/// isVPKUHUMShuffleMask - Return true if this is the shuffle mask for a
/// VPKUHUM instruction.
-bool PPC::isVPKUHUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary) {
+bool PPC::isVPKUHUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary,
+ SelectionDAG &DAG) {
+ unsigned j = DAG.getTarget().getDataLayout()->isLittleEndian() ? 0 : 1;
if (!isUnary) {
for (unsigned i = 0; i != 16; ++i)
- if (!isConstantOrUndef(N->getMaskElt(i), i*2+1))
+ if (!isConstantOrUndef(N->getMaskElt(i), i*2+j))
return false;
} else {
for (unsigned i = 0; i != 8; ++i)
- if (!isConstantOrUndef(N->getMaskElt(i), i*2+1) ||
- !isConstantOrUndef(N->getMaskElt(i+8), i*2+1))
+ if (!isConstantOrUndef(N->getMaskElt(i), i*2+j) ||
+ !isConstantOrUndef(N->getMaskElt(i+8), i*2+j))
return false;
}
return true;
/// isVPKUWUMShuffleMask - Return true if this is the shuffle mask for a
/// VPKUWUM instruction.
-bool PPC::isVPKUWUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary) {
+bool PPC::isVPKUWUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary,
+ SelectionDAG &DAG) {
+ unsigned j, k;
+ if (DAG.getTarget().getDataLayout()->isLittleEndian()) {
+ j = 0;
+ k = 1;
+ } else {
+ j = 2;
+ k = 3;
+ }
if (!isUnary) {
for (unsigned i = 0; i != 16; i += 2)
- if (!isConstantOrUndef(N->getMaskElt(i ), i*2+2) ||
- !isConstantOrUndef(N->getMaskElt(i+1), i*2+3))
+ if (!isConstantOrUndef(N->getMaskElt(i ), i*2+j) ||
+ !isConstantOrUndef(N->getMaskElt(i+1), i*2+k))
return false;
} else {
for (unsigned i = 0; i != 8; i += 2)
- if (!isConstantOrUndef(N->getMaskElt(i ), i*2+2) ||
- !isConstantOrUndef(N->getMaskElt(i+1), i*2+3) ||
- !isConstantOrUndef(N->getMaskElt(i+8), i*2+2) ||
- !isConstantOrUndef(N->getMaskElt(i+9), i*2+3))
+ if (!isConstantOrUndef(N->getMaskElt(i ), i*2+j) ||
+ !isConstantOrUndef(N->getMaskElt(i+1), i*2+k) ||
+ !isConstantOrUndef(N->getMaskElt(i+8), i*2+j) ||
+ !isConstantOrUndef(N->getMaskElt(i+9), i*2+k))
return false;
}
return true;
}
/// isVMRGLShuffleMask - Return true if this is a shuffle mask suitable for
-/// a VRGL* instruction with the specified unit size (1,2 or 4 bytes).
+/// a VMRGL* instruction with the specified unit size (1,2 or 4 bytes).
bool PPC::isVMRGLShuffleMask(ShuffleVectorSDNode *N, unsigned UnitSize,
- bool isUnary) {
- if (!isUnary)
- return isVMerge(N, UnitSize, 8, 24);
- return isVMerge(N, UnitSize, 8, 8);
+ bool isUnary, SelectionDAG &DAG) {
+ if (DAG.getTarget().getDataLayout()->isLittleEndian()) {
+ if (!isUnary)
+ return isVMerge(N, UnitSize, 0, 16);
+ return isVMerge(N, UnitSize, 0, 0);
+ } else {
+ if (!isUnary)
+ return isVMerge(N, UnitSize, 8, 24);
+ return isVMerge(N, UnitSize, 8, 8);
+ }
}
/// isVMRGHShuffleMask - Return true if this is a shuffle mask suitable for
-/// a VRGH* instruction with the specified unit size (1,2 or 4 bytes).
+/// a VMRGH* instruction with the specified unit size (1,2 or 4 bytes).
bool PPC::isVMRGHShuffleMask(ShuffleVectorSDNode *N, unsigned UnitSize,
- bool isUnary) {
- if (!isUnary)
- return isVMerge(N, UnitSize, 0, 16);
- return isVMerge(N, UnitSize, 0, 0);
+ bool isUnary, SelectionDAG &DAG) {
+ if (DAG.getTarget().getDataLayout()->isLittleEndian()) {
+ if (!isUnary)
+ return isVMerge(N, UnitSize, 8, 24);
+ return isVMerge(N, UnitSize, 8, 8);
+ } else {
+ if (!isUnary)
+ return isVMerge(N, UnitSize, 0, 16);
+ return isVMerge(N, UnitSize, 0, 0);
+ }
}
/// isVSLDOIShuffleMask - If this is a vsldoi shuffle mask, return the shift
/// amount, otherwise return -1.
