llvm-6502/test/CodeGen/Hexagon/dsub.ll

19 lines
598 B
LLVM
Raw Normal View History

; RUN: llc -march=hexagon -mcpu=hexagonv5 < %s | FileCheck %s
; Check that we generate double precision floating point subtract in V5.
; CHECK: r{{[0-9]+}}:{{[0-9]+}} = dfsub(r{{[0-9]+}}:{{[0-9]+}}, r{{[0-9]+}}:{{[0-9]+}})
define i32 @main() nounwind {
entry:
%a = alloca double, align 8
%b = alloca double, align 8
%c = alloca double, align 8
store double 1.540000e+01, double* %a, align 8
store double 9.100000e+00, double* %b, align 8
%0 = load double* %b, align 8
%1 = load double* %a, align 8
%sub = fsub double %0, %1
store double %sub, double* %c, align 8
ret i32 0
}