Index: include/llvm/CodeGen/BasicTTIImpl.h =================================================================== --- include/llvm/CodeGen/BasicTTIImpl.h +++ include/llvm/CodeGen/BasicTTIImpl.h @@ -1116,6 +1116,7 @@ unsigned getIntrinsicInstrCost( Intrinsic::ID IID, Type *RetTy, ArrayRef Tys, FastMathFlags FMF, unsigned ScalarizationCostPassed = std::numeric_limits::max()) { + unsigned RetVF = (RetTy->isVectorTy() ? RetTy->getVectorNumElements() : 1); auto *ConcreteTTI = static_cast(this); SmallVector ISDs; @@ -1266,6 +1267,20 @@ return ConcreteTTI->getMinMaxReductionCost( Tys[0], CmpInst::makeCmpResultType(Tys[0]), /*IsPairwiseForm=*/false, /*IsSigned=*/false); + case Intrinsic::usub_sat: { + Type *CondTy = Type::getInt1Ty(RetTy->getContext()); + if (RetVF > 1) + CondTy = VectorType::get(CondTy, RetVF); + + // usub.sat(x, y) -> umax(x, y) - y + unsigned Cost = 0; + Cost += ConcreteTTI->getCmpSelInstrCost(BinaryOperator::ICmp, RetTy, + CondTy, nullptr); + Cost += ConcreteTTI->getCmpSelInstrCost(BinaryOperator::Select, RetTy, + CondTy, nullptr); + Cost += ConcreteTTI->getArithmeticInstrCost(BinaryOperator::Sub, RetTy); + return Cost; + } case Intrinsic::ctpop: ISDs.push_back(ISD::CTPOP); // In case of legalization use TCC_Expensive. This is cheaper than a Index: test/Analysis/CostModel/X86/arith-usat.ll =================================================================== --- test/Analysis/CostModel/X86/arith-usat.ll +++ test/Analysis/CostModel/X86/arith-usat.ll @@ -251,190 +251,190 @@ define i32 @sub(i32 %arg) { ; SSSE3-LABEL: 'sub' -; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; SSSE3-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; SSSE3-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; SSE42-LABEL: 'sub' -; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; AVX1-LABEL: 'sub' -; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; AVX1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; AVX1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; AVX2-LABEL: 'sub' -; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; AVX2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; AVX2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; AVX2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; AVX2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; AVX512F-LABEL: 'sub' -; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; AVX512F-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; AVX512F-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; AVX512F-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; AVX512F-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; AVX512F-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; AVX512BW-LABEL: 'sub' -; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; AVX512BW-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; AVX512BW-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; AVX512BW-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; AVX512BW-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; AVX512BW-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; AVX512DQ-LABEL: 'sub' -; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; SLM-LABEL: 'sub' -; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; SLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; GLM-LABEL: 'sub' -; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; GLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; GLM-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; ; BTVER2-LABEL: 'sub' -; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) -; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) -; BTVER2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) -; BTVER2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) -; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef) -; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef) -; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) +; BTVER2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef) ; BTVER2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef) Index: test/Transforms/SLPVectorizer/X86/arith-sub-usat.ll =================================================================== --- test/Transforms/SLPVectorizer/X86/arith-sub-usat.ll +++ test/Transforms/SLPVectorizer/X86/arith-sub-usat.ll @@ -62,109 +62,34 @@ ; SSE-NEXT: ret void ; ; SLM-LABEL: @sub_v8i64( -; SLM-NEXT: [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8 -; SLM-NEXT: [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8 -; SLM-NEXT: [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8 -; SLM-NEXT: [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8 -; SLM-NEXT: [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8 -; SLM-NEXT: [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8 -; SLM-NEXT: [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8 -; SLM-NEXT: [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8 -; SLM-NEXT: [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8 -; SLM-NEXT: [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8 -; SLM-NEXT: [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8 -; SLM-NEXT: [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8 -; SLM-NEXT: [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8 -; SLM-NEXT: [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8 -; SLM-NEXT: [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8 -; SLM-NEXT: [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8 -; SLM-NEXT: [[R0:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A0]], i64 [[B0]]) -; SLM-NEXT: [[R1:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A1]], i64 [[B1]]) -; SLM-NEXT: [[R2:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A2]], i64 [[B2]]) -; SLM-NEXT: [[R3:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A3]], i64 [[B3]]) -; SLM-NEXT: [[R4:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A4]], i64 [[B4]]) -; SLM-NEXT: [[R5:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A5]], i64 [[B5]]) -; SLM-NEXT: [[R6:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A6]], i64 [[B6]]) -; SLM-NEXT: [[R7:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A7]], i64 [[B7]]) -; SLM-NEXT: store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8 -; SLM-NEXT: store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8 -; SLM-NEXT: store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8 -; SLM-NEXT: store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8 -; SLM-NEXT: store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8 -; SLM-NEXT: store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8 -; SLM-NEXT: store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8 -; SLM-NEXT: store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8 +; SLM-NEXT: [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @b64 to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP6:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2) to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP8:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6) to <2 x i64>*), align 8 +; SLM-NEXT: [[TMP9:%.*]] = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP5]]) +; SLM-NEXT: [[TMP10:%.*]] = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> [[TMP2]], <2 x i64> [[TMP6]]) +; SLM-NEXT: [[TMP11:%.*]] = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> [[TMP3]], <2 x i64> [[TMP7]]) +; SLM-NEXT: [[TMP12:%.*]] = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP8]]) +; SLM-NEXT: store <2 x i64> [[TMP9]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8 +; SLM-NEXT: store <2 x i64> [[TMP10]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8 +; SLM-NEXT: store <2 x i64> [[TMP11]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8 +; SLM-NEXT: store <2 x i64> [[TMP12]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8 ; SLM-NEXT: ret void ; -; AVX1-LABEL: @sub_v8i64( -; AVX1-NEXT: [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8 -; AVX1-NEXT: [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8 -; AVX1-NEXT: [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8 -; AVX1-NEXT: [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8 -; AVX1-NEXT: [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8 -; AVX1-NEXT: [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8 -; AVX1-NEXT: [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8 -; AVX1-NEXT: [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8 -; AVX1-NEXT: [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8 -; AVX1-NEXT: [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8 -; AVX1-NEXT: [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8 -; AVX1-NEXT: [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8 -; AVX1-NEXT: [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8 -; AVX1-NEXT: [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8 -; AVX1-NEXT: [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8 -; AVX1-NEXT: [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8 -; AVX1-NEXT: [[R0:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A0]], i64 [[B0]]) -; AVX1-NEXT: [[R1:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A1]], i64 [[B1]]) -; AVX1-NEXT: [[R2:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A2]], i64 [[B2]]) -; AVX1-NEXT: [[R3:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A3]], i64 [[B3]]) -; AVX1-NEXT: [[R4:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A4]], i64 [[B4]]) -; AVX1-NEXT: [[R5:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A5]], i64 [[B5]]) -; AVX1-NEXT: [[R6:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A6]], i64 [[B6]]) -; AVX1-NEXT: [[R7:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A7]], i64 [[B7]]) -; AVX1-NEXT: store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8 -; AVX1-NEXT: store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8 -; AVX1-NEXT: store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8 -; AVX1-NEXT: store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8 -; AVX1-NEXT: store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8 -; AVX1-NEXT: store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8 -; AVX1-NEXT: store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8 -; AVX1-NEXT: store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8 -; AVX1-NEXT: ret void -; -; AVX2-LABEL: @sub_v8i64( -; AVX2-NEXT: [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8 -; AVX2-NEXT: [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8 -; AVX2-NEXT: [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8 -; AVX2-NEXT: [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8 -; AVX2-NEXT: [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8 -; AVX2-NEXT: [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8 -; AVX2-NEXT: [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8 -; AVX2-NEXT: [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8 -; AVX2-NEXT: [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8 -; AVX2-NEXT: [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8 -; AVX2-NEXT: [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8 -; AVX2-NEXT: [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8 -; AVX2-NEXT: [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8 -; AVX2-NEXT: [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8 -; AVX2-NEXT: [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8 -; AVX2-NEXT: [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8 -; AVX2-NEXT: [[R0:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A0]], i64 [[B0]]) -; AVX2-NEXT: [[R1:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A1]], i64 [[B1]]) -; AVX2-NEXT: [[R2:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A2]], i64 [[B2]]) -; AVX2-NEXT: [[R3:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A3]], i64 [[B3]]) -; AVX2-NEXT: [[R4:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A4]], i64 [[B4]]) -; AVX2-NEXT: [[R5:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A5]], i64 [[B5]]) -; AVX2-NEXT: [[R6:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A6]], i64 [[B6]]) -; AVX2-NEXT: [[R7:%.*]] = call i64 @llvm.usub.sat.i64(i64 [[A7]], i64 [[B7]]) -; AVX2-NEXT: store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8 -; AVX2-NEXT: store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8 -; AVX2-NEXT: store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8 -; AVX2-NEXT: store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8 -; AVX2-NEXT: store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8 -; AVX2-NEXT: store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8 -; AVX2-NEXT: store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8 -; AVX2-NEXT: store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8 -; AVX2-NEXT: ret void +; AVX-LABEL: @sub_v8i64( +; AVX-NEXT: [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8 +; AVX-NEXT: [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8 +; AVX-NEXT: [[TMP3:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @b64 to <4 x i64>*), align 8 +; AVX-NEXT: [[TMP4:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <4 x i64>*), align 8 +; AVX-NEXT: [[TMP5:%.*]] = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]]) +; AVX-NEXT: [[TMP6:%.*]] = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]]) +; AVX-NEXT: store <4 x i64> [[TMP5]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8 +; AVX-NEXT: store <4 x i64> [[TMP6]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8 +; AVX-NEXT: ret void ; ; AVX512-LABEL: @sub_v8i64( ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @a64 to <8 x i64>*), align 8 @@ -173,17 +98,6 @@ ; AVX512-NEXT: store <8 x i64> [[TMP3]], <8 x i64>* bitcast ([8 x i64]* @c64 to <8 x i64>*), align 8 ; AVX512-NEXT: ret void ; -; AVX256BW-LABEL: @sub_v8i64( -; AVX256BW-NEXT: [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8 -; AVX256BW-NEXT: [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8 -; AVX256BW-NEXT: [[TMP3:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @b64 to <4 x i64>*), align 8 -; AVX256BW-NEXT: [[TMP4:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <4 x i64>*), align 8 -; AVX256BW-NEXT: [[TMP5:%.*]] = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]]) -; AVX256BW-NEXT: [[TMP6:%.*]] = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]]) -; AVX256BW-NEXT: store <4 x i64> [[TMP5]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8 -; AVX256BW-NEXT: store <4 x i64> [[TMP6]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8 -; AVX256BW-NEXT: ret void -; %a0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8 %a1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8 %a2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8 @@ -221,70 +135,22 @@ define void @sub_v16i32() { ; SSE-LABEL: @sub_v16i32( -; SSE-NEXT: [[A0:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 0), align 4 -; SSE-NEXT: [[A1:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 1), align 4 -; SSE-NEXT: [[A2:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 2), align 4 -; SSE-NEXT: [[A3:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 3), align 4 -; SSE-NEXT: [[A4:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4), align 4 -; SSE-NEXT: [[A5:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 5), align 4 -; SSE-NEXT: [[A6:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 6), align 4 -; SSE-NEXT: [[A7:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 7), align 4 -; SSE-NEXT: [[A8:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8), align 4 -; SSE-NEXT: [[A9:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 9), align 4 -; SSE-NEXT: [[A10:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 10), align 4 -; SSE-NEXT: [[A11:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 11), align 4 -; SSE-NEXT: [[A12:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12), align 