[llvm] r238635 - [x86] Restructure the parallel bitmath lowering of popcount into
Bruno Cardoso Lopes
bruno.cardoso at gmail.com
Mon Jun 1 08:10:14 PDT 2015
Hi Chandler,
Thanks for all the work, all this came out pretty nice :-)
Did you had a chance to look at http://reviews.llvm.org/D10002 ?
I was waiting for the x86 part of the work to land in order to update it.
Given the way you nicely layout the x86 vector ctpop lowering, looks
like we could have a target independent expansion for vector ctpop as
such:
T = invec type
X = bitcast in_vec to vXi8
A = CTPOPvXi8(X)
B = HorizontalByteSum(A, T)
where, when not custom lowered:
CTPOPvXi8 => parallel bitmath
HorizontalByteSum => shifts + adds.
I guess that would require HorizontalByteSum to be represented by a
new ISD node.
Let me know your thoughts on this.
Cheers,
On Sat, May 30, 2015 at 12:20 AM, Chandler Carruth <chandlerc at gmail.com> wrote:
> Author: chandlerc
> Date: Fri May 29 22:20:55 2015
> New Revision: 238635
>
> URL: http://llvm.org/viewvc/llvm-project?rev=238635&view=rev
> Log:
> [x86] Restructure the parallel bitmath lowering of popcount into
> a separate routine, generalize it to work for all the integer vector
> sizes, and do general code cleanups.
>
> This dramatically improves lowerings of byte and short element vector
> popcount, but more importantly it will make the introduction of the
> LUT-approach much cleaner.
>
> The biggest cleanup I've done is to just force the legalizer to do the
> bitcasting we need. We run these iteratively now and it makes the code
> much simpler IMO. Other changes were minor, and mostly naming and
> splitting things up in a way that makes it more clear what is going on.
>
> The other significant change is to use a different final horizontal sum
> approach. This is the same number of instructions as the old method, but
> shifts left instead of right so that we can clear everything but the
> final sum with a single shift right. This seems likely better than
> a mask which will usually have to read the mask from memory. It is
> certaily fewer u-ops. Also, this will be temporary. This and the LUT
> approach share the need of horizontal adds to finish the computation,
> and we have more clever approaches than this one that I'll switch over
> to.
>
> Modified:
> llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
> llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll
> llvm/trunk/test/CodeGen/X86/vector-popcnt-256.ll
>
> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=238635&r1=238634&r2=238635&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Fri May 29 22:20:55 2015
> @@ -846,8 +846,10 @@ X86TargetLowering::X86TargetLowering(con
> // know to perform better than using the popcnt instructions on each vector
> // element. If popcnt isn't supported, always provide the custom version.
> if (!Subtarget->hasPOPCNT()) {
> - setOperationAction(ISD::CTPOP, MVT::v4i32, Custom);
> setOperationAction(ISD::CTPOP, MVT::v2i64, Custom);
> + setOperationAction(ISD::CTPOP, MVT::v4i32, Custom);
> + setOperationAction(ISD::CTPOP, MVT::v8i16, Custom);
> + setOperationAction(ISD::CTPOP, MVT::v16i8, Custom);
> }
>
> // Custom lower build_vector, vector_shuffle, and extract_vector_elt.
> @@ -17327,141 +17329,131 @@ static SDValue LowerBITCAST(SDValue Op,
> return SDValue();
> }
>
> -static SDValue LowerCTPOP(SDValue Op, const X86Subtarget *Subtarget,
> - SelectionDAG &DAG) {
> - SDNode *Node = Op.getNode();
> - SDLoc dl(Node);
> -
> - Op = Op.getOperand(0);
> - EVT VT = Op.getValueType();
> +static SDValue LowerVectorCTPOPBitmath(SDValue Op, SDLoc DL,
> + const X86Subtarget *Subtarget,
> + SelectionDAG &DAG) {
> + MVT VT = Op.getSimpleValueType();
> assert((VT.is128BitVector() || VT.is256BitVector()) &&
> "CTPOP lowering only implemented for 128/256-bit wide vector types");
>
> - unsigned NumElts = VT.getVectorNumElements();
> - EVT EltVT = VT.getVectorElementType();
> - unsigned Len = EltVT.getSizeInBits();
> + int VecSize = VT.getSizeInBits();
> + int NumElts = VT.getVectorNumElements();
> + MVT EltVT = VT.getVectorElementType();
> + int Len = EltVT.getSizeInBits();
>
> // This is the vectorized version of the "best" algorithm from
> // http://graphics.stanford.edu/~seander/bithacks.html#CountBitsSetParallel
> // with a minor tweak to use a series of adds + shifts instead of vector
> - // multiplications. Implemented for the v2i64, v4i64, v4i32, v8i32 types:
> - //
> - // v2i64, v4i64, v4i32 => Only profitable w/ popcnt disabled
> - // v8i32 => Always profitable
> - //
> - // FIXME: There a couple of possible improvements:
> - //
> - // 1) Support for i8 and i16 vectors (needs measurements if popcnt enabled).
> - // 2) Use strategies from http://wm.ite.pl/articles/sse-popcount.html
> + // multiplications. Implemented for all integer vector types.
> //
> - assert(EltVT.isInteger() && (Len == 32 || Len == 64) && Len % 8 == 0 &&
> - "CTPOP not implemented for this vector element type.");
> + // FIXME: Use strategies from http://wm.ite.pl/articles/sse-popcount.html
>
> - // X86 canonicalize ANDs to vXi64, generate the appropriate bitcasts to avoid
> - // extra legalization.
> - bool NeedsBitcast = EltVT == MVT::i32;
> - MVT BitcastVT = VT.is256BitVector() ? MVT::v4i64 : MVT::v2i64;
> -
> - SDValue Cst55 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x55)), dl,
> + SDValue Cst55 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x55)), DL,
> EltVT);
> - SDValue Cst33 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x33)), dl,
> + SDValue Cst33 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x33)), DL,
> EltVT);
> - SDValue Cst0F = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x0F)), dl,
> + SDValue Cst0F = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x0F)), DL,
> EltVT);
>
> + SDValue V = Op;
> +
> // v = v - ((v >> 1) & 0x55555555...)
> - SmallVector<SDValue, 8> Ones(NumElts, DAG.getConstant(1, dl, EltVT));
> - SDValue OnesV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Ones);
> - SDValue Srl = DAG.getNode(ISD::SRL, dl, VT, Op, OnesV);
> - if (NeedsBitcast)
> - Srl = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Srl);
> + SmallVector<SDValue, 8> Ones(NumElts, DAG.getConstant(1, DL, EltVT));
> + SDValue OnesV = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Ones);
> + SDValue Srl = DAG.getNode(ISD::SRL, DL, VT, V, OnesV);
>
> SmallVector<SDValue, 8> Mask55(NumElts, Cst55);
> - SDValue M55 = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Mask55);
> - if (NeedsBitcast)
> - M55 = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M55);
> -
> - SDValue And = DAG.getNode(ISD::AND, dl, Srl.getValueType(), Srl, M55);
> - if (VT != And.getValueType())
> - And = DAG.getNode(ISD::BITCAST, dl, VT, And);
> - SDValue Sub = DAG.getNode(ISD::SUB, dl, VT, Op, And);
> + SDValue M55 = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Mask55);
> + SDValue And = DAG.getNode(ISD::AND, DL, Srl.getValueType(), Srl, M55);
> +
> + V = DAG.getNode(ISD::SUB, DL, VT, V, And);
>
> // v = (v & 0x33333333...) + ((v >> 2) & 0x33333333...)
> SmallVector<SDValue, 8> Mask33(NumElts, Cst33);
> - SDValue M33 = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Mask33);
> - SmallVector<SDValue, 8> Twos(NumElts, DAG.getConstant(2, dl, EltVT));
> - SDValue TwosV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Twos);
> -
> - Srl = DAG.getNode(ISD::SRL, dl, VT, Sub, TwosV);
> - if (NeedsBitcast) {
> - Srl = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Srl);
> - M33 = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M33);
> - Sub = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Sub);
> - }
> -
> - SDValue AndRHS = DAG.getNode(ISD::AND, dl, M33.getValueType(), Srl, M33);
> - SDValue AndLHS = DAG.getNode(ISD::AND, dl, M33.getValueType(), Sub, M33);
> - if (VT != AndRHS.getValueType()) {
> - AndRHS = DAG.getNode(ISD::BITCAST, dl, VT, AndRHS);
> - AndLHS = DAG.getNode(ISD::BITCAST, dl, VT, AndLHS);
> - }
> - SDValue Add = DAG.getNode(ISD::ADD, dl, VT, AndLHS, AndRHS);
> + SDValue M33 = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Mask33);
> + SDValue AndLHS = DAG.getNode(ISD::AND, DL, M33.getValueType(), V, M33);
> +
> + SmallVector<SDValue, 8> Twos(NumElts, DAG.getConstant(2, DL, EltVT));
> + SDValue TwosV = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Twos);
> + Srl = DAG.getNode(ISD::SRL, DL, VT, V, TwosV);
> + SDValue AndRHS = DAG.getNode(ISD::AND, DL, M33.getValueType(), Srl, M33);
> +
> + V = DAG.getNode(ISD::ADD, DL, VT, AndLHS, AndRHS);
>
> // v = (v + (v >> 4)) & 0x0F0F0F0F...
> - SmallVector<SDValue, 8> Fours(NumElts, DAG.getConstant(4, dl, EltVT));
> - SDValue FoursV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Fours);
> - Srl = DAG.getNode(ISD::SRL, dl, VT, Add, FoursV);
> - Add = DAG.getNode(ISD::ADD, dl, VT, Add, Srl);
> + SmallVector<SDValue, 8> Fours(NumElts, DAG.getConstant(4, DL, EltVT));
> + SDValue FoursV = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Fours);
> + Srl = DAG.getNode(ISD::SRL, DL, VT, V, FoursV);
> + SDValue Add = DAG.getNode(ISD::ADD, DL, VT, V, Srl);
>
> SmallVector<SDValue, 8> Mask0F(NumElts, Cst0F);
> - SDValue M0F = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Mask0F);
> - if (NeedsBitcast) {
> - Add = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Add);
> - M0F = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M0F);
> - }
> - And = DAG.getNode(ISD::AND, dl, M0F.getValueType(), Add, M0F);
> - if (VT != And.getValueType())
> - And = DAG.getNode(ISD::BITCAST, dl, VT, And);
> + SDValue M0F = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Mask0F);
>
> - // The algorithm mentioned above uses:
> - // v = (v * 0x01010101...) >> (Len - 8)
> - //
> - // Change it to use vector adds + vector shifts which yield faster results on
> - // Haswell than using vector integer multiplication.
> - //
> - // For i32 elements:
> - // v = v + (v >> 8)
> - // v = v + (v >> 16)
> - //
> - // For i64 elements:
> - // v = v + (v >> 8)
> - // v = v + (v >> 16)
> - // v = v + (v >> 32)
> + V = DAG.getNode(ISD::AND, DL, M0F.getValueType(), Add, M0F);
> +
> + // At this point, V contains the byte-wise population count, and we are
> + // merely doing a horizontal sum if necessary to get the wider element
> + // counts.
> //
> - Add = And;
> + // FIXME: There is a different lowering strategy above for the horizontal sum
> + // of byte-wise population counts. This one and that one should be merged,
> + // using the fastest of the two for each size.
> + MVT ByteVT = MVT::getVectorVT(MVT::i8, VecSize / 8);
> + MVT ShiftVT = MVT::getVectorVT(MVT::i64, VecSize / 64);
> + V = DAG.getNode(ISD::BITCAST, DL, ByteVT, V);
> SmallVector<SDValue, 8> Csts;
> - for (unsigned i = 8; i <= Len/2; i *= 2) {
> - Csts.assign(NumElts, DAG.getConstant(i, dl, EltVT));
> - SDValue CstsV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Csts);
> - Srl = DAG.getNode(ISD::SRL, dl, VT, Add, CstsV);
> - Add = DAG.getNode(ISD::ADD, dl, VT, Add, Srl);
> - Csts.clear();
> + assert(Len <= 64 && "We don't support element sizes of more than 64 bits!");
> + assert(isPowerOf2_32(Len) && "Only power of two element sizes supported!");
> + for (int i = Len; i > 8; i /= 2) {
> + Csts.assign(VecSize / 64, DAG.getConstant(i / 2, DL, MVT::i64));
> + SDValue Shl = DAG.getNode(
> + ISD::SHL, DL, ShiftVT, DAG.getNode(ISD::BITCAST, DL, ShiftVT, V),
> + DAG.getNode(ISD::BUILD_VECTOR, DL, ShiftVT, Csts));
> + V = DAG.getNode(ISD::ADD, DL, ByteVT, V,
> + DAG.getNode(ISD::BITCAST, DL, ByteVT, Shl));
> + }
> +
> + // The high byte now contains the sum of the element bytes. Shift it right
> + // (if needed) to make it the low byte.
