRnd : SDNode<"X86ISD::FMUL_RND", SDTFPBinOpRound>; +def X86fdivRnd : SDNode<"X86ISD::FDIV_RND", SDTFPBinOpRound>; +def X86fmaxRnd : SDNode<"X86ISD::FMAX", SDTFPBinOpRound>; +def X86fminRnd : SDNode<"X86ISD::FMIN", SDTFPBinOpRound>; + def X86Fmadd : SDNode<"X86ISD::FMADD", SDTFma>; def X86Fnmadd : SDNode<"X86ISD::FNMADD", SDTFma>; def X86Fmsub : SDNode<"X86ISD::FMSUB", SDTFma>; @@ -195,6 +292,21 @@ def X86Fnmsub : SDNode<"X86ISD::FNMSUB", SDTFma>; def X86Fmaddsub : SDNode<"X86ISD::FMADDSUB", SDTFma>; def X86Fmsubadd : SDNode<"X86ISD::FMSUBADD", SDTFma>; +def X86FmaddRnd : SDNode<"X86ISD::FMADD_RND", SDTFmaRound>; +def X86FnmaddRnd : SDNode<"X86ISD::FNMADD_RND", SDTFmaRound>; +def X86FmsubRnd : SDNode<"X86ISD::FMSUB_RND", SDTFmaRound>; +def X86FnmsubRnd : SDNode<"X86ISD::FNMSUB_RND", SDTFmaRound>; +def X86FmaddsubRnd : SDNode<"X86ISD::FMADDSUB_RND", SDTFmaRound>; +def X86FmsubaddRnd : SDNode<"X86ISD::FMSUBADD_RND", SDTFmaRound>; + +def X86rsqrt28 : SDNode<"X86ISD::RSQRT28", STDFp1SrcRm>; +def X86rcp28 : SDNode<"X86ISD::RCP28", STDFp1SrcRm>; +def X86exp2 : SDNode<"X86ISD::EXP2", STDFp1SrcRm>; + +def X86rsqrt28s : SDNode<"X86ISD::RSQRT28", STDFp2SrcRm>; +def X86rcp28s : SDNode<"X86ISD::RCP28", STDFp2SrcRm>; +def X86RndScale : SDNode<"X86ISD::RNDSCALE", STDFp3SrcRm>; + def SDT_PCMPISTRI : SDTypeProfile<2, 3, [SDTCisVT<0, i32>, SDTCisVT<1, i32>, SDTCisVT<2, v16i8>, SDTCisVT<3, v16i8>, SDTCisVT<4, i8>]>; @@ -206,6 +318,13 @@ def SDT_PCMPESTRI : SDTypeProfile<2, 5, [SDTCisVT<0, i32>, SDTCisVT<1, i32>, def X86pcmpistri : SDNode<"X86ISD::PCMPISTRI", SDT_PCMPISTRI>; def X86pcmpestri : SDNode<"X86ISD::PCMPESTRI", SDT_PCMPESTRI>; +def X86compress: SDNode<"X86ISD::COMPRESS", SDTypeProfile<1, 3, + [SDTCisSameAs<0, 2>, SDTCisSameAs<0, 3>, + SDTCisVec<3>, SDTCisVec<1>, SDTCisInt<1>]>, []>; +def X86expand : SDNode<"X86ISD::EXPAND", SDTypeProfile<1, 3, + [SDTCisSameAs<0, 3>, + SDTCisVec<3>, SDTCisVec<1>, SDTCisInt<1>]>, []>; + //===----------------------------------------------------------------------===// // SSE Complex Patterns //===----------------------------------------------------------------------===// @@ -223,13 +342,13 @@ def sse_load_f64 : ComplexPattern { let PrintMethod = "printf32mem"; let MIOperandInfo = (ops ptr_rc, i8imm, ptr_rc_nosp, i32imm, i8imm); - let ParserMatchClass = X86MemAsmOperand; + let ParserMatchClass = X86Mem32AsmOperand; let OperandType = "OPERAND_MEMORY"; } def sdmem : Operand { let PrintMethod = "printf64mem"; let MIOperandInfo = (ops ptr_rc, i8imm, ptr_rc_nosp, i32imm, i8imm); - let ParserMatchClass = X86MemAsmOperand; + let ParserMatchClass = X86Mem64AsmOperand; let OperandType = "OPERAND_MEMORY"; } @@ -249,9 +368,27 @@ def loadv8f32 : PatFrag<(ops node:$ptr), (v8f32 (load node:$ptr))>; def loadv4f64 : PatFrag<(ops node:$ptr), (v4f64 (load node:$ptr))>; def loadv4i64 : PatFrag<(ops node:$ptr), (v4i64 (load node:$ptr))>; -// 128-/256-bit extload pattern fragments +// 512-bit load pattern fragments +def loadv16f32 : PatFrag<(ops node:$ptr), (v16f32 (load node:$ptr))>; +def loadv8f64 : PatFrag<(ops node:$ptr), (v8f64 (load node:$ptr))>; +def loadv64i8 : PatFrag<(ops node:$ptr), (v64i8 (load node:$ptr))>; +def loadv32i16 : PatFrag<(ops node:$ptr), (v32i16 (load node:$ptr))>; +def loadv16i32 : PatFrag<(ops node:$ptr), (v16i32 (load node:$ptr))>; +def loadv8i64 : PatFrag<(ops node:$ptr), (v8i64 (load node:$ptr))>; + +// 128-/256-/512-bit extload pattern fragments def extloadv2f32 : PatFrag<(ops node:$ptr), (v2f64 (extloadvf32 node:$ptr))>; def extloadv4f32 : PatFrag<(ops node:$ptr), (v4f64 (extloadvf32 node:$ptr))>; +def extloadv8f32 : PatFrag<(ops node:$ptr), (v8f64 (extloadvf32 node:$ptr))>; + +// These are needed to match a scalar load that is used in a vector-only +// math instruction such as the FP logical ops: andps, andnps, orps, xorps. +// The memory operand is required to be a 128-bit load, so it must be converted +// from a vector to a scalar. +def loadf32_128 : PatFrag<(ops node:$ptr), + (f32 (vector_extract (loadv4f32 node:$ptr), (iPTR 0)))>; +def loadf64_128 : PatFrag<(ops node:$ptr), + (f64 (vector_extract (loadv2f64 node:$ptr), (iPTR 0)))>; // Like 'store', but always requires 128-bit vector alignment. def alignedstore : PatFrag<(ops node:$val, node:$ptr), @@ -265,6 +402,12 @@ def alignedstore256 : PatFrag<(ops node:$val, node:$ptr), return cast(N)->getAlignment() >= 32; }]>; +// Like 'store', but always requires 512-bit vector alignment. +def alignedstore512 : PatFrag<(ops node:$val, node:$ptr), + (store node:$val, node:$ptr), [{ + return cast(N)->getAlignment() >= 64; +}]>; + // Like 'load', but always requires 128-bit vector alignment. def alignedload : PatFrag<(ops node:$ptr), (load node:$ptr), [{ return cast(N)->getAlignment() >= 16; @@ -280,6 +423,11 @@ def alignedload256 : PatFrag<(ops node:$ptr), (load node:$ptr), [{ return cast(N)->getAlignment() >= 32; }]>; +// Like 'load', but always requires 512-bit vector alignment. +def alignedload512 : PatFrag<(ops node:$ptr), (load node:$ptr), [{ + return cast(N)->getAlignment() >= 64; +}]>; + def alignedloadfsf32 : PatFrag<(ops node:$ptr), (f32 (alignedload node:$ptr))>; def alignedloadfsf64 : PatFrag<(ops node:$ptr), @@ -303,6 +451,16 @@ def alignedloadv4f64 : PatFrag<(ops node:$ptr), def alignedloadv4i64 : PatFrag<(ops node:$ptr), (v4i64 (alignedload256 node:$ptr))>; +// 512-bit aligned load pattern fragments +def alignedloadv16f32 : PatFrag<(ops node:$ptr), + (v16f32 (alignedload512 node:$ptr))>; +def alignedloadv16i32 : PatFrag<(ops node:$ptr), + (v16i32 (alignedload512 node:$ptr))>; +def alignedloadv8f64 : PatFrag<(ops node:$ptr), + (v8f64 (alignedload512 node:$ptr))>; +def alignedloadv8i64 : PatFrag<(ops node:$ptr), + (v8i64 (alignedload512 node:$ptr))>; + // Like 'load', but uses special alignment checks suitable for use in // memory operands in most SSE instructions, which are required to // be naturally aligned on some targets but not on others. If the subtarget @@ -310,7 +468,7 @@ def alignedloadv4i64 : PatFrag<(ops node:$ptr), // setting a feature bit in the processor (on startup, for example). // Opteron 10h and later implement such a feature. def memop : PatFrag<(ops node:$ptr), (load node:$ptr), [{ - return Subtarget->hasVectorUAMem() + return Subtarget->hasSSEUnalignedMem() || cast(N)->getAlignment() >= 16; }]>; @@ -323,11 +481,15 @@ def memopv4f32 : PatFrag<(ops node:$ptr), (v4f32 (memop node:$ptr))>; def memopv2f64 : PatFrag<(ops node:$ptr), (v2f64 (memop node:$ptr))>; def memopv2i64 : PatFrag<(ops node:$ptr), (v2i64 (memop node:$ptr))>; -// 256-bit memop pattern fragments -// NOTE: all 256-bit integer vector loads are promoted to v4i64 -def memopv8f32 : PatFrag<(ops node:$ptr), (v8f32 (memop node:$ptr))>; -def memopv4f64 : PatFrag<(ops node:$ptr), (v4f64 (memop node:$ptr))>; -def memopv4i64 : PatFrag<(ops node:$ptr), (v4i64 (memop node:$ptr))>; +// These are needed to match a scalar memop that is used in a vector-only +// math instruction such as the FP logical ops: andps, andnps, orps, xorps. +// The memory operand is required to be a 128-bit load, so it must be converted +// from a vector to a scalar. +def memopfsf32_128 : PatFrag<(ops node:$ptr), + (f32 (vector_extract (memopv4f32 node:$ptr), (iPTR 0)))>; +def memopfsf64_128 : PatFrag<(ops node:$ptr), + (f64 (vector_extract (memopv2f64 node:$ptr), (iPTR 0)))>; + // SSSE3 uses MMX registers for some instructions. They aren't aligned on a // 16-byte boundary. @@ -377,6 +539,13 @@ def bc_v32i8 : PatFrag<(ops node:$in), (v32i8 (bitconvert node:$in))>; def bc_v16i16 : PatFrag<(ops node:$in), (v16i16 (bitconvert node:$in))>; def bc_v8i32 : PatFrag<(ops node:$in), (v8i32 (bitconvert node:$in))>; def bc_v4i64 : PatFrag<(ops node:$in), (v4i64 (bitconvert node:$in))>; +def bc_v8f32 : PatFrag<(ops node:$in), (v8f32 (bitconvert node:$in))>; + +// 512-bit bitconvert pattern fragments +def bc_v16i32 : PatFrag<(ops node:$in), (v16i32 (bitconvert node:$in))>; +def bc_v8i64 : PatFrag<(ops node:$in), (v8i64 (bitconvert node:$in))>; +def bc_v8f64 : PatFrag<(ops node:$in), (v8f64 (bitconvert node:$in))>; +def bc_v16f32 : PatFrag<(ops node:$in), (v16f32 (bitconvert node:$in))>; def vzmovl_v2i64 : PatFrag<(ops node:$src), (bitconvert (v2i64 (X86vzmovl @@ -393,34 +562,122 @@ def fp32imm0 : PatLeaf<(f32 fpimm), [{ return N->isExactlyValue(+0.0); }]>; +def I8Imm : SDNodeXFormgetZExtValue()); +}]>; + +def FROUND_NO_EXC : ImmLeaf; +def FROUND_CURRENT : ImmLeaf; + // BYTE_imm - Transform bit immediates into byte immediates. def BYTE_imm : SDNodeXForm> 3 return getI32Imm(N->getZExtValue() >> 3); }]>; -// EXTRACT_get_vextractf128_imm xform function: convert extract_subvector index -// to VEXTRACTF128 imm. -def EXTRACT_get_vextractf128_imm : SDNodeXForm; + +// INSERT_get_vinsert128_imm xform function: convert insert_subvector index to +// VINSERTF128/VINSERTI128 imm. +def INSERT_get_vinsert128_imm : SDNodeXForm; -// INSERT_get_vinsertf128_imm xform function: convert insert_subvector