1 // REQUIRES: powerpc-registered-target 2 // RUN: %clang_cc1 -triple powerpc64-unknown-unknown \ 3 // RUN: -emit-llvm %s -o - -target-cpu pwr7 | FileCheck %s 4 // RUN: %clang_cc1 -triple powerpc64le-unknown-unknown \ 5 // RUN: -emit-llvm %s -o - -target-cpu pwr8 | FileCheck %s 6 // RUN: %clang_cc1 -triple powerpc64-unknown-aix \ 7 // RUN: -emit-llvm %s -o - -target-cpu pwr7 | FileCheck %s 8 // RUN: %clang_cc1 -triple powerpc-unknown-aix \ 9 // RUN: -emit-llvm %s -o - -target-cpu pwr7 | FileCheck %s 10 11 extern vector float a; 12 extern vector float b; 13 extern vector float c; 14 extern vector double d; 15 extern vector double e; 16 extern vector double f; 17 18 // CHECK-LABEL: @test_flags_recipdivf( 19 // CHECK: [[TMP0:%.*]] = load <4 x float>, <4 x float>* @a, align 16 20 // CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* @b, align 16 21 // CHECK-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* @a, align 16 22 // CHECK-NEXT: [[TMP3:%.*]] = load <4 x float>, <4 x float>* @b, align 16 23 // CHECK-NEXT: [[RECIPDIV:%.*]] = fdiv fast <4 x float> [[TMP2]], [[TMP3]] 24 // CHECK-NEXT: [[TMP4:%.*]] = load <4 x float>, <4 x float>* @c, align 16 25 // CHECK-NEXT: [[ADD:%.*]] = fadd <4 x float> [[RECIPDIV]], [[TMP4]] 26 // CHECK-NEXT: ret <4 x float> [[ADD]] 27 // 28 vector float test_flags_recipdivf() { 29 return __builtin_ppc_recipdivf(a, b) + c; 30 } 31 32 // CHECK-LABEL: @test_flags_recipdivd( 33 // CHECK: [[TMP0:%.*]] = load <2 x double>, <2 x double>* @d, align 16 34 // CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, <2 x double>* @e, align 16 35 // CHECK-NEXT: [[TMP2:%.*]] = load <2 x double>, <2 x double>* @d, align 16 36 // CHECK-NEXT: [[TMP3:%.*]] = load <2 x double>, <2 x double>* @e, align 16 37 // CHECK-NEXT: [[RECIPDIV:%.*]] = fdiv fast <2 x double> [[TMP2]], [[TMP3]] 38 // CHECK-NEXT: [[TMP4:%.*]] = load <2 x double>, <2 x double>* @f, align 16 39 // CHECK-NEXT: [[ADD:%.*]] = fadd <2 x double> [[RECIPDIV]], [[TMP4]] 40 // CHECK-NEXT: ret <2 x double> [[ADD]] 41 // 42 vector double test_flags_recipdivd() { 43 return __builtin_ppc_recipdivd(d, e) + f; 44 } 45 46 // CHECK-LABEL: @test_flags_rsqrtf( 47 // CHECK: [[TMP0:%.*]] = load <4 x float>, <4 x float>* @a, align 16 48 // CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* @a, align 16 49 // CHECK-NEXT: [[TMP2:%.*]] = call fast <4 x float> @llvm.sqrt.v4f32(<4 x float> [[TMP1]]) 50 // CHECK-NEXT: [[RSQRT:%.*]] = fdiv fast <4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>, [[TMP2]] 51 // CHECK-NEXT: [[TMP3:%.*]] = load <4 x float>, <4 x float>* @b, align 16 52 // CHECK-NEXT: [[ADD:%.*]] = fadd <4 x float> [[RSQRT]], [[TMP3]] 53 // CHECK-NEXT: ret <4 x float> [[ADD]] 54 // 55 vector float test_flags_rsqrtf() { 56 return __builtin_ppc_rsqrtf(a) + b; 57 } 58 59 // CHECK-LABEL: @test_flags_rsqrtd( 60 // CHECK: [[TMP0:%.*]] = load <2 x double>, <2 x double>* @d, align 16 61 // CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, <2 x double>* @d, align 16 62 // CHECK-NEXT: [[TMP2:%.*]] = call fast <2 x double> @llvm.sqrt.v2f64(<2 x double> [[TMP1]]) 63 // CHECK-NEXT: [[RSQRT:%.*]] = fdiv fast <2 x double> <double 1.000000e+00, double 1.000000e+00>, [[TMP2]] 64 // CHECK-NEXT: [[TMP3:%.*]] = load <2 x double>, <2 x double>* @e, align 16 65 // CHECK-NEXT: [[ADD:%.*]] = fadd <2 x double> [[RSQRT]], [[TMP3]] 66 // CHECK-NEXT: ret <2 x double> [[ADD]] 67 // 68 vector double test_flags_rsqrtd() { 69 return __builtin_ppc_rsqrtd(d) + e; 70 } 71