> + V = DAG.getNode(ISD::BITCAST, DL, VT, V);
> + if (Len > 8) {
> + Csts.assign(NumElts, DAG.getConstant(Len - 8, DL, EltVT));
> + V = DAG.getNode(ISD::SRL, DL, VT, V,
> + DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Csts));
> }
> + return V;
> +}
>
> - // The result is on the least significant 6-bits on i32 and 7-bits on i64.
> - SDValue Cst3F = DAG.getConstant(APInt(Len, Len == 32 ? 0x3F : 0x7F), dl,
> - EltVT);
> - SmallVector<SDValue, 8> Cst3FV(NumElts, Cst3F);
> - SDValue M3F = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Cst3FV);
> - if (NeedsBitcast) {
> - Add = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Add);
> - M3F = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M3F);
> - }
> - And = DAG.getNode(ISD::AND, dl, M3F.getValueType(), Add, M3F);
> - if (VT != And.getValueType())
> - And = DAG.getNode(ISD::BITCAST, dl, VT, And);
>
> - return And;
> +static SDValue LowerVectorCTPOP(SDValue Op, const X86Subtarget *Subtarget,
> + SelectionDAG &DAG) {
> + MVT VT = Op.getSimpleValueType();
> + // FIXME: Need to add AVX-512 support here!
> + assert((VT.is256BitVector() || VT.is128BitVector()) &&
> + "Unknown CTPOP type to handle");
> + SDLoc DL(Op.getNode());
> + SDValue Op0 = Op.getOperand(0);
> +
> + if (VT.is256BitVector() && !Subtarget->hasInt256()) {
> + unsigned NumElems = VT.getVectorNumElements();
> +
> + // Extract each 128-bit vector, compute pop count and concat the result.
> + SDValue LHS = Extract128BitVector(Op0, 0, DAG, DL);
> + SDValue RHS = Extract128BitVector(Op0, NumElems/2, DAG, DL);
> +
> + return DAG.getNode(ISD::CONCAT_VECTORS, DL, VT,
> + LowerVectorCTPOPBitmath(LHS, DL, Subtarget, DAG),
> + LowerVectorCTPOPBitmath(RHS, DL, Subtarget, DAG));
> + }
> +
> + return LowerVectorCTPOPBitmath(Op0, DL, Subtarget, DAG);
> +}
> +
> +static SDValue LowerCTPOP(SDValue Op, const X86Subtarget *Subtarget,
> + SelectionDAG &DAG) {
> + assert(Op.getValueType().isVector() &&
> + "We only do custom lowering for vector population count.");
> + return LowerVectorCTPOP(Op, Subtarget, DAG);
> }
>
> static SDValue LowerLOAD_SUB(SDValue Op, SelectionDAG &DAG) {
>
> Modified: llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll?rev=238635&r1=238634&r2=238635&view=diff
> ==============================================================================
> --- llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll (original)
> +++ llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll Fri May 29 22:20:55 2015
> @@ -25,15 +25,15 @@ define <2 x i64> @testv2i64(<2 x i64> %i
> ; SSE-NEXT: paddq %xmm0, %xmm1
> ; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> ; SSE-NEXT: movdqa %xmm1, %xmm0
> -; SSE-NEXT: psrlq $8, %xmm0
> -; SSE-NEXT: paddq %xmm1, %xmm0
> +; SSE-NEXT: psllq $32, %xmm0
> +; SSE-NEXT: paddb %xmm1, %xmm0
> ; SSE-NEXT: movdqa %xmm0, %xmm1
> -; SSE-NEXT: psrlq $16, %xmm1
> -; SSE-NEXT: paddq %xmm0, %xmm1
> +; SSE-NEXT: psllq $16, %xmm1
> +; SSE-NEXT: paddb %xmm0, %xmm1
> ; SSE-NEXT: movdqa %xmm1, %xmm0
> -; SSE-NEXT: psrlq $32, %xmm0
> -; SSE-NEXT: paddq %xmm1, %xmm0
> -; SSE-NEXT: pand {{.*}}(%rip), %xmm0
> +; SSE-NEXT: psllq $8, %xmm0
> +; SSE-NEXT: paddb %xmm1, %xmm0
> +; SSE-NEXT: psrlq $56, %xmm0
> ; SSE-NEXT: retq
> ;
> ; AVX-LABEL: testv2i64:
> @@ -49,13 +49,13 @@ define <2 x i64> @testv2i64(<2 x i64> %i
> ; AVX-NEXT: vpsrlq $4, %xmm0, %xmm1
> ; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> ; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> -; AVX-NEXT: vpsrlq $8, %xmm0, %xmm1
> -; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> -; AVX-NEXT: vpsrlq $16, %xmm0, %xmm1
> -; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> -; AVX-NEXT: vpsrlq $32, %xmm0, %xmm1
> -; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> -; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $32, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $16, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsrlq $56, %xmm0, %xmm0
> ; AVX-NEXT: retq
> %out = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %in)
> ret <2 x i64> %out
> @@ -79,12 +79,12 @@ define <4 x i32> @testv4i32(<4 x i32> %i
> ; SSE-NEXT: paddd %xmm0, %xmm1
> ; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> ; SSE-NEXT: movdqa %xmm1, %xmm2
> -; SSE-NEXT: psrld $8, %xmm2
> -; SSE-NEXT: paddd %xmm1, %xmm2
> +; SSE-NEXT: psllq $16, %xmm2
> +; SSE-NEXT: paddb %xmm1, %xmm2
> ; SSE-NEXT: movdqa %xmm2, %xmm0
> -; SSE-NEXT: psrld $16, %xmm0
> -; SSE-NEXT: paddd %xmm2, %xmm0
> -; SSE-NEXT: pand {{.*}}(%rip), %xmm0
> +; SSE-NEXT: psllq $8, %xmm0
> +; SSE-NEXT: paddb %xmm2, %xmm0
> +; SSE-NEXT: psrld $24, %xmm0
> ; SSE-NEXT: retq
> ;
> ; AVX1-LABEL: testv4i32:
> @@ -100,11 +100,11 @@ define <4 x i32> @testv4i32(<4 x i32> %i
> ; AVX1-NEXT: vpsrld $4, %xmm0, %xmm1
> ; AVX1-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> ; AVX1-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> -; AVX1-NEXT: vpsrld $8, %xmm0, %xmm1
> -; AVX1-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX1-NEXT: vpsrld $16, %xmm0, %xmm1
> -; AVX1-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX1-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX1-NEXT: vpsllq $16, %xmm0, %xmm1
> +; AVX1-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX1-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX1-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX1-NEXT: vpsrld $24, %xmm0, %xmm0
> ; AVX1-NEXT: retq
> ;
> ; AVX2-LABEL: testv4i32:
> @@ -122,2167 +122,101 @@ define <4 x i32> @testv4i32(<4 x i32> %i
> ; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> ; AVX2-NEXT: vpbroadcastd {{.*}}(%rip), %xmm1
> ; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
> -; AVX2-NEXT: vpsrld $8, %xmm0, %xmm1
> -; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
> -; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX2-NEXT: vpbroadcastd {{.*}}(%rip), %xmm1
> -; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
> +; AVX2-NEXT: vpsllq $16, %xmm0, %xmm1
> +; AVX2-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX2-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX2-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX2-NEXT: vpsrld $24, %xmm0, %xmm0
> ; AVX2-NEXT: retq
> %out = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> %in)
> ret <4 x i32> %out
> }
>
> define <8 x i16> @testv8i16(<8 x i16> %in) {
> -; SSE2-LABEL: testv8i16:
> -; SSE2: # BB#0:
> -; SSE2-NEXT: pextrw $7, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: pextrw $3, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
> -; SSE2-NEXT: pextrw $5, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: pextrw $1, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
> -; SSE2-NEXT: pextrw $6, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: pextrw $2, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
> -; SSE2-NEXT: pextrw $4, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: movd %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
> -; SSE2-NEXT: retq
> -;
> -; SSE3-LABEL: testv8i16:
> -; SSE3: # BB#0:
> -; SSE3-NEXT: pextrw $7, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: pextrw $3, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
> -; SSE3-NEXT: pextrw $5, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: pextrw $1, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
> -; SSE3-NEXT: pextrw $6, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: pextrw $2, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
> -; SSE3-NEXT: pextrw $4, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: movd %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
> -; SSE3-NEXT: retq
> -;
> -; SSSE3-LABEL: testv8i16:
> -; SSSE3: # BB#0:
> -; SSSE3-NEXT: pextrw $7, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: pextrw $3, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
> -; SSSE3-NEXT: pextrw $5, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: pextrw $1, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
> -; SSSE3-NEXT: pextrw $6, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: pextrw $2, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
> -; SSSE3-NEXT: pextrw $4, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: movd %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
> -; SSSE3-NEXT: retq
> -;
> -; SSE41-LABEL: testv8i16:
> -; SSE41: # BB#0:
> -; SSE41-NEXT: pextrw $1, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: movd %xmm0, %ecx
> -; SSE41-NEXT: movl %ecx, %edx
> -; SSE41-NEXT: shrl %edx
> -; SSE41-NEXT: andl $21845, %edx # imm = 0x5555
> -; SSE41-NEXT: subl %edx, %ecx
> -; SSE41-NEXT: movl %ecx, %edx
> -; SSE41-NEXT: andl $13107, %edx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: addl %edx, %ecx
> -; SSE41-NEXT: movl %ecx, %edx
> -; SSE41-NEXT: andl $65520, %edx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %edx
> -; SSE41-NEXT: addl %ecx, %edx
> -; SSE41-NEXT: andl $3855, %edx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %edx, %ecx # imm = 0x101
> -; SSE41-NEXT: movzbl %ch, %ecx # NOREX
> -; SSE41-NEXT: movd %ecx, %xmm1
> -; SSE41-NEXT: pinsrw $1, %eax, %xmm1
> -; SSE41-NEXT: pextrw $2, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $2, %eax, %xmm1
> -; SSE41-NEXT: pextrw $3, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $3, %eax, %xmm1
> -; SSE41-NEXT: pextrw $4, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $4, %eax, %xmm1
> -; SSE41-NEXT: pextrw $5, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $5, %eax, %xmm1
> -; SSE41-NEXT: pextrw $6, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $6, %eax, %xmm1
> -; SSE41-NEXT: pextrw $7, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $7, %eax, %xmm1
> -; SSE41-NEXT: movdqa %xmm1, %xmm0
> -; SSE41-NEXT: retq
> +; SSE-LABEL: testv8i16:
> +; SSE: # BB#0:
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $1, %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: psubw %xmm1, %xmm0
> +; SSE-NEXT: movdqa {{.*#+}} xmm1 = [13107,13107,13107,13107,13107,13107,13107,13107]
> +; SSE-NEXT: movdqa %xmm0, %xmm2
> +; SSE-NEXT: pand %xmm1, %xmm2
> +; SSE-NEXT: psrlw $2, %xmm0
> +; SSE-NEXT: pand %xmm1, %xmm0
> +; SSE-NEXT: paddw %xmm2, %xmm0
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $4, %xmm1
> +; SSE-NEXT: paddw %xmm0, %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: movdqa %xmm1, %xmm0
> +; SSE-NEXT: psllq $8, %xmm0
> +; SSE-NEXT: paddb %xmm1, %xmm0
> +; SSE-NEXT: psrlw $8, %xmm0
> +; SSE-NEXT: retq
> ;
> ; AVX-LABEL: testv8i16:
> ; AVX: # BB#0:
> -; AVX-NEXT: vpextrw $1, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vmovd %xmm0, %ecx
> -; AVX-NEXT: movl %ecx, %edx
> -; AVX-NEXT: shrl %edx
> -; AVX-NEXT: andl $21845, %edx # imm = 0x5555
> -; AVX-NEXT: subl %edx, %ecx
> -; AVX-NEXT: movl %ecx, %edx
> -; AVX-NEXT: andl $13107, %edx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: addl %edx, %ecx
> -; AVX-NEXT: movl %ecx, %edx
> -; AVX-NEXT: andl $65520, %edx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %edx
> -; AVX-NEXT: addl %ecx, %edx
> -; AVX-NEXT: andl $3855, %edx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %edx, %ecx # imm = 0x101
> -; AVX-NEXT: movzbl %ch, %ecx # NOREX
> -; AVX-NEXT: vmovd %ecx, %xmm1
> -; AVX-NEXT: vpinsrw $1, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $2, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $2, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $3, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $3, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $4, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $4, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $5, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $5, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $6, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $6, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $7, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $7, %eax, %xmm1, %xmm0
> +; AVX-NEXT: vpsrlw $1, %xmm0, %xmm1
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm1, %xmm1
> +; AVX-NEXT: vpsubw %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [13107,13107,13107,13107,13107,13107,13107,13107]
> +; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
> +; AVX-NEXT: vpsrlw $2, %xmm0, %xmm0
> +; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpaddw %xmm0, %xmm2, %xmm0
> +; AVX-NEXT: vpsrlw $4, %xmm0, %xmm1
> +; AVX-NEXT: vpaddw %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0
> ; AVX-NEXT: retq
> %out = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> %in)
> ret <8 x i16> %out
> }
>
> define <16 x i8> @testv16i8(<16 x i8> %in) {
> -; SSE2-LABEL: testv16i8:
> -; SSE2: # BB#0:
> -; SSE2-NEXT: pushq %rbp
> -; SSE2-NEXT: .Ltmp0:
> -; SSE2-NEXT: .cfi_def_cfa_offset 16
> -; SSE2-NEXT: pushq %rbx
> -; SSE2-NEXT: .Ltmp1:
> -; SSE2-NEXT: .cfi_def_cfa_offset 24
> -; SSE2-NEXT: .Ltmp2:
> -; SSE2-NEXT: .cfi_offset %rbx, -24
> -; SSE2-NEXT: .Ltmp3:
> -; SSE2-NEXT: .cfi_offset %rbp, -16
> -; SSE2-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r10b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %dil
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r9b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %bpl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %sil
