llvm-for-llvmta/test/CodeGen/ARM/a15-mla.ll

38 lines
1.2 KiB
LLVM

; RUN: llc -mtriple=arm-eabi -float-abi=hard -mcpu=cortex-a15 -mattr=+neon,+neonfp %s -o - \
; RUN: | FileCheck %s
; This test checks that the VMLxForwarting feature is disabled for A15.
; CHECK: fun_a:
define <4 x i32> @fun_a(<4 x i32> %x, <4 x i32> %y) nounwind{
%1 = add <4 x i32> %x, %y
; CHECK-NOT: vmul
; CHECK: vmla
%2 = mul <4 x i32> %1, %1
%3 = add <4 x i32> %y, %2
ret <4 x i32> %3
}
; This tests checks that VMLA FP patterns can be matched in instruction selection when targeting
; Cortex-A15.
; CHECK: fun_b:
define <4 x float> @fun_b(<4 x float> %x, <4 x float> %y, <4 x float> %z) nounwind{
; CHECK: vmla.f32
%t = fmul <4 x float> %x, %y
%r = fadd <4 x float> %t, %z
ret <4 x float> %r
}
; This tests checks that FP VMLA instructions are not expanded into separate multiply/addition
; operations when targeting Cortex-A15.
; CHECK: fun_c:
define <4 x float> @fun_c(<4 x float> %x, <4 x float> %y, <4 x float> %z, <4 x float> %u, <4 x float> %v) nounwind{
; CHECK: vmla.f32
%t1 = fmul <4 x float> %x, %y
%r1 = fadd <4 x float> %t1, %z
; CHECK: vmla.f32
%t2 = fmul <4 x float> %u, %v
%r2 = fadd <4 x float> %t2, %r1
ret <4 x float> %r2
}