index to -// VINSERTF128 imm. -def INSERT_get_vinsertf128_imm : SDNodeXForm; -def vextractf128_extract : PatFrag<(ops node:$bigvec, node:$index), +// INSERT_get_vinsert256_imm xform function: convert insert_subvector index to +// VINSERTF64x4 imm. +def INSERT_get_vinsert256_imm : SDNodeXForm; + +def vextract128_extract : PatFrag<(ops node:$bigvec, node:$index), (extract_subvector node:$bigvec, node:$index), [{ - return X86::isVEXTRACTF128Index(N); -}], EXTRACT_get_vextractf128_imm>; + return X86::isVEXTRACT128Index(N); +}], EXTRACT_get_vextract128_imm>; -def vinsertf128_insert : PatFrag<(ops node:$bigvec, node:$smallvec, +def vinsert128_insert : PatFrag<(ops node:$bigvec, node:$smallvec, node:$index), (insert_subvector node:$bigvec, node:$smallvec, node:$index), [{ - return X86::isVINSERTF128Index(N); -}], INSERT_get_vinsertf128_imm>; + return X86::isVINSERT128Index(N); +}], INSERT_get_vinsert128_imm>; + + +def vextract256_extract : PatFrag<(ops node:$bigvec, node:$index), + (extract_subvector node:$bigvec, + node:$index), [{ + return X86::isVEXTRACT256Index(N); +}], EXTRACT_get_vextract256_imm>; + +def vinsert256_insert : PatFrag<(ops node:$bigvec, node:$smallvec, + node:$index), + (insert_subvector node:$bigvec, node:$smallvec, + node:$index), [{ + return X86::isVINSERT256Index(N); +}], INSERT_get_vinsert256_imm>; + +def masked_load_aligned128 : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_load node:$src1, node:$src2, node:$src3), [{ + if (auto *Load = dyn_cast(N)) + return Load->getAlignment() >= 16; + return false; +}]>; + +def masked_load_aligned256 : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_load node:$src1, node:$src2, node:$src3), [{ + if (auto *Load = dyn_cast(N)) + return Load->getAlignment() >= 32; + return false; +}]>; + +def masked_load_aligned512 : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_load node:$src1, node:$src2, node:$src3), [{ + if (auto *Load = dyn_cast(N)) + return Load->getAlignment() >= 64; + return false; +}]>; + +def masked_load_unaligned : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_load node:$src1, node:$src2, node:$src3), [{ + return isa(N); +}]>; + +def masked_store_aligned128 : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_store node:$src1, node:$src2, node:$src3), [{ + if (auto *Store = dyn_cast(N)) + return Store->getAlignment() >= 16; + return false; +}]>; + +def masked_store_aligned256 : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_store node:$src1, node:$src2, node:$src3), [{ + if (auto *Store = dyn_cast(N)) + return Store->getAlignment() >= 32; + return false; +}]>; + +def masked_store_aligned512 : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_store node:$src1, node:$src2, node:$src3), [{ + if (auto *Store = dyn_cast(N)) + return Store->getAlignment() >= 64; + return false; +}]>; + +def masked_store_unaligned : PatFrag<(ops node:$src1, node:$src2, node:$src3), + (masked_store node:$src1, node:$src2, node:$src3), [{ + return isa(N); +}]>;