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %bl
> -; SSE2-NEXT: movb %bl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %bl
> -; SSE2-NEXT: movb %bl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %bl
> -; SSE2-NEXT: andb $51, %bl
> -; SSE2-NEXT: addb %al, %bl
> -; SSE2-NEXT: movb %bl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %bl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %dl
> -; SSE2-NEXT: andb $51, %dl
> -; SSE2-NEXT: addb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %dl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r11b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r8b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb %al, %bl
> -; SSE2-NEXT: shrb %bl
> -; SSE2-NEXT: andb $85, %bl
> -; SSE2-NEXT: subb %bl, %al
> -; SSE2-NEXT: movb %al, %bl
> -; SSE2-NEXT: andb $51, %bl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %bl, %al
> -; SSE2-NEXT: movb %al, %bl
> -; SSE2-NEXT: shrb $4, %bl
> -; SSE2-NEXT: addb %al, %bl
> -; SSE2-NEXT: andb $15, %bl
> -; SSE2-NEXT: movzbl %bl, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE2-NEXT: movb %cl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %cl
> -; SSE2-NEXT: movb %cl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: movb %cl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %dl
> -; SSE2-NEXT: andb $51, %dl
> -; SSE2-NEXT: addb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %dl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
> -; SSE2-NEXT: movb %bpl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %bpl
> -; SSE2-NEXT: movb %bpl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %bpl
> -; SSE2-NEXT: andb $51, %bpl
> -; SSE2-NEXT: addb %al, %bpl
> -; SSE2-NEXT: movb %bpl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %bpl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE2-NEXT: movb %cl, %dl
> -; SSE2-NEXT: shrb %dl
> -; SSE2-NEXT: andb $85, %dl
> -; SSE2-NEXT: subb %dl, %cl
> -; SSE2-NEXT: movb %cl, %dl
> -; SSE2-NEXT: andb $51, %dl
> -; SSE2-NEXT: shrb $2, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: addb %dl, %cl
> -; SSE2-NEXT: movb %cl, %dl
> -; SSE2-NEXT: shrb $4, %dl
> -; SSE2-NEXT: addb %cl, %dl
> -; SSE2-NEXT: andb $15, %dl
> -; SSE2-NEXT: movzbl %dl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm1
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE2-NEXT: movb %r10b, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %r10b
> -; SSE2-NEXT: movb %r10b, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %r10b
> -; SSE2-NEXT: andb $51, %r10b
> -; SSE2-NEXT: addb %cl, %r10b
> -; SSE2-NEXT: movb %r10b, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %r10b, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm2
> -; SSE2-NEXT: movb %r11b, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %r11b
> -; SSE2-NEXT: movb %r11b, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %r11b
> -; SSE2-NEXT: andb $51, %r11b
> -; SSE2-NEXT: addb %cl, %r11b
> -; SSE2-NEXT: movb %r11b, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %r11b, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm0
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE2-NEXT: movb %r9b, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %r9b
> -; SSE2-NEXT: movb %r9b, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %r9b
> -; SSE2-NEXT: andb $51, %r9b
> -; SSE2-NEXT: addb %cl, %r9b
> -; SSE2-NEXT: movb %r9b, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %r9b, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm3
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %eax
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
> -; SSE2-NEXT: movb %dil, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %dil
> -; SSE2-NEXT: movb %dil, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %dil
> -; SSE2-NEXT: andb $51, %dil
> -; SSE2-NEXT: addb %al, %dil
> -; SSE2-NEXT: movb %dil, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %dil, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: movb %r8b, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %r8b
> -; SSE2-NEXT: movb %r8b, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %r8b
> -; SSE2-NEXT: andb $51, %r8b
> -; SSE2-NEXT: addb %al, %r8b
> -; SSE2-NEXT: movb %r8b, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %r8b, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
> -; SSE2-NEXT: movb %sil, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %sil
> -; SSE2-NEXT: movb %sil, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %sil
> -; SSE2-NEXT: andb $51, %sil
> -; SSE2-NEXT: addb %al, %sil
> -; SSE2-NEXT: movb %sil, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %sil, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm4
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE2-NEXT: popq %rbx
> -; SSE2-NEXT: popq %rbp
> -; SSE2-NEXT: retq
> -;
> -; SSE3-LABEL: testv16i8:
> -; SSE3: # BB#0:
> -; SSE3-NEXT: pushq %rbp
> -; SSE3-NEXT: .Ltmp0:
> -; SSE3-NEXT: .cfi_def_cfa_offset 16
> -; SSE3-NEXT: pushq %rbx
> -; SSE3-NEXT: .Ltmp1:
> -; SSE3-NEXT: .cfi_def_cfa_offset 24
> -; SSE3-NEXT: .Ltmp2:
> -; SSE3-NEXT: .cfi_offset %rbx, -24
> -; SSE3-NEXT: .Ltmp3:
> -; SSE3-NEXT: .cfi_offset %rbp, -16
> -; SSE3-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r10b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dil
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r9b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bpl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %sil
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bl
> -; SSE3-NEXT: movb %bl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %bl
> -; SSE3-NEXT: movb %bl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %bl
> -; SSE3-NEXT: andb $51, %bl
> -; SSE3-NEXT: addb %al, %bl
> -; SSE3-NEXT: movb %bl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %bl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %dl
> -; SSE3-NEXT: andb $51, %dl
> -; SSE3-NEXT: addb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %dl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r11b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r8b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb %al, %bl
> -; SSE3-NEXT: shrb %bl
> -; SSE3-NEXT: andb $85, %bl
> -; SSE3-NEXT: subb %bl, %al
> -; SSE3-NEXT: movb %al, %bl
> -; SSE3-NEXT: andb $51, %bl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %bl, %al
> -; SSE3-NEXT: movb %al, %bl
> -; SSE3-NEXT: shrb $4, %bl
> -; SSE3-NEXT: addb %al, %bl
> -; SSE3-NEXT: andb $15, %bl
> -; SSE3-NEXT: movzbl %bl, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE3-NEXT: movb %cl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %cl
> -; SSE3-NEXT: movb %cl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: movb %cl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %dl
> -; SSE3-NEXT: andb $51, %dl
> -; SSE3-NEXT: addb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %dl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
> -; SSE3-NEXT: movb %bpl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %bpl
> -; SSE3-NEXT: movb %bpl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %bpl
> -; SSE3-NEXT: andb $51, %bpl
> -; SSE3-NEXT: addb %al, %bpl
> -; SSE3-NEXT: movb %bpl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %bpl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE3-NEXT: movb %cl, %dl
> -; SSE3-NEXT: shrb %dl
> -; SSE3-NEXT: andb $85, %dl
> -; SSE3-NEXT: subb %dl, %cl
> -; SSE3-NEXT: movb %cl, %dl
> -; SSE3-NEXT: andb $51, %dl
> -; SSE3-NEXT: shrb $2, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: addb %dl, %cl
> -; SSE3-NEXT: movb %cl, %dl
> -; SSE3-NEXT: shrb $4, %dl
> -; SSE3-NEXT: addb %cl, %dl
> -; SSE3-NEXT: andb $15, %dl
> -; SSE3-NEXT: movzbl %dl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm1
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE3-NEXT: movb %r10b, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %r10b
> -; SSE3-NEXT: movb %r10b, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %r10b
> -; SSE3-NEXT: andb $51, %r10b
> -; SSE3-NEXT: addb %cl, %r10b
> -; SSE3-NEXT: movb %r10b, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %r10b, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm2
> -; SSE3-NEXT: movb %r11b, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %r11b
> -; SSE3-NEXT: movb %r11b, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %r11b
> -; SSE3-NEXT: andb $51, %r11b
> -; SSE3-NEXT: addb %cl, %r11b
> -; SSE3-NEXT: movb %r11b, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %r11b, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm0
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE3-NEXT: movb %r9b, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %r9b
> -; SSE3-NEXT: movb %r9b, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %r9b
> -; SSE3-NEXT: andb $51, %r9b
> -; SSE3-NEXT: addb %cl, %r9b
> -; SSE3-NEXT: movb %r9b, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %r9b, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm3
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %eax
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
> -; SSE3-NEXT: movb %dil, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %dil
> -; SSE3-NEXT: movb %dil, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %dil
> -; SSE3-NEXT: andb $51, %dil
> -; SSE3-NEXT: addb %al, %dil
> -; SSE3-NEXT: movb %dil, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %dil, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: movb %r8b, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %r8b
> -; SSE3-NEXT: movb %r8b, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %r8b
> -; SSE3-NEXT: andb $51, %r8b
> -; SSE3-NEXT: addb %al, %r8b
> -; SSE3-NEXT: movb %r8b, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %r8b, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
> -; SSE3-NEXT: movb %sil, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %sil
> -; SSE3-NEXT: movb %sil, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %sil
> -; SSE3-NEXT: andb $51, %sil
> -; SSE3-NEXT: addb %al, %sil
> -; SSE3-NEXT: movb %sil, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %sil, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm4
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE3-NEXT: popq %rbx
> -; SSE3-NEXT: popq %rbp
> -; SSE3-NEXT: retq
> -;
> -; SSSE3-LABEL: testv16i8:
> -; SSSE3: # BB#0:
> -; SSSE3-NEXT: pushq %rbp
> -; SSSE3-NEXT: .Ltmp0:
> -; SSSE3-NEXT: .cfi_def_cfa_offset 16
> -; SSSE3-NEXT: pushq %rbx
> -; SSSE3-NEXT: .Ltmp1:
> -; SSSE3-NEXT: .cfi_def_cfa_offset 24
> -; SSSE3-NEXT: .Ltmp2:
> -; SSSE3-NEXT: .cfi_offset %rbx, -24
> -; SSSE3-NEXT: .Ltmp3:
> -; SSSE3-NEXT: .cfi_offset %rbp, -16
> -; SSSE3-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r10b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dil
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r9b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bpl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %sil
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bl
> -; SSSE3-NEXT: movb %bl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %bl
> -; SSSE3-NEXT: movb %bl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %bl
> -; SSSE3-NEXT: andb $51, %bl
> -; SSSE3-NEXT: addb %al, %bl
> -; SSSE3-NEXT: movb %bl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %bl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %dl
> -; SSSE3-NEXT: andb $51, %dl
> -; SSSE3-NEXT: addb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %dl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r11b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r8b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb %al, %bl
> -; SSSE3-NEXT: shrb %bl
> -; SSSE3-NEXT: andb $85, %bl
> -; SSSE3-NEXT: subb %bl, %al
> -; SSSE3-NEXT: movb %al, %bl
> -; SSSE3-NEXT: andb $51, %bl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %bl, %al
> -; SSSE3-NEXT: movb %al, %bl
> -; SSSE3-NEXT: shrb $4, %bl
> -; SSSE3-NEXT: addb %al, %bl
> -; SSSE3-NEXT: andb $15, %bl
> -; SSSE3-NEXT: movzbl %bl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSSE3-NEXT: movb %cl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %cl
> -; SSSE3-NEXT: movb %cl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: movb %cl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %dl
> -; SSSE3-NEXT: andb $51, %dl
> -; SSSE3-NEXT: addb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %dl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
> -; SSSE3-NEXT: movb %bpl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %bpl
> -; SSSE3-NEXT: movb %bpl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %bpl
> -; SSSE3-NEXT: andb $51, %bpl
> -; SSSE3-NEXT: addb %al, %bpl
> -; SSSE3-NEXT: movb %bpl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %bpl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSSE3-NEXT: movb %cl, %dl
> -; SSSE3-NEXT: shrb %dl
> -; SSSE3-NEXT: andb $85, %dl
> -; SSSE3-NEXT: subb %dl, %cl
> -; SSSE3-NEXT: movb %cl, %dl
> -; SSSE3-NEXT: andb $51, %dl
> -; SSSE3-NEXT: shrb $2, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: addb %dl, %cl
> -; SSSE3-NEXT: movb %cl, %dl
> -; SSSE3-NEXT: shrb $4, %dl
