replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove AVX vpermil intrinsics. I removed their uses from clang headers and builtins a while back. 

llvm-svn: 154985
This commit is contained in:
Craig Topper 2012-04-18 05:24:00 +00:00
parent 26b47e24fe
commit d3c9e404ba
3 changed files with 38 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
llvm/include/llvm/IntrinsicsX86.td
View File

@ -1091,20 +1091,6 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.".
GCCBuiltin<"__builtin_ia32_vperm2f128_si256">,
Intrinsic<[llvm_v8i32_ty], [llvm_v8i32_ty,
llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx_vpermil_pd :
Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty,
llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx_vpermil_ps :
Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty,
llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx_vpermil_pd_256 :
Intrinsic<[llvm_v4f64_ty], [llvm_v4f64_ty,
llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx_vpermil_ps_256 :
Intrinsic<[llvm_v8f32_ty], [llvm_v8f32_ty,
llvm_i8_ty], [IntrNoMem]>;
}
// Vector blend

6
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -9592,12 +9592,6 @@ X86TargetLowering::LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG) const
case Intrinsic::x86_avx2_vperm2i128:
return DAG.getNode(X86ISD::VPERM2X128, dl, Op.getValueType(),
Op.getOperand(1), Op.getOperand(2), Op.getOperand(3));
case Intrinsic::x86_avx_vpermil_ps:
case Intrinsic::x86_avx_vpermil_pd:
case Intrinsic::x86_avx_vpermil_ps_256:
case Intrinsic::x86_avx_vpermil_pd_256:
return DAG.getNode(X86ISD::VPERMILP, dl, Op.getValueType(),
Op.getOperand(1), Op.getOperand(2));
case Intrinsic::x86_avx2_permd:
case Intrinsic::x86_avx2_permps:
// Operands intentionally swapped. Mask is last operand to intrinsic,

43
llvm/lib/VMCore/AutoUpgrade.cpp
View File

@ -18,9 +18,6 @@
#include "llvm/LLVMContext.h"
#include "llvm/Module.h"
#include "llvm/IntrinsicInst.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/Support/CallSite.h"
#include "llvm/Support/CFG.h"
#include "llvm/Support/ErrorHandling.h"
@ -59,7 +56,8 @@ static bool UpgradeIntrinsicFunction1(Function *F, Function *&NewFn) {
if (Name.startswith("x86.sse2.pcmpeq.") ||
Name.startswith("x86.sse2.pcmpgt.") ||
Name.startswith("x86.avx2.pcmpeq.") ||
Name.startswith("x86.avx2.pcmpgt.")) {
Name.startswith("x86.avx2.pcmpgt.") ||
Name.startswith("x86.avx.vpermil.")) {
NewFn = 0;
return true;
}
@ -121,7 +119,42 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
// need to sign extend since icmp returns vector of i1
Rep = Builder.CreateSExt(Rep, CI->getType(), "");
} else {
llvm_unreachable("Unknown function for CallInst upgrade.");
bool PD128 = false, PD256 = false, PS128 = false, PS256 = false;
if (Name.startswith("llvm.x86.avx.vpermil.pd.256"))
PD256 = true;
else if (Name.startswith("llvm.x86.avx.vpermil.pd"))
PD128 = true;
else if (Name.startswith("llvm.x86.avx.vpermil.ps.256"))
PS256 = true;
else if (Name.startswith("llvm.x86.avx.vpermil.ps"))
PS128 = true;
if (PD256 || PD128 || PS256 || PS128) {
Value *Op0 = CI->getArgOperand(0);
unsigned Imm = cast<ConstantInt>(CI->getArgOperand(1))->getZExtValue();
SmallVector<Constant*, 8> Idxs;
if (PD128)
for (unsigned i = 0; i != 2; ++i)
Idxs.push_back(Builder.getInt32((Imm >> i) & 0x1));
else if (PD256)
for (unsigned l = 0; l != 4; l+=2)
for (unsigned i = 0; i != 2; ++i)
Idxs.push_back(Builder.getInt32(((Imm >> (l+i)) & 0x1) + l));
else if (PS128)
for (unsigned i = 0; i != 4; ++i)
Idxs.push_back(Builder.getInt32((Imm >> (2 * i)) & 0x3));
else if (PS256)
for (unsigned l = 0; l != 8; l+=4)
for (unsigned i = 0; i != 4; ++i)
Idxs.push_back(Builder.getInt32(((Imm >> (2 * i)) & 0x3) + l));
else
llvm_unreachable("Unexpected function");
Rep = Builder.CreateShuffleVector(Op0, Op0, ConstantVector::get(Idxs));
} else {
llvm_unreachable("Unknown function for CallInst upgrade.");
}
}
CI->replaceAllUsesWith(Rep);