forked from OSchip/llvm-project
19 lines
541 B
LLVM
19 lines
541 B
LLVM
; RUN: llc -march=hexagon -mcpu=hexagonv5 < %s | FileCheck %s
|
|
; Check that we generate double precision floating point subtract in V5.
|
|
|
|
; CHECK: call __hexagon_subdf3
|
|
|
|
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
|
|
}
|