> -; SSSE3-NEXT: addb %cl, %dl
> -; SSSE3-NEXT: andb $15, %dl
> -; SSSE3-NEXT: movzbl %dl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm1
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSSE3-NEXT: movb %r10b, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %r10b
> -; SSSE3-NEXT: movb %r10b, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %r10b
> -; SSSE3-NEXT: andb $51, %r10b
> -; SSSE3-NEXT: addb %cl, %r10b
> -; SSSE3-NEXT: movb %r10b, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %r10b, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm2
> -; SSSE3-NEXT: movb %r11b, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %r11b
> -; SSSE3-NEXT: movb %r11b, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %r11b
> -; SSSE3-NEXT: andb $51, %r11b
> -; SSSE3-NEXT: addb %cl, %r11b
> -; SSSE3-NEXT: movb %r11b, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %r11b, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm0
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSSE3-NEXT: movb %r9b, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %r9b
> -; SSSE3-NEXT: movb %r9b, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %r9b
> -; SSSE3-NEXT: andb $51, %r9b
> -; SSSE3-NEXT: addb %cl, %r9b
> -; SSSE3-NEXT: movb %r9b, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %r9b, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm3
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
> -; SSSE3-NEXT: movb %dil, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %dil
> -; SSSE3-NEXT: movb %dil, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %dil
> -; SSSE3-NEXT: andb $51, %dil
> -; SSSE3-NEXT: addb %al, %dil
> -; SSSE3-NEXT: movb %dil, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %dil, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: movb %r8b, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %r8b
> -; SSSE3-NEXT: movb %r8b, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %r8b
> -; SSSE3-NEXT: andb $51, %r8b
> -; SSSE3-NEXT: addb %al, %r8b
> -; SSSE3-NEXT: movb %r8b, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %r8b, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
> -; SSSE3-NEXT: movb %sil, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %sil
> -; SSSE3-NEXT: movb %sil, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %sil
> -; SSSE3-NEXT: andb $51, %sil
> -; SSSE3-NEXT: addb %al, %sil
> -; SSSE3-NEXT: movb %sil, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %sil, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm4
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSSE3-NEXT: popq %rbx
> -; SSSE3-NEXT: popq %rbp
> -; SSSE3-NEXT: retq
> -;
> -; SSE41-LABEL: testv16i8:
> -; SSE41: # BB#0:
> -; SSE41-NEXT: pextrb $1, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pextrb $0, %xmm0, %ecx
> -; SSE41-NEXT: movb %cl, %dl
> -; SSE41-NEXT: shrb %dl
> -; SSE41-NEXT: andb $85, %dl
> -; SSE41-NEXT: subb %dl, %cl
> -; SSE41-NEXT: movb %cl, %dl
> -; SSE41-NEXT: andb $51, %dl
> -; SSE41-NEXT: shrb $2, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: addb %dl, %cl
> -; SSE41-NEXT: movb %cl, %dl
> -; SSE41-NEXT: shrb $4, %dl
> -; SSE41-NEXT: addb %cl, %dl
> -; SSE41-NEXT: andb $15, %dl
> -; SSE41-NEXT: movzbl %dl, %ecx
> -; SSE41-NEXT: movd %ecx, %xmm1
> -; SSE41-NEXT: pinsrb $1, %eax, %xmm1
> -; SSE41-NEXT: pextrb $2, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $2, %eax, %xmm1
> -; SSE41-NEXT: pextrb $3, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $3, %eax, %xmm1
> -; SSE41-NEXT: pextrb $4, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $4, %eax, %xmm1
> -; SSE41-NEXT: pextrb $5, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $5, %eax, %xmm1
> -; SSE41-NEXT: pextrb $6, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $6, %eax, %xmm1
> -; SSE41-NEXT: pextrb $7, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $7, %eax, %xmm1
> -; SSE41-NEXT: pextrb $8, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $8, %eax, %xmm1
> -; SSE41-NEXT: pextrb $9, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $9, %eax, %xmm1
> -; SSE41-NEXT: pextrb $10, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $10, %eax, %xmm1
> -; SSE41-NEXT: pextrb $11, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $11, %eax, %xmm1
> -; SSE41-NEXT: pextrb $12, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $12, %eax, %xmm1
> -; SSE41-NEXT: pextrb $13, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $13, %eax, %xmm1
> -; SSE41-NEXT: pextrb $14, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $14, %eax, %xmm1
> -; SSE41-NEXT: pextrb $15, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $15, %eax, %xmm1
> -; SSE41-NEXT: movdqa %xmm1, %xmm0
> -; SSE41-NEXT: retq
> +; SSE-LABEL: testv16i8:
> +; SSE: # BB#0:
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $1, %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: psubb %xmm1, %xmm0
> +; SSE-NEXT: movdqa {{.*#+}} xmm1 = [51,51,51,51,51,51,51,51,51,51,51,51,51,51,51,51]
> +; SSE-NEXT: movdqa %xmm0, %xmm2
> +; SSE-NEXT: pand %xmm1, %xmm2
> +; SSE-NEXT: psrlw $2, %xmm0
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm0
> +; SSE-NEXT: pand %xmm1, %xmm0
> +; SSE-NEXT: paddb %xmm2, %xmm0
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $4, %xmm1
> +; SSE-NEXT: movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
> +; SSE-NEXT: pand %xmm2, %xmm1
> +; SSE-NEXT: paddb %xmm0, %xmm1
> +; SSE-NEXT: pand %xmm2, %xmm1
> +; SSE-NEXT: movdqa %xmm1, %xmm0
> +; SSE-NEXT: retq
> ;
> ; AVX-LABEL: testv16i8:
> ; AVX: # BB#0:
> -; AVX-NEXT: vpextrb $1, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpextrb $0, %xmm0, %ecx
> -; AVX-NEXT: movb %cl, %dl
> -; AVX-NEXT: shrb %dl
> -; AVX-NEXT: andb $85, %dl
> -; AVX-NEXT: subb %dl, %cl
> -; AVX-NEXT: movb %cl, %dl
> -; AVX-NEXT: andb $51, %dl
> -; AVX-NEXT: shrb $2, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: addb %dl, %cl
> -; AVX-NEXT: movb %cl, %dl
> -; AVX-NEXT: shrb $4, %dl
> -; AVX-NEXT: addb %cl, %dl
> -; AVX-NEXT: andb $15, %dl
> -; AVX-NEXT: movzbl %dl, %ecx
> -; AVX-NEXT: vmovd %ecx, %xmm1
> -; AVX-NEXT: vpinsrb $1, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $2, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $2, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $3, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $3, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $4, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $4, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $5, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $5, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $6, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $6, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $7, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $7, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $8, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $8, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $9, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $9, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $10, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $10, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $11, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $11, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $12, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $12, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $13, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $13, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $14, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $14, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $15, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $15, %eax, %xmm1, %xmm0
> +; AVX-NEXT: vpsrlw $1, %xmm0, %xmm1
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm1, %xmm1
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm1, %xmm1
> +; AVX-NEXT: vpsubb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [51,51,51,51,51,51,51,51,51,51,51,51,51,51,51,51]
> +; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
> +; AVX-NEXT: vpsrlw $2, %xmm0, %xmm0
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpaddb %xmm0, %xmm2, %xmm0
> +; AVX-NEXT: vpsrlw $4, %xmm0, %xmm1
> +; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
> +; AVX-NEXT: vpand %xmm2, %xmm1, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpand %xmm2, %xmm0, %xmm0
> ; AVX-NEXT: retq
> %out = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> %in)
> ret <16 x i8> %out
>
> Modified: llvm/trunk/test/CodeGen/X86/vector-popcnt-256.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-popcnt-256.ll?rev=238635&r1=238634&r2=238635&view=diff
> ==============================================================================
> --- llvm/trunk/test/CodeGen/X86/vector-popcnt-256.ll (original)
> +++ llvm/trunk/test/CodeGen/X86/vector-popcnt-256.ll Fri May 29 22:20:55 2015
> @@ -99,14 +99,13 @@ define <4 x i64> @testv4i64(<4 x i64> %i
> ; AVX2-NEXT: vpaddq %ymm1, %ymm0, %ymm0
> ; AVX2-NEXT: vpbroadcastq {{.*}}(%rip), %ymm1
> ; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpsrlq $8, %ymm0, %ymm1
> -; AVX2-NEXT: vpaddq %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpsrlq $16, %ymm0, %ymm1
> -; AVX2-NEXT: vpaddq %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpsrlq $32, %ymm0, %ymm1
> -; AVX2-NEXT: vpaddq %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpbroadcastq {{.*}}(%rip), %ymm1
> -; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsllq $32, %ymm0, %ymm1
> +; AVX2-NEXT: vpaddb %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsllq $16, %ymm0, %ymm1
> +; AVX2-NEXT: vpaddb %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsllq $8, %ymm0, %ymm1
> +; AVX2-NEXT: vpaddb %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsrlq $56, %ymm0, %ymm0
> ; AVX2-NEXT: retq
> %out = call <4 x i64> @llvm.ctpop.v4i64(<4 x i64> %in)
> ret <4 x i64> %out
> @@ -270,12 +269,11 @@ define <8 x i32> @testv8i32(<8 x i32> %i
> ; AVX2-NEXT: vpaddd %ymm1, %ymm0, %ymm0
> ; AVX2-NEXT: vpbroadcastd {{.*}}(%rip), %ymm1
> ; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpsrld $8, %ymm0, %ymm1
> -; AVX2-NEXT: vpaddd %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpsrld $16, %ymm0, %ymm1
> -; AVX2-NEXT: vpaddd %ymm1, %ymm0, %ymm0
> -; AVX2-NEXT: vpbroadcastd {{.*}}(%rip), %ymm1
> -; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsllq $16, %ymm0, %ymm1
> +; AVX2-NEXT: vpaddb %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsllq $8, %ymm0, %ymm1
> +; AVX2-NEXT: vpaddb %ymm1, %ymm0, %ymm0
> +; AVX2-NEXT: vpsrld $24, %ymm0, %ymm0
> ; AVX2-NEXT: retq
> %out = call <8 x i32> @llvm.ctpop.v8i32(<8 x i32> %in)
> ret <8 x i32> %out
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
--
Bruno Cardoso Lopes
http://www.brunocardoso.cc
On Sat, May 30, 2015 at 12:20 AM, Chandler Carruth <chandlerc at gmail.com> wrote:
> Author: chandlerc
> Date: Fri May 29 22:20:55 2015
> New Revision: 238635
>
> URL: http://llvm.org/viewvc/llvm-project?rev=238635&view=rev
> Log:
> [x86] Restructure the parallel bitmath lowering of popcount into
> a separate routine, generalize it to work for all the integer vector
> sizes, and do general code cleanups.
>
> This dramatically improves lowerings of byte and short element vector
> popcount, but more importantly it will make the introduction of the
> LUT-approach much cleaner.
>
> The biggest cleanup I've done is to just force the legalizer to do the
> bitcasting we need. We run these iteratively now and it makes the code
> much simpler IMO. Other changes were minor, and mostly naming and
> splitting things up in a way that makes it more clear what is going on.
>
> The other significant change is to use a different final horizontal sum
> approach. This is the same number of instructions as the old method, but
> shifts left instead of right so that we can clear everything but the
> final sum with a single shift right. This seems likely better than
> a mask which will usually have to read the mask from memory. It is
> certaily fewer u-ops. Also, this will be temporary. This and the LUT
> approach share the need of horizontal adds to finish the computation,
> and we have more clever approaches than this one that I'll switch over
> to.
>
> Modified:
> llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
> llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll
> llvm/trunk/test/CodeGen/X86/vector-popcnt-256.ll
>
> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=238635&r1=238634&r2=238635&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Fri May 29 22:20:55 2015
> @@ -846,8 +846,10 @@ X86TargetLowering::X86TargetLowering(con
> // know to perform better than using the popcnt instructions on each vector
> // element. If popcnt isn't supported, always provide the custom version.
> if (!Subtarget->hasPOPCNT()) {
> - setOperationAction(ISD::CTPOP, MVT::v4i32, Custom);
> setOperationAction(ISD::CTPOP, MVT::v2i64, Custom);
> + setOperationAction(ISD::CTPOP, MVT::v4i32, Custom);
> + setOperationAction(ISD::CTPOP, MVT::v8i16, Custom);
> + setOperationAction(ISD::CTPOP, MVT::v16i8, Custom);
> }
>
> // Custom lower build_vector, vector_shuffle, and extract_vector_elt.