-int PPC::isVSLDOIShuffleMask(SDNode *N, bool isUnary) {
+int PPC::isVSLDOIShuffleMask(SDNode *N, bool isUnary, SelectionDAG &DAG) {
if (N->getValueType(0) != MVT::v16i8)
return -1;
// numbered from this value.
unsigned ShiftAmt = SVOp->getMaskElt(i);
if (ShiftAmt < i) return -1;
- ShiftAmt -= i;
- if (!isUnary) {
- // Check the rest of the elements to see if they are consecutive.
- for (++i; i != 16; ++i)
- if (!isConstantOrUndef(SVOp->getMaskElt(i), ShiftAmt+i))
- return -1;
- } else {
- // Check the rest of the elements to see if they are consecutive.
- for (++i; i != 16; ++i)
- if (!isConstantOrUndef(SVOp->getMaskElt(i), (ShiftAmt+i) & 15))
- return -1;
+ if (DAG.getTarget().getDataLayout()->isLittleEndian()) {
+
+ ShiftAmt += i;
+
+ if (!isUnary) {
+ // Check the rest of the elements to see if they are consecutive.
+ for (++i; i != 16; ++i)
+ if (!isConstantOrUndef(SVOp->getMaskElt(i), ShiftAmt - i))
+ return -1;
+ } else {
+ // Check the rest of the elements to see if they are consecutive.
+ for (++i; i != 16; ++i)
+ if (!isConstantOrUndef(SVOp->getMaskElt(i), (ShiftAmt - i) & 15))
+ return -1;
+ }
+
+ } else { // Big Endian
+
+ ShiftAmt -= i;
+
+ if (!isUnary) {
+ // Check the rest of the elements to see if they are consecutive.
+ for (++i; i != 16; ++i)
+ if (!isConstantOrUndef(SVOp->getMaskElt(i), ShiftAmt+i))
+ return -1;
+ } else {
+ // Check the rest of the elements to see if they are consecutive.
+ for (++i; i != 16; ++i)
+ if (!isConstantOrUndef(SVOp->getMaskElt(i), (ShiftAmt+i) & 15))
+ return -1;
+ }
}
return ShiftAmt;
}
/// getVSPLTImmediate - Return the appropriate VSPLT* immediate to splat the
/// specified isSplatShuffleMask VECTOR_SHUFFLE mask.
-unsigned PPC::getVSPLTImmediate(SDNode *N, unsigned EltSize) {
+unsigned PPC::getVSPLTImmediate(SDNode *N, unsigned EltSize,
+ SelectionDAG &DAG) {
ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
assert(isSplatShuffleMask(SVOp, EltSize));
- return SVOp->getMaskElt(0) / EltSize;
+ if (DAG.getTarget().getDataLayout()->isLittleEndian())
+ return (16 / EltSize) - 1 - (SVOp->getMaskElt(0) / EltSize);
+ else
+ return SVOp->getMaskElt(0) / EltSize;
}
/// get_VSPLTI_elt - If this is a build_vector of constants which can be formed
SDValue V2 = Op.getOperand(1);
ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(Op);
EVT VT = Op.getValueType();
+ bool isLittleEndian = PPCSubTarget.isLittleEndian();
// Cases that are handled by instructions that take permute immediates
// (such as vsplt*) should be left as VECTOR_SHUFFLE nodes so they can be
if (PPC::isSplatShuffleMask(SVOp, 1) ||
PPC::isSplatShuffleMask(SVOp, 2) ||
PPC::isSplatShuffleMask(SVOp, 4) ||
- PPC::isVPKUWUMShuffleMask(SVOp, true) ||
- PPC::isVPKUHUMShuffleMask(SVOp, true) ||
- PPC::isVSLDOIShuffleMask(SVOp, true) != -1 ||
- PPC::isVMRGLShuffleMask(SVOp, 1, true) ||
- PPC::isVMRGLShuffleMask(SVOp, 2, true) ||
- PPC::isVMRGLShuffleMask(SVOp, 4, true) ||
- PPC::isVMRGHShuffleMask(SVOp, 1, true) ||
- PPC::isVMRGHShuffleMask(SVOp, 2, true) ||
- PPC::isVMRGHShuffleMask(SVOp, 4, true)) {
+ PPC::isVPKUWUMShuffleMask(SVOp, true, DAG) ||
+ PPC::isVPKUHUMShuffleMask(SVOp, true, DAG) ||
+ PPC::isVSLDOIShuffleMask(SVOp, true, DAG) != -1 ||
+ PPC::isVMRGLShuffleMask(SVOp, 1, true, DAG) ||
+ PPC::isVMRGLShuffleMask(SVOp, 2, true, DAG) ||
+ PPC::isVMRGLShuffleMask(SVOp, 4, true, DAG) ||
+ PPC::isVMRGHShuffleMask(SVOp, 1, true, DAG) ||
+ PPC::isVMRGHShuffleMask(SVOp, 2, true, DAG) ||
+ PPC::isVMRGHShuffleMask(SVOp, 4, true, DAG)) {
return Op;
}
}
// Altivec has a variety of "shuffle immediates" that take two vector inputs
// and produce a fixed permutation. If any of these match, do not lower to
// VPERM.
