1 ; Test vector multiply-and-add. 2 ; 3 ; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s 4 5 declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) 6 7 ; Test a v16i8 multiply-and-add. 8 define <16 x i8> @f1(<16 x i8> %dummy, <16 x i8> %val1, <16 x i8> %val2, 9 <16 x i8> %val3) { 10 ; CHECK-LABEL: f1: 11 ; CHECK: vmalb %v24, %v26, %v28, %v30 12 ; CHECK: br %r14 13 %mul = mul <16 x i8> %val1, %val2 14 %ret = add <16 x i8> %mul, %val3 15 ret <16 x i8> %ret 16 } 17 18 ; Test a v8i16 multiply-and-add. 19 define <8 x i16> @f2(<8 x i16> %dummy, <8 x i16> %val1, <8 x i16> %val2, 20 <8 x i16> %val3) { 21 ; CHECK-LABEL: f2: 22 ; CHECK: vmalhw %v24, %v26, %v28, %v30 23 ; CHECK: br %r14 24 %mul = mul <8 x i16> %val1, %val2 25 %ret = add <8 x i16> %mul, %val3 26 ret <8 x i16> %ret 27 } 28 29 ; Test a v4i32 multiply-and-add. 30 define <4 x i32> @f3(<4 x i32> %dummy, <4 x i32> %val1, <4 x i32> %val2, 31 <4 x i32> %val3) { 32 ; CHECK-LABEL: f3: 33 ; CHECK: vmalf %v24, %v26, %v28, %v30 34 ; CHECK: br %r14 35 %mul = mul <4 x i32> %val1, %val2 36 %ret = add <4 x i32> %mul, %val3 37 ret <4 x i32> %ret 38 } 39 40 ; Test a v2f64 multiply-and-add. 41 define <2 x double> @f4(<2 x double> %dummy, <2 x double> %val1, 42 <2 x double> %val2, <2 x double> %val3) { 43 ; CHECK-LABEL: f4: 44 ; CHECK: vfmadb %v24, %v26, %v28, %v30 45 ; CHECK: br %r14 46 %ret = call <2 x double> @llvm.fma.v2f64 (<2 x double> %val1, 47 <2 x double> %val2, 48 <2 x double> %val3) 49 ret <2 x double> %ret 50 } 51 52 ; Test a v2f64 multiply-and-subtract. 53 define <2 x double> @f5(<2 x double> %dummy, <2 x double> %val1, 54 <2 x double> %val2, <2 x double> %val3) { 55 ; CHECK-LABEL: f5: 56 ; CHECK: vfmsdb %v24, %v26, %v28, %v30 57 ; CHECK: br %r14 58 %negval3 = fsub <2 x double> <double -0.0, double -0.0>, %val3 59 %ret = call <2 x double> @llvm.fma.v2f64 (<2 x double> %val1, 60 <2 x double> %val2, 61 <2 x double> %negval3) 62 ret <2 x double> %ret 63 } 64