> @@ -17327,141 +17329,131 @@ static SDValue LowerBITCAST(SDValue Op,
> return SDValue();
> }
>
> -static SDValue LowerCTPOP(SDValue Op, const X86Subtarget *Subtarget,
> - SelectionDAG &DAG) {
> - SDNode *Node = Op.getNode();
> - SDLoc dl(Node);
> -
> - Op = Op.getOperand(0);
> - EVT VT = Op.getValueType();
> +static SDValue LowerVectorCTPOPBitmath(SDValue Op, SDLoc DL,
> + const X86Subtarget *Subtarget,
> + SelectionDAG &DAG) {
> + MVT VT = Op.getSimpleValueType();
> assert((VT.is128BitVector() || VT.is256BitVector()) &&
> "CTPOP lowering only implemented for 128/256-bit wide vector types");
>
> - unsigned NumElts = VT.getVectorNumElements();
> - EVT EltVT = VT.getVectorElementType();
> - unsigned Len = EltVT.getSizeInBits();
> + int VecSize = VT.getSizeInBits();
> + int NumElts = VT.getVectorNumElements();
> + MVT EltVT = VT.getVectorElementType();
> + int Len = EltVT.getSizeInBits();
>
> // This is the vectorized version of the "best" algorithm from
> // http://graphics.stanford.edu/~seander/bithacks.html#CountBitsSetParallel
> // with a minor tweak to use a series of adds + shifts instead of vector
> - // multiplications. Implemented for the v2i64, v4i64, v4i32, v8i32 types:
> - //
> - // v2i64, v4i64, v4i32 => Only profitable w/ popcnt disabled
> - // v8i32 => Always profitable
> - //
> - // FIXME: There a couple of possible improvements:
> - //
> - // 1) Support for i8 and i16 vectors (needs measurements if popcnt enabled).
> - // 2) Use strategies from http://wm.ite.pl/articles/sse-popcount.html
> + // multiplications. Implemented for all integer vector types.
> //
> - assert(EltVT.isInteger() && (Len == 32 || Len == 64) && Len % 8 == 0 &&
> - "CTPOP not implemented for this vector element type.");
> + // FIXME: Use strategies from http://wm.ite.pl/articles/sse-popcount.html
>
> - // X86 canonicalize ANDs to vXi64, generate the appropriate bitcasts to avoid
> - // extra legalization.
> - bool NeedsBitcast = EltVT == MVT::i32;
> - MVT BitcastVT = VT.is256BitVector() ? MVT::v4i64 : MVT::v2i64;
> -
> - SDValue Cst55 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x55)), dl,
> + SDValue Cst55 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x55)), DL,
> EltVT);
> - SDValue Cst33 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x33)), dl,
> + SDValue Cst33 = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x33)), DL,
> EltVT);
> - SDValue Cst0F = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x0F)), dl,
> + SDValue Cst0F = DAG.getConstant(APInt::getSplat(Len, APInt(8, 0x0F)), DL,
> EltVT);
>
> + SDValue V = Op;
> +
> // v = v - ((v >> 1) & 0x55555555...)
> - SmallVector<SDValue, 8> Ones(NumElts, DAG.getConstant(1, dl, EltVT));
> - SDValue OnesV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Ones);
> - SDValue Srl = DAG.getNode(ISD::SRL, dl, VT, Op, OnesV);
> - if (NeedsBitcast)
> - Srl = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Srl);
> + SmallVector<SDValue, 8> Ones(NumElts, DAG.getConstant(1, DL, EltVT));
> + SDValue OnesV = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Ones);
> + SDValue Srl = DAG.getNode(ISD::SRL, DL, VT, V, OnesV);
>
> SmallVector<SDValue, 8> Mask55(NumElts, Cst55);
> - SDValue M55 = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Mask55);
> - if (NeedsBitcast)
> - M55 = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M55);
> -
> - SDValue And = DAG.getNode(ISD::AND, dl, Srl.getValueType(), Srl, M55);
> - if (VT != And.getValueType())
> - And = DAG.getNode(ISD::BITCAST, dl, VT, And);
> - SDValue Sub = DAG.getNode(ISD::SUB, dl, VT, Op, And);
> + SDValue M55 = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Mask55);
> + SDValue And = DAG.getNode(ISD::AND, DL, Srl.getValueType(), Srl, M55);
> +
> + V = DAG.getNode(ISD::SUB, DL, VT, V, And);
>
> // v = (v & 0x33333333...) + ((v >> 2) & 0x33333333...)
> SmallVector<SDValue, 8> Mask33(NumElts, Cst33);
> - SDValue M33 = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Mask33);
> - SmallVector<SDValue, 8> Twos(NumElts, DAG.getConstant(2, dl, EltVT));
> - SDValue TwosV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Twos);
> -
> - Srl = DAG.getNode(ISD::SRL, dl, VT, Sub, TwosV);
> - if (NeedsBitcast) {
> - Srl = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Srl);
> - M33 = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M33);
> - Sub = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Sub);
> - }
> -
> - SDValue AndRHS = DAG.getNode(ISD::AND, dl, M33.getValueType(), Srl, M33);
> - SDValue AndLHS = DAG.getNode(ISD::AND, dl, M33.getValueType(), Sub, M33);
> - if (VT != AndRHS.getValueType()) {
> - AndRHS = DAG.getNode(ISD::BITCAST, dl, VT, AndRHS);
> - AndLHS = DAG.getNode(ISD::BITCAST, dl, VT, AndLHS);
> - }
> - SDValue Add = DAG.getNode(ISD::ADD, dl, VT, AndLHS, AndRHS);
> + SDValue M33 = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Mask33);
> + SDValue AndLHS = DAG.getNode(ISD::AND, DL, M33.getValueType(), V, M33);
> +
> + SmallVector<SDValue, 8> Twos(NumElts, DAG.getConstant(2, DL, EltVT));
> + SDValue TwosV = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Twos);
> + Srl = DAG.getNode(ISD::SRL, DL, VT, V, TwosV);
> + SDValue AndRHS = DAG.getNode(ISD::AND, DL, M33.getValueType(), Srl, M33);
> +
> + V = DAG.getNode(ISD::ADD, DL, VT, AndLHS, AndRHS);
>
> // v = (v + (v >> 4)) & 0x0F0F0F0F...
> - SmallVector<SDValue, 8> Fours(NumElts, DAG.getConstant(4, dl, EltVT));
> - SDValue FoursV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Fours);
> - Srl = DAG.getNode(ISD::SRL, dl, VT, Add, FoursV);
> - Add = DAG.getNode(ISD::ADD, dl, VT, Add, Srl);
> + SmallVector<SDValue, 8> Fours(NumElts, DAG.getConstant(4, DL, EltVT));
> + SDValue FoursV = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Fours);
> + Srl = DAG.getNode(ISD::SRL, DL, VT, V, FoursV);
> + SDValue Add = DAG.getNode(ISD::ADD, DL, VT, V, Srl);
>
> SmallVector<SDValue, 8> Mask0F(NumElts, Cst0F);
> - SDValue M0F = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Mask0F);
> - if (NeedsBitcast) {
> - Add = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Add);
> - M0F = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M0F);
> - }
> - And = DAG.getNode(ISD::AND, dl, M0F.getValueType(), Add, M0F);
> - if (VT != And.getValueType())
> - And = DAG.getNode(ISD::BITCAST, dl, VT, And);
> + SDValue M0F = DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Mask0F);
>
> - // The algorithm mentioned above uses:
> - // v = (v * 0x01010101...) >> (Len - 8)
> - //
> - // Change it to use vector adds + vector shifts which yield faster results on
> - // Haswell than using vector integer multiplication.
> - //
> - // For i32 elements:
> - // v = v + (v >> 8)
> - // v = v + (v >> 16)
> - //
> - // For i64 elements:
> - // v = v + (v >> 8)
> - // v = v + (v >> 16)
> - // v = v + (v >> 32)
> + V = DAG.getNode(ISD::AND, DL, M0F.getValueType(), Add, M0F);
> +
> + // At this point, V contains the byte-wise population count, and we are
> + // merely doing a horizontal sum if necessary to get the wider element
> + // counts.
> //
> - Add = And;
> + // FIXME: There is a different lowering strategy above for the horizontal sum
> + // of byte-wise population counts. This one and that one should be merged,
> + // using the fastest of the two for each size.
> + MVT ByteVT = MVT::getVectorVT(MVT::i8, VecSize / 8);
> + MVT ShiftVT = MVT::getVectorVT(MVT::i64, VecSize / 64);
> + V = DAG.getNode(ISD::BITCAST, DL, ByteVT, V);
> SmallVector<SDValue, 8> Csts;
> - for (unsigned i = 8; i <= Len/2; i *= 2) {
> - Csts.assign(NumElts, DAG.getConstant(i, dl, EltVT));
> - SDValue CstsV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Csts);
> - Srl = DAG.getNode(ISD::SRL, dl, VT, Add, CstsV);
> - Add = DAG.getNode(ISD::ADD, dl, VT, Add, Srl);
> - Csts.clear();
> + assert(Len <= 64 && "We don't support element sizes of more than 64 bits!");
> + assert(isPowerOf2_32(Len) && "Only power of two element sizes supported!");
> + for (int i = Len; i > 8; i /= 2) {
> + Csts.assign(VecSize / 64, DAG.getConstant(i / 2, DL, MVT::i64));
> + SDValue Shl = DAG.getNode(
> + ISD::SHL, DL, ShiftVT, DAG.getNode(ISD::BITCAST, DL, ShiftVT, V),
> + DAG.getNode(ISD::BUILD_VECTOR, DL, ShiftVT, Csts));
> + V = DAG.getNode(ISD::ADD, DL, ByteVT, V,
> + DAG.getNode(ISD::BITCAST, DL, ByteVT, Shl));
> + }
> +
> + // The high byte now contains the sum of the element bytes. Shift it right
> + // (if needed) to make it the low byte.
> + V = DAG.getNode(ISD::BITCAST, DL, VT, V);
> + if (Len > 8) {
> + Csts.assign(NumElts, DAG.getConstant(Len - 8, DL, EltVT));
> + V = DAG.getNode(ISD::SRL, DL, VT, V,
> + DAG.getNode(ISD::BUILD_VECTOR, DL, VT, Csts));
> }
> + return V;
> +}
>
> - // The result is on the least significant 6-bits on i32 and 7-bits on i64.
> - SDValue Cst3F = DAG.getConstant(APInt(Len, Len == 32 ? 0x3F : 0x7F), dl,
> - EltVT);
> - SmallVector<SDValue, 8> Cst3FV(NumElts, Cst3F);
> - SDValue M3F = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Cst3FV);
> - if (NeedsBitcast) {
> - Add = DAG.getNode(ISD::BITCAST, dl, BitcastVT, Add);
> - M3F = DAG.getNode(ISD::BITCAST, dl, BitcastVT, M3F);
> - }
> - And = DAG.getNode(ISD::AND, dl, M3F.getValueType(), Add, M3F);
> - if (VT != And.getValueType())
> - And = DAG.getNode(ISD::BITCAST, dl, VT, And);
>
> - return And;
> +static SDValue LowerVectorCTPOP(SDValue Op, const X86Subtarget *Subtarget,
> + SelectionDAG &DAG) {
> + MVT VT = Op.getSimpleValueType();
> + // FIXME: Need to add AVX-512 support here!
> + assert((VT.is256BitVector() || VT.is128BitVector()) &&
> + "Unknown CTPOP type to handle");
> + SDLoc DL(Op.getNode());
> + SDValue Op0 = Op.getOperand(0);
> +
> + if (VT.is256BitVector() && !Subtarget->hasInt256()) {
> + unsigned NumElems = VT.getVectorNumElements();
> +
> + // Extract each 128-bit vector, compute pop count and concat the result.