- if (PPC::isVPKUWUMShuffleMask(SVOp, false) ||
- PPC::isVPKUHUMShuffleMask(SVOp, false) ||
- PPC::isVSLDOIShuffleMask(SVOp, false) != -1 ||
- PPC::isVMRGLShuffleMask(SVOp, 1, false) ||
- PPC::isVMRGLShuffleMask(SVOp, 2, false) ||
- PPC::isVMRGLShuffleMask(SVOp, 4, false) ||
- PPC::isVMRGHShuffleMask(SVOp, 1, false) ||
- PPC::isVMRGHShuffleMask(SVOp, 2, false) ||
- PPC::isVMRGHShuffleMask(SVOp, 4, false))
+ if (PPC::isVPKUWUMShuffleMask(SVOp, false, DAG) ||
+ PPC::isVPKUHUMShuffleMask(SVOp, false, DAG) ||
+ PPC::isVSLDOIShuffleMask(SVOp, false, DAG) != -1 ||
+ PPC::isVMRGLShuffleMask(SVOp, 1, false, DAG) ||
+ PPC::isVMRGLShuffleMask(SVOp, 2, false, DAG) ||
+ PPC::isVMRGLShuffleMask(SVOp, 4, false, DAG) ||
+ PPC::isVMRGHShuffleMask(SVOp, 1, false, DAG) ||
+ PPC::isVMRGHShuffleMask(SVOp, 2, false, DAG) ||
+ PPC::isVMRGHShuffleMask(SVOp, 4, false, DAG))
return Op;
// Check to see if this is a shuffle of 4-byte values. If so, we can use our
// If this shuffle can be expressed as a shuffle of 4-byte elements, use the
// perfect shuffle vector to determine if it is cost effective to do this as
// discrete instructions, or whether we should use a vperm.
- if (isFourElementShuffle) {
+ // For now, we skip this for little endian until such time as we have a
+ // little-endian perfect shuffle table.
+ if (isFourElementShuffle && !isLittleEndian) {
// Compute the index in the perfect shuffle table.
unsigned PFTableIndex =
PFIndexes[0]*9*9*9+PFIndexes[1]*9*9+PFIndexes[2]*9+PFIndexes[3];
// instruction.
EVT EltVT = V1.getValueType().getVectorElementType();
unsigned BytesPerElement = EltVT.getSizeInBits()/8;
- bool isLittleEndian = PPCSubTarget.isLittleEndian();
SmallVector<SDValue, 16> ResultMask;
for (unsigned i = 0, e = VT.getVectorNumElements(); i != e; ++i) {
namespace PPC {
/// isVPKUHUMShuffleMask - Return true if this is the shuffle mask for a
/// VPKUHUM instruction.
- bool isVPKUHUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary);
+ bool isVPKUHUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary,
+ SelectionDAG &DAG);
/// isVPKUWUMShuffleMask - Return true if this is the shuffle mask for a
/// VPKUWUM instruction.
- bool isVPKUWUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary);
+ bool isVPKUWUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary,
+ SelectionDAG &DAG);
/// isVMRGLShuffleMask - Return true if this is a shuffle mask suitable for
/// a VRGL* instruction with the specified unit size (1,2 or 4 bytes).
bool isVMRGLShuffleMask(ShuffleVectorSDNode *N, unsigned UnitSize,
- bool isUnary);
+ bool isUnary, SelectionDAG &DAG);
/// isVMRGHShuffleMask - Return true if this is a shuffle mask suitable for
/// a VRGH* instruction with the specified unit size (1,2 or 4 bytes).
bool isVMRGHShuffleMask(ShuffleVectorSDNode *N, unsigned UnitSize,
- bool isUnary);
+ bool isUnary, SelectionDAG &DAG);
/// isVSLDOIShuffleMask - If this is a vsldoi shuffle mask, return the shift
/// amount, otherwise return -1.