4 -; SSE-NEXT: [[A13:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 13), align 4 -; SSE-NEXT: [[A14:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 14), align 4 -; SSE-NEXT: [[A15:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 15), align 4 -; SSE-NEXT: [[B0:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 0), align 4 -; SSE-NEXT: [[B1:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 1), align 4 -; SSE-NEXT: [[B2:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 2), align 4 -; SSE-NEXT: [[B3:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 3), align 4 -; SSE-NEXT: [[B4:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4), align 4 -; SSE-NEXT: [[B5:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 5), align 4 -; SSE-NEXT: [[B6:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 6), align 4 -; SSE-NEXT: [[B7:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 7), align 4 -; SSE-NEXT: [[B8:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8), align 4 -; SSE-NEXT: [[B9:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 9), align 4 -; SSE-NEXT: [[B10:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 10), align 4 -; SSE-NEXT: [[B11:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 11), align 4 -; SSE-NEXT: [[B12:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12), align 4 -; SSE-NEXT: [[B13:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 13), align 4 -; SSE-NEXT: [[B14:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 14), align 4 -; SSE-NEXT: [[B15:%.*]] = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 15), align 4 -; SSE-NEXT: [[R0:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A0]], i32 [[B0]]) -; SSE-NEXT: [[R1:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A1]], i32 [[B1]]) -; SSE-NEXT: [[R2:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A2]], i32 [[B2]]) -; SSE-NEXT: [[R3:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A3]], i32 [[B3]]) -; SSE-NEXT: [[R4:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A4]], i32 [[B4]]) -; SSE-NEXT: [[R5:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A5]], i32 [[B5]]) -; SSE-NEXT: [[R6:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A6]], i32 [[B6]]) -; SSE-NEXT: [[R7:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A7]], i32 [[B7]]) -; SSE-NEXT: [[R8:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A8]], i32 [[B8]]) -; SSE-NEXT: [[R9:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A9]], i32 [[B9]]) -; SSE-NEXT: [[R10:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A10]], i32 [[B10]]) -; SSE-NEXT: [[R11:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A11]], i32 [[B11]]) -; SSE-NEXT: [[R12:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A12]], i32 [[B12]]) -; SSE-NEXT: [[R13:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A13]], i32 [[B13]]) -; SSE-NEXT: [[R14:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A14]], i32 [[B14]]) -; SSE-NEXT: [[R15:%.*]] = call i32 @llvm.usub.sat.i32(i32 [[A15]], i32 [[B15]]) -; SSE-NEXT: store i32 [[R0]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 0), align 4 -; SSE-NEXT: store i32 [[R1]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 1), align 4 -; SSE-NEXT: store i32 [[R2]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 2), align 4 -; SSE-NEXT: store i32 [[R3]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 3), align 4 -; SSE-NEXT: store i32 [[R4]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4), align 4 -; SSE-NEXT: store i32 [[R5]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 5), align 4 -; SSE-NEXT: store i32 [[R6]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 6), align 4 -; SSE-NEXT: store i32 [[R7]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 7), align 4 -; SSE-NEXT: store i32 [[R8]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8), align 4 -; SSE-NEXT: store i32 [[R9]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 9), align 4 -; SSE-NEXT: store i32 [[R10]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 10), align 4 -; SSE-NEXT: store i32 [[R11]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 11), align 4 -; SSE-NEXT: store i32 [[R12]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12), align 4 -; SSE-NEXT: store i32 [[R13]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 13), align 4 -; SSE-NEXT: store i32 [[R14]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 14), align 4 -; SSE-NEXT: store i32 [[R15]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 15), align 4 +; SSE-NEXT: [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4 +; SSE-NEXT: [[TMP9:%.*]] = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]]) +; SSE-NEXT: [[TMP10:%.*]] = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]]) +; SSE-NEXT: [[TMP11:%.*]] = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]]) +; SSE-NEXT: [[TMP12:%.*]] = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]]) +; SSE-NEXT: store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4 +; SSE-NEXT: store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4 +; SSE-NEXT: store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4 +; SSE-NEXT: store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4 ; SSE-NEXT: ret void ; ; SLM-LABEL: @sub_v16i32(