> + SDValue LHS = Extract128BitVector(Op0, 0, DAG, DL);
> + SDValue RHS = Extract128BitVector(Op0, NumElems/2, DAG, DL);
> +
> + return DAG.getNode(ISD::CONCAT_VECTORS, DL, VT,
> + LowerVectorCTPOPBitmath(LHS, DL, Subtarget, DAG),
> + LowerVectorCTPOPBitmath(RHS, DL, Subtarget, DAG));
> + }
> +
> + return LowerVectorCTPOPBitmath(Op0, DL, Subtarget, DAG);
> +}
> +
> +static SDValue LowerCTPOP(SDValue Op, const X86Subtarget *Subtarget,
> + SelectionDAG &DAG) {
> + assert(Op.getValueType().isVector() &&
> + "We only do custom lowering for vector population count.");
> + return LowerVectorCTPOP(Op, Subtarget, DAG);
> }
>
> static SDValue LowerLOAD_SUB(SDValue Op, SelectionDAG &DAG) {
>
> Modified: llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll?rev=238635&r1=238634&r2=238635&view=diff
> ==============================================================================
> --- llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll (original)
> +++ llvm/trunk/test/CodeGen/X86/vector-popcnt-128.ll Fri May 29 22:20:55 2015
> @@ -25,15 +25,15 @@ define <2 x i64> @testv2i64(<2 x i64> %i
> ; SSE-NEXT: paddq %xmm0, %xmm1
> ; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> ; SSE-NEXT: movdqa %xmm1, %xmm0
> -; SSE-NEXT: psrlq $8, %xmm0
> -; SSE-NEXT: paddq %xmm1, %xmm0
> +; SSE-NEXT: psllq $32, %xmm0
> +; SSE-NEXT: paddb %xmm1, %xmm0
> ; SSE-NEXT: movdqa %xmm0, %xmm1
> -; SSE-NEXT: psrlq $16, %xmm1
> -; SSE-NEXT: paddq %xmm0, %xmm1
> +; SSE-NEXT: psllq $16, %xmm1
> +; SSE-NEXT: paddb %xmm0, %xmm1
> ; SSE-NEXT: movdqa %xmm1, %xmm0
> -; SSE-NEXT: psrlq $32, %xmm0
> -; SSE-NEXT: paddq %xmm1, %xmm0
> -; SSE-NEXT: pand {{.*}}(%rip), %xmm0
> +; SSE-NEXT: psllq $8, %xmm0
> +; SSE-NEXT: paddb %xmm1, %xmm0
> +; SSE-NEXT: psrlq $56, %xmm0
> ; SSE-NEXT: retq
> ;
> ; AVX-LABEL: testv2i64:
> @@ -49,13 +49,13 @@ define <2 x i64> @testv2i64(<2 x i64> %i
> ; AVX-NEXT: vpsrlq $4, %xmm0, %xmm1
> ; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> ; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> -; AVX-NEXT: vpsrlq $8, %xmm0, %xmm1
> -; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> -; AVX-NEXT: vpsrlq $16, %xmm0, %xmm1
> -; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> -; AVX-NEXT: vpsrlq $32, %xmm0, %xmm1
> -; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
> -; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $32, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $16, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsrlq $56, %xmm0, %xmm0
> ; AVX-NEXT: retq
> %out = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %in)
> ret <2 x i64> %out
> @@ -79,12 +79,12 @@ define <4 x i32> @testv4i32(<4 x i32> %i
> ; SSE-NEXT: paddd %xmm0, %xmm1
> ; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> ; SSE-NEXT: movdqa %xmm1, %xmm2
> -; SSE-NEXT: psrld $8, %xmm2
> -; SSE-NEXT: paddd %xmm1, %xmm2
> +; SSE-NEXT: psllq $16, %xmm2
> +; SSE-NEXT: paddb %xmm1, %xmm2
> ; SSE-NEXT: movdqa %xmm2, %xmm0
> -; SSE-NEXT: psrld $16, %xmm0
> -; SSE-NEXT: paddd %xmm2, %xmm0
> -; SSE-NEXT: pand {{.*}}(%rip), %xmm0
> +; SSE-NEXT: psllq $8, %xmm0
> +; SSE-NEXT: paddb %xmm2, %xmm0
> +; SSE-NEXT: psrld $24, %xmm0
> ; SSE-NEXT: retq
> ;
> ; AVX1-LABEL: testv4i32:
> @@ -100,11 +100,11 @@ define <4 x i32> @testv4i32(<4 x i32> %i
> ; AVX1-NEXT: vpsrld $4, %xmm0, %xmm1
> ; AVX1-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> ; AVX1-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> -; AVX1-NEXT: vpsrld $8, %xmm0, %xmm1
> -; AVX1-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX1-NEXT: vpsrld $16, %xmm0, %xmm1
> -; AVX1-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX1-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX1-NEXT: vpsllq $16, %xmm0, %xmm1
> +; AVX1-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX1-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX1-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX1-NEXT: vpsrld $24, %xmm0, %xmm0
> ; AVX1-NEXT: retq
> ;
> ; AVX2-LABEL: testv4i32:
> @@ -122,2167 +122,101 @@ define <4 x i32> @testv4i32(<4 x i32> %i
> ; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> ; AVX2-NEXT: vpbroadcastd {{.*}}(%rip), %xmm1
> ; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
> -; AVX2-NEXT: vpsrld $8, %xmm0, %xmm1
> -; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
> -; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
> -; AVX2-NEXT: vpbroadcastd {{.*}}(%rip), %xmm1
> -; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
> +; AVX2-NEXT: vpsllq $16, %xmm0, %xmm1
> +; AVX2-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX2-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX2-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX2-NEXT: vpsrld $24, %xmm0, %xmm0
> ; AVX2-NEXT: retq
> %out = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> %in)
> ret <4 x i32> %out
> }
>
> define <8 x i16> @testv8i16(<8 x i16> %in) {
> -; SSE2-LABEL: testv8i16:
> -; SSE2: # BB#0:
> -; SSE2-NEXT: pextrw $7, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: pextrw $3, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
> -; SSE2-NEXT: pextrw $5, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: pextrw $1, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
> -; SSE2-NEXT: pextrw $6, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: pextrw $2, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
> -; SSE2-NEXT: pextrw $4, %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: movd %xmm0, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: shrl %ecx
> -; SSE2-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE2-NEXT: subl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE2-NEXT: shrl $2, %eax
> -; SSE2-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE2-NEXT: addl %ecx, %eax
> -; SSE2-NEXT: movl %eax, %ecx
> -; SSE2-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE2-NEXT: shrl $4, %ecx
> -; SSE2-NEXT: addl %eax, %ecx
> -; SSE2-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE2-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE2-NEXT: movzbl %ah, %eax # NOREX
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
> -; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
> -; SSE2-NEXT: retq
> -;
> -; SSE3-LABEL: testv8i16:
> -; SSE3: # BB#0:
> -; SSE3-NEXT: pextrw $7, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: pextrw $3, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
> -; SSE3-NEXT: pextrw $5, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: pextrw $1, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
> -; SSE3-NEXT: pextrw $6, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: pextrw $2, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
> -; SSE3-NEXT: pextrw $4, %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: movd %xmm0, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: shrl %ecx
> -; SSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE3-NEXT: subl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE3-NEXT: shrl $2, %eax
> -; SSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE3-NEXT: addl %ecx, %eax
> -; SSE3-NEXT: movl %eax, %ecx
> -; SSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE3-NEXT: shrl $4, %ecx
> -; SSE3-NEXT: addl %eax, %ecx
> -; SSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
> -; SSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
> -; SSE3-NEXT: retq
> -;
> -; SSSE3-LABEL: testv8i16:
> -; SSSE3: # BB#0:
> -; SSSE3-NEXT: pextrw $7, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: pextrw $3, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
> -; SSSE3-NEXT: pextrw $5, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: pextrw $1, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
> -; SSSE3-NEXT: pextrw $6, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: pextrw $2, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
> -; SSSE3-NEXT: pextrw $4, %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: movd %xmm0, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: shrl %ecx
> -; SSSE3-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSSE3-NEXT: subl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSSE3-NEXT: shrl $2, %eax
> -; SSSE3-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSSE3-NEXT: addl %ecx, %eax
> -; SSSE3-NEXT: movl %eax, %ecx
> -; SSSE3-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSSE3-NEXT: shrl $4, %ecx
> -; SSSE3-NEXT: addl %eax, %ecx
> -; SSSE3-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSSE3-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSSE3-NEXT: movzbl %ah, %eax # NOREX
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
> -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
> -; SSSE3-NEXT: retq
> -;
> -; SSE41-LABEL: testv8i16:
> -; SSE41: # BB#0:
> -; SSE41-NEXT: pextrw $1, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: movd %xmm0, %ecx
> -; SSE41-NEXT: movl %ecx, %edx
> -; SSE41-NEXT: shrl %edx
> -; SSE41-NEXT: andl $21845, %edx # imm = 0x5555
> -; SSE41-NEXT: subl %edx, %ecx
> -; SSE41-NEXT: movl %ecx, %edx
> -; SSE41-NEXT: andl $13107, %edx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: addl %edx, %ecx
> -; SSE41-NEXT: movl %ecx, %edx
> -; SSE41-NEXT: andl $65520, %edx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %edx
> -; SSE41-NEXT: addl %ecx, %edx
> -; SSE41-NEXT: andl $3855, %edx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %edx, %ecx # imm = 0x101
> -; SSE41-NEXT: movzbl %ch, %ecx # NOREX
> -; SSE41-NEXT: movd %ecx, %xmm1
> -; SSE41-NEXT: pinsrw $1, %eax, %xmm1
> -; SSE41-NEXT: pextrw $2, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $2, %eax, %xmm1
> -; SSE41-NEXT: pextrw $3, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $3, %eax, %xmm1
> -; SSE41-NEXT: pextrw $4, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $4, %eax, %xmm1
> -; SSE41-NEXT: pextrw $5, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $5, %eax, %xmm1
> -; SSE41-NEXT: pextrw $6, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $6, %eax, %xmm1
> -; SSE41-NEXT: pextrw $7, %xmm0, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: shrl %ecx
> -; SSE41-NEXT: andl $21845, %ecx # imm = 0x5555
> -; SSE41-NEXT: subl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $13107, %ecx # imm = 0x3333
> -; SSE41-NEXT: shrl $2, %eax
> -; SSE41-NEXT: andl $13107, %eax # imm = 0x3333
> -; SSE41-NEXT: addl %ecx, %eax
> -; SSE41-NEXT: movl %eax, %ecx
> -; SSE41-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; SSE41-NEXT: shrl $4, %ecx
> -; SSE41-NEXT: addl %eax, %ecx
> -; SSE41-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; SSE41-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; SSE41-NEXT: movzbl %ah, %eax # NOREX
> -; SSE41-NEXT: pinsrw $7, %eax, %xmm1
> -; SSE41-NEXT: movdqa %xmm1, %xmm0
> -; SSE41-NEXT: retq
> +; SSE-LABEL: testv8i16:
> +; SSE: # BB#0:
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $1, %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: psubw %xmm1, %xmm0
> +; SSE-NEXT: movdqa {{.*#+}} xmm1 = [13107,13107,13107,13107,13107,13107,13107,13107]
> +; SSE-NEXT: movdqa %xmm0, %xmm2
> +; SSE-NEXT: pand %xmm1, %xmm2
> +; SSE-NEXT: psrlw $2, %xmm0
> +; SSE-NEXT: pand %xmm1, %xmm0
> +; SSE-NEXT: paddw %xmm2, %xmm0
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $4, %xmm1
> +; SSE-NEXT: paddw %xmm0, %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: movdqa %xmm1, %xmm0
> +; SSE-NEXT: psllq $8, %xmm0
> +; SSE-NEXT: paddb %xmm1, %xmm0
> +; SSE-NEXT: psrlw $8, %xmm0
> +; SSE-NEXT: retq
> ;
> ; AVX-LABEL: testv8i16:
> ; AVX: # BB#0:
> -; AVX-NEXT: vpextrw $1, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vmovd %xmm0, %ecx
> -; AVX-NEXT: movl %ecx, %edx
> -; AVX-NEXT: shrl %edx
> -; AVX-NEXT: andl $21845, %edx # imm = 0x5555
> -; AVX-NEXT: subl %edx, %ecx
> -; AVX-NEXT: movl %ecx, %edx
> -; AVX-NEXT: andl $13107, %edx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: addl %edx, %ecx
> -; AVX-NEXT: movl %ecx, %edx
> -; AVX-NEXT: andl $65520, %edx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %edx
> -; AVX-NEXT: addl %ecx, %edx
> -; AVX-NEXT: andl $3855, %edx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %edx, %ecx # imm = 0x101
> -; AVX-NEXT: movzbl %ch, %ecx # NOREX
> -; AVX-NEXT: vmovd %ecx, %xmm1
> -; AVX-NEXT: vpinsrw $1, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $2, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $2, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $3, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $3, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $4, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $4, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $5, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $5, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $6, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $6, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrw $7, %xmm0, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: shrl %ecx
> -; AVX-NEXT: andl $21845, %ecx # imm = 0x5555