- int isVSLDOIShuffleMask(SDNode *N, bool isUnary);
+ int isVSLDOIShuffleMask(SDNode *N, bool isUnary, SelectionDAG &DAG);
/// isSplatShuffleMask - Return true if the specified VECTOR_SHUFFLE operand
/// specifies a splat of a single element that is suitable for input to
/// getVSPLTImmediate - Return the appropriate VSPLT* immediate to splat the
/// specified isSplatShuffleMask VECTOR_SHUFFLE mask.
- unsigned getVSPLTImmediate(SDNode *N, unsigned EltSize);
+ unsigned getVSPLTImmediate(SDNode *N, unsigned EltSize, SelectionDAG &DAG);
/// get_VSPLTI_elt - If this is a build_vector of constants which can be
/// formed by using a vspltis[bhw] instruction of the specified element
def vpkuhum_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVPKUHUMShuffleMask(cast<ShuffleVectorSDNode>(N), false);
+ return PPC::isVPKUHUMShuffleMask(cast<ShuffleVectorSDNode>(N), false,
+ *CurDAG);
}]>;
def vpkuwum_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), false);
+ return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), false,
+ *CurDAG);
}]>;
def vpkuhum_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVPKUHUMShuffleMask(cast<ShuffleVectorSDNode>(N), true);
+ return PPC::isVPKUHUMShuffleMask(cast<ShuffleVectorSDNode>(N), true,
+ *CurDAG);
}]>;
def vpkuwum_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), true);
+ return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), true,
+ *CurDAG);
}]>;
def vmrglb_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 1, false);
+ return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 1, false,
+ *CurDAG);
}]>;
def vmrglh_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 2, false);
+ return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 2, false,
+ *CurDAG);
}]>;
def vmrglw_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 4, false);
+ return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 4, false,
+ *CurDAG);
}]>;
def vmrghb_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 1, false);
+ return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 1, false,
+ *CurDAG);
}]>;
def vmrghh_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 2, false);
+ return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 2, false,
+ *CurDAG);
}]>;
def vmrghw_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 4, false);
+ return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 4, false,
+ *CurDAG);
}]>;
def vmrglb_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
- return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 1, true);
+ return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 1, true,
+ *CurDAG);
}]>;
def vmrglh_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 2, true);
+ return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 2, true,
+ *CurDAG);
}]>;
def vmrglw_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 4, true);
+ return PPC::isVMRGLShuffleMask(cast<ShuffleVectorSDNode>(N), 4, true,
+ *CurDAG);
}]>;
def vmrghb_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 1, true);
+ return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 1, true,
+ *CurDAG);
}]>;
def vmrghh_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 2, true);
+ return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 2, true,
+ *CurDAG);
}]>;
def vmrghw_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 4, true);
+ return PPC::isVMRGHShuffleMask(cast<ShuffleVectorSDNode>(N), 4, true,
+ *CurDAG);
}]>;
def VSLDOI_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::isVSLDOIShuffleMask(N, false));
+ return getI32Imm(PPC::isVSLDOIShuffleMask(N, false, *CurDAG));
}]>;
def vsldoi_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVSLDOIShuffleMask(N, false) != -1;
+ return PPC::isVSLDOIShuffleMask(N, false, *CurDAG) != -1;
}], VSLDOI_get_imm>;
/// VSLDOI_unary* - These are used to match vsldoi(X,X), which is turned into
/// vector_shuffle(X,undef,mask) by the dag combiner.
def VSLDOI_unary_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::isVSLDOIShuffleMask(N, true));
+ return getI32Imm(PPC::isVSLDOIShuffleMask(N, true, *CurDAG));
}]>;
def vsldoi_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
- return PPC::isVSLDOIShuffleMask(N, true) != -1;
+ return PPC::isVSLDOIShuffleMask(N, true, *CurDAG) != -1;
}], VSLDOI_unary_get_imm>;
// VSPLT*_get_imm xform function: convert vector_shuffle mask to VSPLT* imm.
