1; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx -mattr=+avx -x86-experimental-vector-widening-legalization | FileCheck %s 2 3; Check that we perform a scalar XOR on i32. 4 5; CHECK: pull_bitcast 6; CHECK: xorl 7; CHECK: ret 8define void @pull_bitcast (<4 x i8>* %pA, <4 x i8>* %pB) { 9 %A = load <4 x i8>, <4 x i8>* %pA 10 %B = load <4 x i8>, <4 x i8>* %pB 11 %C = xor <4 x i8> %A, %B 12 store <4 x i8> %C, <4 x i8>* %pA 13 ret void 14} 15 16; CHECK: multi_use_swizzle 17; CHECK: pshufd 18; CHECK-NEXT: pshufd 19; CHECK-NEXT: pblendw 20; CHECK-NEXT: pshufd 21; CHECK-NEXT: pshufd 22; CHECK-NEXT: pxor 23; CHECK-NEXT: ret 24define <4 x i32> @multi_use_swizzle (<4 x i32>* %pA, <4 x i32>* %pB) { 25 %A = load <4 x i32>, <4 x i32>* %pA 26 %B = load <4 x i32>, <4 x i32>* %pB 27 %S = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 1, i32 1, i32 5, i32 6> 28 %S1 = shufflevector <4 x i32> %S, <4 x i32> undef, <4 x i32> <i32 1, i32 3, i32 2, i32 2> 29 %S2 = shufflevector <4 x i32> %S, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 2> 30 %R = xor <4 x i32> %S1, %S2 31 ret <4 x i32> %R 32} 33 34; CHECK: pull_bitcast2 35; CHECK: xorl 36; CHECK: ret 37define <4 x i8> @pull_bitcast2 (<4 x i8>* %pA, <4 x i8>* %pB, <4 x i8>* %pC) { 38 %A = load <4 x i8>, <4 x i8>* %pA 39 store <4 x i8> %A, <4 x i8>* %pC 40 %B = load <4 x i8>, <4 x i8>* %pB 41 %C = xor <4 x i8> %A, %B 42 store <4 x i8> %C, <4 x i8>* %pA 43 ret <4 x i8> %C 44} 45 46 47 48; CHECK: reverse_1 49; CHECK-NOT: pshufd 50; CHECK: ret 51define <4 x i32> @reverse_1 (<4 x i32>* %pA, <4 x i32>* %pB) { 52 %A = load <4 x i32>, <4 x i32>* %pA 53 %B = load <4 x i32>, <4 x i32>* %pB 54 %S = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 1, i32 0, i32 3, i32 2> 55 %S1 = shufflevector <4 x i32> %S, <4 x i32> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2> 56 ret <4 x i32> %S1 57} 58 59 60; CHECK: no_reverse_shuff 61; CHECK: pshufd 62; CHECK: ret 63define <4 x i32> @no_reverse_shuff (<4 x i32>* %pA, <4 x i32>* %pB) { 64 %A = load <4 x i32>, <4 x i32>* %pA 65 %B = load <4 x i32>, <4 x i32>* %pB 66 %S = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 1, i32 0, i32 3, i32 2> 67 %S1 = shufflevector <4 x i32> %S, <4 x i32> undef, <4 x i32> <i32 3, i32 2, i32 3, i32 2> 68 ret <4 x i32> %S1 69} 70