1 ; RUN: llc < %s -march=nvptx -mcpu=sm_20 -fp-contract=fast | FileCheck %s 2 3 declare float @dummy_f32(float, float) #0 4 declare double @dummy_f64(double, double) #0 5 6 define ptx_device float @t1_f32(float %x, float %y, float %z) { 7 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}; 8 ; CHECK: ret; 9 %a = fmul float %x, %y 10 %b = fadd float %a, %z 11 ret float %b 12 } 13 14 define ptx_device float @t2_f32(float %x, float %y, float %z, float %w) { 15 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}; 16 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}; 17 ; CHECK: ret; 18 %a = fmul float %x, %y 19 %b = fadd float %a, %z 20 %c = fadd float %a, %w 21 %d = call float @dummy_f32(float %b, float %c) 22 ret float %d 23 } 24 25 define ptx_device double @t1_f64(double %x, double %y, double %z) { 26 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}; 27 ; CHECK: ret; 28 %a = fmul double %x, %y 29 %b = fadd double %a, %z 30 ret double %b 31 } 32 33 define ptx_device double @t2_f64(double %x, double %y, double %z, double %w) { 34 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}; 35 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}; 36 ; CHECK: ret; 37 %a = fmul double %x, %y 38 %b = fadd double %a, %z 39 %c = fadd double %a, %w 40 %d = call double @dummy_f64(double %b, double %c) 41 ret double %d 42 } 43
