1 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=knl --show-mc-encoding| FileCheck %s 2 3 define <16 x float> @floor_v16f32(<16 x float> %a) { 4 ; CHECK-LABEL: floor_v16f32 5 ; CHECK: vrndscaleps $1, {{.*}}encoding: [0x62,0xf3,0x7d,0x48,0x08,0xc0,0x01] 6 %res = call <16 x float> @llvm.floor.v16f32(<16 x float> %a) 7 ret <16 x float> %res 8 } 9 declare <16 x float> @llvm.floor.v16f32(<16 x float> %p) 10 11 define <8 x double> @floor_v8f64(<8 x double> %a) { 12 ; CHECK-LABEL: floor_v8f64 13 ; CHECK: vrndscalepd $1, {{.*}}encoding: [0x62,0xf3,0xfd,0x48,0x09,0xc0,0x01] 14 %res = call <8 x double> @llvm.floor.v8f64(<8 x double> %a) 15 ret <8 x double> %res 16 } 17 declare <8 x double> @llvm.floor.v8f64(<8 x double> %p) 18 19 define <16 x float> @ceil_v16f32(<16 x float> %a) { 20 ; CHECK-LABEL: ceil_v16f32 21 ; CHECK: vrndscaleps $2, {{.*}}encoding: [0x62,0xf3,0x7d,0x48,0x08,0xc0,0x02] 22 %res = call <16 x float> @llvm.ceil.v16f32(<16 x float> %a) 23 ret <16 x float> %res 24 } 25 declare <16 x float> @llvm.ceil.v16f32(<16 x float> %p) 26 27 define <8 x double> @ceil_v8f64(<8 x double> %a) { 28 ; CHECK-LABEL: ceil_v8f64 29 ; CHECK: vrndscalepd $2, {{.*}}encoding: [0x62,0xf3,0xfd,0x48,0x09,0xc0,0x02] 30 %res = call <8 x double> @llvm.ceil.v8f64(<8 x double> %a) 31 ret <8 x double> %res 32 } 33 declare <8 x double> @llvm.ceil.v8f64(<8 x double> %p) 34 35 define <16 x float> @trunc_v16f32(<16 x float> %a) { 36 ; CHECK-LABEL: trunc_v16f32 37 ; CHECK: vrndscaleps $3, {{.*}}encoding: [0x62,0xf3,0x7d,0x48,0x08,0xc0,0x03] 38 %res = call <16 x float> @llvm.trunc.v16f32(<16 x float> %a) 39 ret <16 x float> %res 40 } 41 declare <16 x float> @llvm.trunc.v16f32(<16 x float> %p) 42 43 define <8 x double> @trunc_v8f64(<8 x double> %a) { 44 ; CHECK-LABEL: trunc_v8f64 45 ; CHECK: vrndscalepd $3, {{.*}}encoding: [0x62,0xf3,0xfd,0x48,0x09,0xc0,0x03] 46 %res = call <8 x double> @llvm.trunc.v8f64(<8 x double> %a) 47 ret <8 x double> %res 48 } 49 declare <8 x double> @llvm.trunc.v8f64(<8 x double> %p) 50 51 define <16 x float> @rint_v16f32(<16 x float> %a) { 52 ; CHECK-LABEL: rint_v16f32 53 ; CHECK: vrndscaleps $4, {{.*}}encoding: [0x62,0xf3,0x7d,0x48,0x08,0xc0,0x04] 54 %res = call <16 x float> @llvm.rint.v16f32(<16 x float> %a) 55 ret <16 x float> %res 56 } 57 declare <16 x float> @llvm.rint.v16f32(<16 x float> %p) 58 59 define <8 x double> @rint_v8f64(<8 x double> %a) { 60 ; CHECK-LABEL: rint_v8f64 61 ; CHECK: vrndscalepd $4, {{.*}}encoding: [0x62,0xf3,0xfd,0x48,0x09,0xc0,0x04] 62 %res = call <8 x double> @llvm.rint.v8f64(<8 x double> %a) 63 ret <8 x double> %res 64 } 65 declare <8 x double> @llvm.rint.v8f64(<8 x double> %p) 66 67 define <16 x float> @nearbyint_v16f32(<16 x float> %a) { 68 ; CHECK-LABEL: nearbyint_v16f32 69 ; CHECK: vrndscaleps $12, {{.*}}encoding: [0x62,0xf3,0x7d,0x48,0x08,0xc0,0x0c] 70 %res = call <16 x float> @llvm.nearbyint.v16f32(<16 x float> %a) 71 ret <16 x float> %res 72 } 73 declare <16 x float> @llvm.nearbyint.v16f32(<16 x float> %p) 74 75 define <8 x double> @nearbyint_v8f64(<8 x double> %a) { 76 ; CHECK-LABEL: nearbyint_v8f64 77 ; CHECK: vrndscalepd $12, {{.*}}encoding: [0x62,0xf3,0xfd,0x48,0x09,0xc0,0x0c] 78 %res = call <8 x double> @llvm.nearbyint.v8f64(<8 x double> %a) 79 ret <8 x double> %res 80 } 81 declare <8 x double> @llvm.nearbyint.v8f64(<8 x double> %p) 82 83 define double @nearbyint_f64(double %a) { 84 ; CHECK-LABEL: nearbyint_f64 85 ; CHECK: vrndscalesd $12, {{.*}}encoding: [0x62,0xf3,0xfd,0x08,0x0b,0xc0,0x0c] 86 %res = call double @llvm.nearbyint.f64(double %a) 87 ret double %res 88 } 89 declare double @llvm.nearbyint.f64(double %p) 90 91 define float @floor_f32(float %a) { 92 ; CHECK-LABEL: floor_f32 93 ; CHECK: vrndscaless $1, {{.*}}encoding: [0x62,0xf3,0x7d,0x08,0x0a,0xc0,0x01] 94 %res = call float @llvm.floor.f32(float %a) 95 ret float %res 96 } 97 declare float @llvm.floor.f32(float %p) 98 99 define float @floor_f32m(float* %aptr) { 100 ; CHECK-LABEL: floor_f32m 101 ; CHECK: vrndscaless $1, (%rdi), {{.*}}encoding: [0x62,0xf3,0x7d,0x08,0x0a,0x07,0x01] 102 %a = load float, float* %aptr, align 4 103 %res = call float @llvm.floor.f32(float %a) 104 ret float %res 105 } 106 107