> -; AVX-NEXT: subl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $13107, %ecx # imm = 0x3333
> -; AVX-NEXT: shrl $2, %eax
> -; AVX-NEXT: andl $13107, %eax # imm = 0x3333
> -; AVX-NEXT: addl %ecx, %eax
> -; AVX-NEXT: movl %eax, %ecx
> -; AVX-NEXT: andl $65520, %ecx # imm = 0xFFF0
> -; AVX-NEXT: shrl $4, %ecx
> -; AVX-NEXT: addl %eax, %ecx
> -; AVX-NEXT: andl $3855, %ecx # imm = 0xF0F
> -; AVX-NEXT: imull $257, %ecx, %eax # imm = 0x101
> -; AVX-NEXT: movzbl %ah, %eax # NOREX
> -; AVX-NEXT: vpinsrw $7, %eax, %xmm1, %xmm0
> +; AVX-NEXT: vpsrlw $1, %xmm0, %xmm1
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm1, %xmm1
> +; AVX-NEXT: vpsubw %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [13107,13107,13107,13107,13107,13107,13107,13107]
> +; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
> +; AVX-NEXT: vpsrlw $2, %xmm0, %xmm0
> +; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpaddw %xmm0, %xmm2, %xmm0
> +; AVX-NEXT: vpsrlw $4, %xmm0, %xmm1
> +; AVX-NEXT: vpaddw %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
> +; AVX-NEXT: vpsllq $8, %xmm0, %xmm1
> +; AVX-NEXT: vpaddb %xmm1, %xmm0, %xmm0
> +; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0
> ; AVX-NEXT: retq
> %out = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> %in)
> ret <8 x i16> %out
> }
>
> define <16 x i8> @testv16i8(<16 x i8> %in) {
> -; SSE2-LABEL: testv16i8:
> -; SSE2: # BB#0:
> -; SSE2-NEXT: pushq %rbp
> -; SSE2-NEXT: .Ltmp0:
> -; SSE2-NEXT: .cfi_def_cfa_offset 16
> -; SSE2-NEXT: pushq %rbx
> -; SSE2-NEXT: .Ltmp1:
> -; SSE2-NEXT: .cfi_def_cfa_offset 24
> -; SSE2-NEXT: .Ltmp2:
> -; SSE2-NEXT: .cfi_offset %rbx, -24
> -; SSE2-NEXT: .Ltmp3:
> -; SSE2-NEXT: .cfi_offset %rbp, -16
> -; SSE2-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r10b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %dil
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r9b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %bpl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %sil
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %bl
> -; SSE2-NEXT: movb %bl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %bl
> -; SSE2-NEXT: movb %bl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %bl
> -; SSE2-NEXT: andb $51, %bl
> -; SSE2-NEXT: addb %al, %bl
> -; SSE2-NEXT: movb %bl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %bl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %dl
> -; SSE2-NEXT: andb $51, %dl
> -; SSE2-NEXT: addb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %dl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r11b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %r8b
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb %al, %bl
> -; SSE2-NEXT: shrb %bl
> -; SSE2-NEXT: andb $85, %bl
> -; SSE2-NEXT: subb %bl, %al
> -; SSE2-NEXT: movb %al, %bl
> -; SSE2-NEXT: andb $51, %bl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %bl, %al
> -; SSE2-NEXT: movb %al, %bl
> -; SSE2-NEXT: shrb $4, %bl
> -; SSE2-NEXT: addb %al, %bl
> -; SSE2-NEXT: andb $15, %bl
> -; SSE2-NEXT: movzbl %bl, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE2-NEXT: movb %cl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %cl
> -; SSE2-NEXT: movb %cl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: movb %cl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm1
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %dl
> -; SSE2-NEXT: andb $51, %dl
> -; SSE2-NEXT: addb %al, %dl
> -; SSE2-NEXT: movb %dl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %dl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
> -; SSE2-NEXT: movb %bpl, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %bpl
> -; SSE2-NEXT: movb %bpl, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %bpl
> -; SSE2-NEXT: andb $51, %bpl
> -; SSE2-NEXT: addb %al, %bpl
> -; SSE2-NEXT: movb %bpl, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %bpl, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE2-NEXT: movb %cl, %dl
> -; SSE2-NEXT: shrb %dl
> -; SSE2-NEXT: andb $85, %dl
> -; SSE2-NEXT: subb %dl, %cl
> -; SSE2-NEXT: movb %cl, %dl
> -; SSE2-NEXT: andb $51, %dl
> -; SSE2-NEXT: shrb $2, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: addb %dl, %cl
> -; SSE2-NEXT: movb %cl, %dl
> -; SSE2-NEXT: shrb $4, %dl
> -; SSE2-NEXT: addb %cl, %dl
> -; SSE2-NEXT: andb $15, %dl
> -; SSE2-NEXT: movzbl %dl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm1
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE2-NEXT: movb %r10b, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %r10b
> -; SSE2-NEXT: movb %r10b, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %r10b
> -; SSE2-NEXT: andb $51, %r10b
> -; SSE2-NEXT: addb %cl, %r10b
> -; SSE2-NEXT: movb %r10b, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %r10b, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm2
> -; SSE2-NEXT: movb %r11b, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %r11b
> -; SSE2-NEXT: movb %r11b, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %r11b
> -; SSE2-NEXT: andb $51, %r11b
> -; SSE2-NEXT: addb %cl, %r11b
> -; SSE2-NEXT: movb %r11b, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %r11b, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm0
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE2-NEXT: movb %r9b, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %r9b
> -; SSE2-NEXT: movb %r9b, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %r9b
> -; SSE2-NEXT: andb $51, %r9b
> -; SSE2-NEXT: addb %cl, %r9b
> -; SSE2-NEXT: movb %r9b, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %r9b, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %ecx
> -; SSE2-NEXT: movd %ecx, %xmm3
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %eax
> -; SSE2-NEXT: movd %eax, %xmm2
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
> -; SSE2-NEXT: movb %dil, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %dil
> -; SSE2-NEXT: movb %dil, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %dil
> -; SSE2-NEXT: andb $51, %dil
> -; SSE2-NEXT: addb %al, %dil
> -; SSE2-NEXT: movb %dil, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %dil, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: movb %r8b, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %r8b
> -; SSE2-NEXT: movb %r8b, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %r8b
> -; SSE2-NEXT: andb $51, %r8b
> -; SSE2-NEXT: addb %al, %r8b
> -; SSE2-NEXT: movb %r8b, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %r8b, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm3
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
> -; SSE2-NEXT: movb %sil, %al
> -; SSE2-NEXT: shrb %al
> -; SSE2-NEXT: andb $85, %al
> -; SSE2-NEXT: subb %al, %sil
> -; SSE2-NEXT: movb %sil, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: shrb $2, %sil
> -; SSE2-NEXT: andb $51, %sil
> -; SSE2-NEXT: addb %al, %sil
> -; SSE2-NEXT: movb %sil, %al
> -; SSE2-NEXT: shrb $4, %al
> -; SSE2-NEXT: addb %sil, %al
> -; SSE2-NEXT: andb $15, %al
> -; SSE2-NEXT: movzbl %al, %eax
> -; SSE2-NEXT: movd %eax, %xmm4
> -; SSE2-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb %cl
> -; SSE2-NEXT: andb $85, %cl
> -; SSE2-NEXT: subb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: andb $51, %cl
> -; SSE2-NEXT: shrb $2, %al
> -; SSE2-NEXT: andb $51, %al
> -; SSE2-NEXT: addb %cl, %al
> -; SSE2-NEXT: movb %al, %cl
> -; SSE2-NEXT: shrb $4, %cl
> -; SSE2-NEXT: addb %al, %cl
> -; SSE2-NEXT: andb $15, %cl
> -; SSE2-NEXT: movzbl %cl, %eax
> -; SSE2-NEXT: movd %eax, %xmm0
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE2-NEXT: popq %rbx
> -; SSE2-NEXT: popq %rbp
> -; SSE2-NEXT: retq
> -;
> -; SSE3-LABEL: testv16i8:
> -; SSE3: # BB#0:
> -; SSE3-NEXT: pushq %rbp
> -; SSE3-NEXT: .Ltmp0:
> -; SSE3-NEXT: .cfi_def_cfa_offset 16
> -; SSE3-NEXT: pushq %rbx
> -; SSE3-NEXT: .Ltmp1:
> -; SSE3-NEXT: .cfi_def_cfa_offset 24
> -; SSE3-NEXT: .Ltmp2:
> -; SSE3-NEXT: .cfi_offset %rbx, -24
> -; SSE3-NEXT: .Ltmp3:
> -; SSE3-NEXT: .cfi_offset %rbp, -16
> -; SSE3-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r10b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dil
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r9b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bpl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %sil
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bl
> -; SSE3-NEXT: movb %bl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %bl
> -; SSE3-NEXT: movb %bl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %bl
> -; SSE3-NEXT: andb $51, %bl
> -; SSE3-NEXT: addb %al, %bl
> -; SSE3-NEXT: movb %bl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %bl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %dl
> -; SSE3-NEXT: andb $51, %dl
> -; SSE3-NEXT: addb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %dl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r11b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r8b
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb %al, %bl
> -; SSE3-NEXT: shrb %bl
> -; SSE3-NEXT: andb $85, %bl
> -; SSE3-NEXT: subb %bl, %al
> -; SSE3-NEXT: movb %al, %bl
> -; SSE3-NEXT: andb $51, %bl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %bl, %al
> -; SSE3-NEXT: movb %al, %bl
> -; SSE3-NEXT: shrb $4, %bl
> -; SSE3-NEXT: addb %al, %bl
> -; SSE3-NEXT: andb $15, %bl
> -; SSE3-NEXT: movzbl %bl, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE3-NEXT: movb %cl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %cl
> -; SSE3-NEXT: movb %cl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: movb %cl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm1
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %dl
> -; SSE3-NEXT: andb $51, %dl
> -; SSE3-NEXT: addb %al, %dl
> -; SSE3-NEXT: movb %dl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %dl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
> -; SSE3-NEXT: movb %bpl, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %bpl
> -; SSE3-NEXT: movb %bpl, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %bpl
> -; SSE3-NEXT: andb $51, %bpl
> -; SSE3-NEXT: addb %al, %bpl
> -; SSE3-NEXT: movb %bpl, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %bpl, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSE3-NEXT: movb %cl, %dl
> -; SSE3-NEXT: shrb %dl
> -; SSE3-NEXT: andb $85, %dl
> -; SSE3-NEXT: subb %dl, %cl
> -; SSE3-NEXT: movb %cl, %dl
> -; SSE3-NEXT: andb $51, %dl
> -; SSE3-NEXT: shrb $2, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: addb %dl, %cl
> -; SSE3-NEXT: movb %cl, %dl
> -; SSE3-NEXT: shrb $4, %dl
> -; SSE3-NEXT: addb %cl, %dl
> -; SSE3-NEXT: andb $15, %dl
> -; SSE3-NEXT: movzbl %dl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm1
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSE3-NEXT: movb %r10b, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %r10b
> -; SSE3-NEXT: movb %r10b, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %r10b
> -; SSE3-NEXT: andb $51, %r10b
> -; SSE3-NEXT: addb %cl, %r10b
> -; SSE3-NEXT: movb %r10b, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %r10b, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm2
> -; SSE3-NEXT: movb %r11b, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %r11b
> -; SSE3-NEXT: movb %r11b, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %r11b
> -; SSE3-NEXT: andb $51, %r11b
> -; SSE3-NEXT: addb %cl, %r11b
> -; SSE3-NEXT: movb %r11b, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %r11b, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm0
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE3-NEXT: movb %r9b, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %r9b
> -; SSE3-NEXT: movb %r9b, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %r9b
> -; SSE3-NEXT: andb $51, %r9b
> -; SSE3-NEXT: addb %cl, %r9b
> -; SSE3-NEXT: movb %r9b, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %r9b, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %ecx
> -; SSE3-NEXT: movd %ecx, %xmm3
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %eax
> -; SSE3-NEXT: movd %eax, %xmm2
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
> -; SSE3-NEXT: movb %dil, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %dil
> -; SSE3-NEXT: movb %dil, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %dil
> -; SSE3-NEXT: andb $51, %dil
> -; SSE3-NEXT: addb %al, %dil
> -; SSE3-NEXT: movb %dil, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %dil, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: movb %r8b, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %r8b
> -; SSE3-NEXT: movb %r8b, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %r8b
> -; SSE3-NEXT: andb $51, %r8b
> -; SSE3-NEXT: addb %al, %r8b
> -; SSE3-NEXT: movb %r8b, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %r8b, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm3
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
> -; SSE3-NEXT: movb %sil, %al
> -; SSE3-NEXT: shrb %al
> -; SSE3-NEXT: andb $85, %al
> -; SSE3-NEXT: subb %al, %sil
> -; SSE3-NEXT: movb %sil, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: shrb $2, %sil
