1; RUN: llc -mtriple armv8a-none-linux-gnu -mattr=+dotprod -float-abi=hard < %s | FileCheck %s 2 3declare <2 x i32> @llvm.arm.neon.udot.v2i32.v8i8(<2 x i32>, <8 x i8>, <8 x i8>) 4declare <4 x i32> @llvm.arm.neon.udot.v4i32.v16i8(<4 x i32>, <16 x i8>, <16 x i8>) 5declare <2 x i32> @llvm.arm.neon.sdot.v2i32.v8i8(<2 x i32>, <8 x i8>, <8 x i8>) 6declare <4 x i32> @llvm.arm.neon.sdot.v4i32.v16i8(<4 x i32>, <16 x i8>, <16 x i8>) 7 8define <2 x i32> @test_vdot_u32(<2 x i32> %a, <8 x i8> %b, <8 x i8> %c) #0 { 9entry: 10; CHECK-LABEL: test_vdot_u32: 11; CHECK: vudot.u8 d0, d1, d2 12 %vdot1.i = call <2 x i32> @llvm.arm.neon.udot.v2i32.v8i8(<2 x i32> %a, <8 x i8> %b, <8 x i8> %c) #2 13 ret <2 x i32> %vdot1.i 14} 15 16define <4 x i32> @test_vdotq_u32(<4 x i32> %a, <16 x i8> %b, <16 x i8> %c) #0 { 17entry: 18; CHECK-LABEL: test_vdotq_u32: 19; CHECK: vudot.u8 q0, q1, q2 20 %vdot1.i = call <4 x i32> @llvm.arm.neon.udot.v4i32.v16i8(<4 x i32> %a, <16 x i8> %b, <16 x i8> %c) #2 21 ret <4 x i32> %vdot1.i 22} 23 24define <2 x i32> @test_vdot_s32(<2 x i32> %a, <8 x i8> %b, <8 x i8> %c) #0 { 25entry: 26; CHECK-LABEL: test_vdot_s32: 27; CHECK: vsdot.s8 d0, d1, d2 28 %vdot1.i = call <2 x i32> @llvm.arm.neon.sdot.v2i32.v8i8(<2 x i32> %a, <8 x i8> %b, <8 x i8> %c) #2 29 ret <2 x i32> %vdot1.i 30} 31 32define <4 x i32> @test_vdotq_s32(<4 x i32> %a, <16 x i8> %b, <16 x i8> %c) #0 { 33entry: 34; CHECK-LABEL: test_vdotq_s32: 35; CHECK: vsdot.s8 q0, q1, q2 36 %vdot1.i = call <4 x i32> @llvm.arm.neon.sdot.v4i32.v16i8(<4 x i32> %a, <16 x i8> %b, <16 x i8> %c) #2 37 ret <4 x i32> %vdot1.i 38} 39 40define <2 x i32> @test_vdot_lane_u32(<2 x i32> %a, <8 x i8> %b, <8 x i8> %c) { 41entry: 42; CHECK-LABEL: test_vdot_lane_u32: 43; CHECK: vudot.u8 d0, d1, d2[1] 44 %.cast = bitcast <8 x i8> %c to <2 x i32> 45 %shuffle = shufflevector <2 x i32> %.cast, <2 x i32> undef, <2 x i32> <i32 1, i32 1> 46 %.cast5 = bitcast <2 x i32> %shuffle to <8 x i8> 47 %vdot1.i = call <2 x i32> @llvm.arm.neon.udot.v2i32.v8i8(<2 x i32> %a, <8 x i8> %b, <8 x i8> %.cast5) #2 48 ret <2 x i32> %vdot1.i 49} 50 51define <4 x i32> @test_vdotq_lane_u32(<4 x i32> %a, <16 x i8> %b, <8 x i8> %c) { 52entry: 53; CHECK-LABEL: test_vdotq_lane_u32: 54; CHECK: vudot.u8 q0, q1, d4[1] 55 %.cast = bitcast <8 x i8> %c to <2 x i32> 56 %shuffle = shufflevector <2 x i32> %.cast, <2 x i32> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1> 57 %.cast3 = bitcast <4 x i32> %shuffle to <16 x i8> 58 %vdot1.i = call <4 x i32> @llvm.arm.neon.udot.v4i32.v16i8(<4 x i32> %a, <16 x i8> %b, <16 x i8> %.cast3) #2 59 ret <4 x i32> %vdot1.i 60} 61 62define <2 x i32> @test_vdot_lane_s32(<2 x i32> %a, <8 x i8> %b, <8 x i8> %c) { 63entry: 64; CHECK-LABEL: test_vdot_lane_s32: 65; CHECK: vsdot.s8 d0, d1, d2[1] 66 %.cast = bitcast <8 x i8> %c to <2 x i32> 67 %shuffle = shufflevector <2 x i32> %.cast, <2 x i32> undef, <2 x i32> <i32 1, i32 1> 68 %.cast5 = bitcast <2 x i32> %shuffle to <8 x i8> 69 %vdot1.i = call <2 x i32> @llvm.arm.neon.sdot.v2i32.v8i8(<2 x i32> %a, <8 x i8> %b, <8 x i8> %.cast5) #2 70 ret <2 x i32> %vdot1.i 71} 72 73define <4 x i32> @test_vdotq_lane_s32(<4 x i32> %a, <16 x i8> %b, <8 x i8> %c) { 74entry: 75; CHECK-LABEL: test_vdotq_lane_s32: 76; CHECK: vsdot.s8 q0, q1, d4[1] 77 %.cast = bitcast <8 x i8> %c to <2 x i32> 78 %shuffle = shufflevector <2 x i32> %.cast, <2 x i32> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1> 79 %.cast3 = bitcast <4 x i32> %shuffle to <16 x i8> 80 %vdot1.i = call <4 x i32> @llvm.arm.neon.sdot.v4i32.v16i8(<4 x i32> %a, <16 x i8> %b, <16 x i8> %.cast3) #2 81 ret <4 x i32> %vdot1.i 82} 83