def VSPLTB_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::getVSPLTImmediate(N, 1));
+ return getI32Imm(PPC::getVSPLTImmediate(N, 1, *CurDAG));
}]>;
def vspltb_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isSplatShuffleMask(cast<ShuffleVectorSDNode>(N), 1);
}], VSPLTB_get_imm>;
def VSPLTH_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::getVSPLTImmediate(N, 2));
+ return getI32Imm(PPC::getVSPLTImmediate(N, 2, *CurDAG));
}]>;
def vsplth_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isSplatShuffleMask(cast<ShuffleVectorSDNode>(N), 2);
}], VSPLTH_get_imm>;
def VSPLTW_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::getVSPLTImmediate(N, 4));
+ return getI32Imm(PPC::getVSPLTImmediate(N, 4, *CurDAG));
}]>;
def vspltw_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
--- /dev/null
+; RUN: llc < %s -mtriple=powerpc64le-unknown-linux-gnu -mattr=+altivec | FileCheck %s
+
+define void @VPKUHUM_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VPKUHUM_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK: vpkuhum
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VPKUHUM_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VPKUHUM_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+; CHECK: vpkuhum
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VPKUWUM_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VPKUWUM_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 0, i32 1, i32 4, i32 5, i32 8, i32 9, i32 12, i32 13, i32 16, i32 17, i32 20, i32 21, i32 24, i32 25, i32 28, i32 29>
+; CHECK: vpkuwum
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VPKUWUM_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VPKUWUM_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 0, i32 1, i32 4, i32 5, i32 8, i32 9, i32 12, i32 13, i32 0, i32 1, i32 4, i32 5, i32 8, i32 9, i32 12, i32 13>
+; CHECK: vpkuwum
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGLB_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VMRGLB_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23>
+; CHECK: vmrglb
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGLB_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VMRGLB_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 0, i32 0, i32 1, i32 1, i32 2, i32 2, i32 3, i32 3, i32 4, i32 4, i32 5, i32 5, i32 6, i32 6, i32 7, i32 7>
+; CHECK: vmrglb
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGHB_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VMRGHB_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27, i32 12, i32 28, i32 13, i32 29, i32 14, i32 30, i32 15, i32 31>
+; CHECK: vmrghb
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGHB_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VMRGHB_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 8, i32 8, i32 9, i32 9, i32 10, i32 10, i32 11, i32 11, i32 12, i32 12, i32 13, i32 13, i32 14, i32 14, i32 15, i32 15>
+; CHECK: vmrghb
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGLH_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VMRGLH_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 2, i32 3, i32 18, i32 19, i32 4, i32 5, i32 20, i32 21, i32 6, i32 7, i32 22, i32 23>
+; CHECK: vmrglh
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGLH_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VMRGLH_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 2, i32 3, i32 2, i32 3, i32 4, i32 5, i32 4, i32 5, i32 6, i32 7, i32 6, i32 7>
+; CHECK: vmrglh
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGHH_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VMRGHH_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 8, i32 9, i32 24, i32 25, i32 10, i32 11, i32 26, i32 27, i32 12, i32 13, i32 28, i32 29, i32 14, i32 15, i32 30, i32 31>
+; CHECK: vmrghh
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGHH_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VMRGHH_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 8, i32 9, i32 8, i32 9, i32 10, i32 11, i32 10, i32 11, i32 12, i32 13, i32 12, i32 13, i32 14, i32 15, i32 14, i32 15>
+; CHECK: vmrghh
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGLW_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VMRGLW_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23>
+; CHECK: vmrglw
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGLW_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VMRGLW_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7>
+; CHECK: vmrglw
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGHW_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VMRGHW_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 24, i32 25, i32 26, i32 27, i32 12, i32 13, i32 14, i32 15, i32 28, i32 29, i32 30, i32 31>
+; CHECK: vmrghw
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VMRGHW_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VMRGHW_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 12, i32 13, i32 14, i32 15>
+; CHECK: vmrghw
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
+define void @VSLDOI_xy(<16 x i8>* %A, <16 x i8>* %B) {
+entry:
+; CHECK: VSLDOI_xy:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = load <16 x i8>* %B
+ %tmp3 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp2, <16 x i32> <i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4>
+; CHECK: vsldoi
+ store <16 x i8> %tmp3, <16 x i8>* %A
+ ret void
+}
+
+define void @VSLDOI_xx(<16 x i8>* %A) {
+entry:
+; CHECK: VSLDOI_xx:
+ %tmp = load <16 x i8>* %A
+ %tmp2 = shufflevector <16 x i8> %tmp, <16 x i8> %tmp, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4>
+; CHECK: vsldoi
+ store <16 x i8> %tmp2, <16 x i8>* %A
+ ret void
+}
+
projects / oota-llvm.git / commitdiff