> -; SSE3-NEXT: andb $51, %sil
> -; SSE3-NEXT: addb %al, %sil
> -; SSE3-NEXT: movb %sil, %al
> -; SSE3-NEXT: shrb $4, %al
> -; SSE3-NEXT: addb %sil, %al
> -; SSE3-NEXT: andb $15, %al
> -; SSE3-NEXT: movzbl %al, %eax
> -; SSE3-NEXT: movd %eax, %xmm4
> -; SSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb %cl
> -; SSE3-NEXT: andb $85, %cl
> -; SSE3-NEXT: subb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: andb $51, %cl
> -; SSE3-NEXT: shrb $2, %al
> -; SSE3-NEXT: andb $51, %al
> -; SSE3-NEXT: addb %cl, %al
> -; SSE3-NEXT: movb %al, %cl
> -; SSE3-NEXT: shrb $4, %cl
> -; SSE3-NEXT: addb %al, %cl
> -; SSE3-NEXT: andb $15, %cl
> -; SSE3-NEXT: movzbl %cl, %eax
> -; SSE3-NEXT: movd %eax, %xmm0
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSE3-NEXT: popq %rbx
> -; SSE3-NEXT: popq %rbp
> -; SSE3-NEXT: retq
> -;
> -; SSSE3-LABEL: testv16i8:
> -; SSSE3: # BB#0:
> -; SSSE3-NEXT: pushq %rbp
> -; SSSE3-NEXT: .Ltmp0:
> -; SSSE3-NEXT: .cfi_def_cfa_offset 16
> -; SSSE3-NEXT: pushq %rbx
> -; SSSE3-NEXT: .Ltmp1:
> -; SSSE3-NEXT: .cfi_def_cfa_offset 24
> -; SSSE3-NEXT: .Ltmp2:
> -; SSSE3-NEXT: .cfi_offset %rbx, -24
> -; SSSE3-NEXT: .Ltmp3:
> -; SSSE3-NEXT: .cfi_offset %rbp, -16
> -; SSSE3-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r10b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dil
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r9b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bpl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %sil
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %bl
> -; SSSE3-NEXT: movb %bl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %bl
> -; SSSE3-NEXT: movb %bl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %bl
> -; SSSE3-NEXT: andb $51, %bl
> -; SSSE3-NEXT: addb %al, %bl
> -; SSSE3-NEXT: movb %bl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %bl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %dl
> -; SSSE3-NEXT: andb $51, %dl
> -; SSSE3-NEXT: addb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %dl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r11b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %dl
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %r8b
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb %al, %bl
> -; SSSE3-NEXT: shrb %bl
> -; SSSE3-NEXT: andb $85, %bl
> -; SSSE3-NEXT: subb %bl, %al
> -; SSSE3-NEXT: movb %al, %bl
> -; SSSE3-NEXT: andb $51, %bl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %bl, %al
> -; SSSE3-NEXT: movb %al, %bl
> -; SSSE3-NEXT: shrb $4, %bl
> -; SSSE3-NEXT: addb %al, %bl
> -; SSSE3-NEXT: andb $15, %bl
> -; SSSE3-NEXT: movzbl %bl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSSE3-NEXT: movb %cl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %cl
> -; SSSE3-NEXT: movb %cl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: movb %cl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm1
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %dl
> -; SSSE3-NEXT: andb $51, %dl
> -; SSSE3-NEXT: addb %al, %dl
> -; SSSE3-NEXT: movb %dl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %dl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
> -; SSSE3-NEXT: movb %bpl, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %bpl
> -; SSSE3-NEXT: movb %bpl, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %bpl
> -; SSSE3-NEXT: andb $51, %bpl
> -; SSSE3-NEXT: addb %al, %bpl
> -; SSSE3-NEXT: movb %bpl, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %bpl, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
> -; SSSE3-NEXT: movb %cl, %dl
> -; SSSE3-NEXT: shrb %dl
> -; SSSE3-NEXT: andb $85, %dl
> -; SSSE3-NEXT: subb %dl, %cl
> -; SSSE3-NEXT: movb %cl, %dl
> -; SSSE3-NEXT: andb $51, %dl
> -; SSSE3-NEXT: shrb $2, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: addb %dl, %cl
> -; SSSE3-NEXT: movb %cl, %dl
> -; SSSE3-NEXT: shrb $4, %dl
> -; SSSE3-NEXT: addb %cl, %dl
> -; SSSE3-NEXT: andb $15, %dl
> -; SSSE3-NEXT: movzbl %dl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm1
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
> -; SSSE3-NEXT: movb %r10b, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %r10b
> -; SSSE3-NEXT: movb %r10b, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %r10b
> -; SSSE3-NEXT: andb $51, %r10b
> -; SSSE3-NEXT: addb %cl, %r10b
> -; SSSE3-NEXT: movb %r10b, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %r10b, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm2
> -; SSSE3-NEXT: movb %r11b, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %r11b
> -; SSSE3-NEXT: movb %r11b, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %r11b
> -; SSSE3-NEXT: andb $51, %r11b
> -; SSSE3-NEXT: addb %cl, %r11b
> -; SSSE3-NEXT: movb %r11b, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %r11b, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm0
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSSE3-NEXT: movb %r9b, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %r9b
> -; SSSE3-NEXT: movb %r9b, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %r9b
> -; SSSE3-NEXT: andb $51, %r9b
> -; SSSE3-NEXT: addb %cl, %r9b
> -; SSSE3-NEXT: movb %r9b, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %r9b, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %ecx
> -; SSSE3-NEXT: movd %ecx, %xmm3
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm2
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3],xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
> -; SSSE3-NEXT: movb %dil, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %dil
> -; SSSE3-NEXT: movb %dil, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %dil
> -; SSSE3-NEXT: andb $51, %dil
> -; SSSE3-NEXT: addb %al, %dil
> -; SSSE3-NEXT: movb %dil, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %dil, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: movb %r8b, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %r8b
> -; SSSE3-NEXT: movb %r8b, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %r8b
> -; SSSE3-NEXT: andb $51, %r8b
> -; SSSE3-NEXT: addb %al, %r8b
> -; SSSE3-NEXT: movb %r8b, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %r8b, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm3
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
> -; SSSE3-NEXT: movb %sil, %al
> -; SSSE3-NEXT: shrb %al
> -; SSSE3-NEXT: andb $85, %al
> -; SSSE3-NEXT: subb %al, %sil
> -; SSSE3-NEXT: movb %sil, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: shrb $2, %sil
> -; SSSE3-NEXT: andb $51, %sil
> -; SSSE3-NEXT: addb %al, %sil
> -; SSSE3-NEXT: movb %sil, %al
> -; SSSE3-NEXT: shrb $4, %al
> -; SSSE3-NEXT: addb %sil, %al
> -; SSSE3-NEXT: andb $15, %al
> -; SSSE3-NEXT: movzbl %al, %eax
> -; SSSE3-NEXT: movd %eax, %xmm4
> -; SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb %cl
> -; SSSE3-NEXT: andb $85, %cl
> -; SSSE3-NEXT: subb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: andb $51, %cl
> -; SSSE3-NEXT: shrb $2, %al
> -; SSSE3-NEXT: andb $51, %al
> -; SSSE3-NEXT: addb %cl, %al
> -; SSSE3-NEXT: movb %al, %cl
> -; SSSE3-NEXT: shrb $4, %cl
> -; SSSE3-NEXT: addb %al, %cl
> -; SSSE3-NEXT: andb $15, %cl
> -; SSSE3-NEXT: movzbl %cl, %eax
> -; SSSE3-NEXT: movd %eax, %xmm0
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
> -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
> -; SSSE3-NEXT: popq %rbx
> -; SSSE3-NEXT: popq %rbp
> -; SSSE3-NEXT: retq
> -;
> -; SSE41-LABEL: testv16i8:
> -; SSE41: # BB#0:
> -; SSE41-NEXT: pextrb $1, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pextrb $0, %xmm0, %ecx
> -; SSE41-NEXT: movb %cl, %dl
> -; SSE41-NEXT: shrb %dl
> -; SSE41-NEXT: andb $85, %dl
> -; SSE41-NEXT: subb %dl, %cl
> -; SSE41-NEXT: movb %cl, %dl
> -; SSE41-NEXT: andb $51, %dl
> -; SSE41-NEXT: shrb $2, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: addb %dl, %cl
> -; SSE41-NEXT: movb %cl, %dl
> -; SSE41-NEXT: shrb $4, %dl
> -; SSE41-NEXT: addb %cl, %dl
> -; SSE41-NEXT: andb $15, %dl
> -; SSE41-NEXT: movzbl %dl, %ecx
> -; SSE41-NEXT: movd %ecx, %xmm1
> -; SSE41-NEXT: pinsrb $1, %eax, %xmm1
> -; SSE41-NEXT: pextrb $2, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $2, %eax, %xmm1
> -; SSE41-NEXT: pextrb $3, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $3, %eax, %xmm1
> -; SSE41-NEXT: pextrb $4, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $4, %eax, %xmm1
> -; SSE41-NEXT: pextrb $5, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $5, %eax, %xmm1
> -; SSE41-NEXT: pextrb $6, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $6, %eax, %xmm1
> -; SSE41-NEXT: pextrb $7, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $7, %eax, %xmm1
> -; SSE41-NEXT: pextrb $8, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $8, %eax, %xmm1
> -; SSE41-NEXT: pextrb $9, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $9, %eax, %xmm1
> -; SSE41-NEXT: pextrb $10, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $10, %eax, %xmm1
> -; SSE41-NEXT: pextrb $11, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $11, %eax, %xmm1
> -; SSE41-NEXT: pextrb $12, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $12, %eax, %xmm1
> -; SSE41-NEXT: pextrb $13, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $13, %eax, %xmm1
> -; SSE41-NEXT: pextrb $14, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $14, %eax, %xmm1
> -; SSE41-NEXT: pextrb $15, %xmm0, %eax
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb %cl
> -; SSE41-NEXT: andb $85, %cl
> -; SSE41-NEXT: subb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: andb $51, %cl
> -; SSE41-NEXT: shrb $2, %al
> -; SSE41-NEXT: andb $51, %al
> -; SSE41-NEXT: addb %cl, %al
> -; SSE41-NEXT: movb %al, %cl
> -; SSE41-NEXT: shrb $4, %cl
> -; SSE41-NEXT: addb %al, %cl
> -; SSE41-NEXT: andb $15, %cl
> -; SSE41-NEXT: movzbl %cl, %eax
> -; SSE41-NEXT: pinsrb $15, %eax, %xmm1
> -; SSE41-NEXT: movdqa %xmm1, %xmm0
> -; SSE41-NEXT: retq
> +; SSE-LABEL: testv16i8:
> +; SSE: # BB#0:
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $1, %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm1
> +; SSE-NEXT: psubb %xmm1, %xmm0
> +; SSE-NEXT: movdqa {{.*#+}} xmm1 = [51,51,51,51,51,51,51,51,51,51,51,51,51,51,51,51]
> +; SSE-NEXT: movdqa %xmm0, %xmm2
> +; SSE-NEXT: pand %xmm1, %xmm2
> +; SSE-NEXT: psrlw $2, %xmm0
> +; SSE-NEXT: pand {{.*}}(%rip), %xmm0
> +; SSE-NEXT: pand %xmm1, %xmm0
> +; SSE-NEXT: paddb %xmm2, %xmm0
> +; SSE-NEXT: movdqa %xmm0, %xmm1
> +; SSE-NEXT: psrlw $4, %xmm1
> +; SSE-NEXT: movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
> +; SSE-NEXT: pand %xmm2, %xmm1
> +; SSE-NEXT: paddb %xmm0, %xmm1
> +; SSE-NEXT: pand %xmm2, %xmm1
> +; SSE-NEXT: movdqa %xmm1, %xmm0
> +; SSE-NEXT: retq
> ;
> ; AVX-LABEL: testv16i8:
> ; AVX: # BB#0:
> -; AVX-NEXT: vpextrb $1, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpextrb $0, %xmm0, %ecx
> -; AVX-NEXT: movb %cl, %dl
> -; AVX-NEXT: shrb %dl
> -; AVX-NEXT: andb $85, %dl
> -; AVX-NEXT: subb %dl, %cl
> -; AVX-NEXT: movb %cl, %dl
> -; AVX-NEXT: andb $51, %dl
> -; AVX-NEXT: shrb $2, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: addb %dl, %cl
> -; AVX-NEXT: movb %cl, %dl
> -; AVX-NEXT: shrb $4, %dl
> -; AVX-NEXT: addb %cl, %dl
> -; AVX-NEXT: andb $15, %dl
> -; AVX-NEXT: movzbl %dl, %ecx
> -; AVX-NEXT: vmovd %ecx, %xmm1
> -; AVX-NEXT: vpinsrb $1, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $2, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $2, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $3, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $3, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $4, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $4, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $5, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $5, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $6, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $6, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $7, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $7, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $8, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $8, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $9, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $9, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $10, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $10, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $11, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb $4, %cl
> -; AVX-NEXT: addb %al, %cl
> -; AVX-NEXT: andb $15, %cl
> -; AVX-NEXT: movzbl %cl, %eax
> -; AVX-NEXT: vpinsrb $11, %eax, %xmm1, %xmm1
> -; AVX-NEXT: vpextrb $12, %xmm0, %eax
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: shrb %cl
> -; AVX-NEXT: andb $85, %cl
> -; AVX-NEXT: subb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: andb $51, %cl
> -; AVX-NEXT: shrb $2, %al
> -; AVX-NEXT: andb $51, %al
> -; AVX-NEXT: addb %cl, %al
> -; AVX-NEXT: movb %al, %cl
> -; AVX-NEXT: ...
>
> [Message clipped]
--
Bruno Cardoso Lopes
http://www.brunocardoso.cc
More information about the llvm-commits
mailing list