1 ; RUN: llc -mcpu=cortex-a8 < %s | FileCheck %s
3 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64"
4 target triple = "armv7-eabi"
6 %foo = type { <4 x float> }
8 define arm_aapcs_vfpcc void @bar(ptr noalias sret(%foo) %agg.result, <4 x float> %quat.0) nounwind {
10 %quat_addr = alloca %foo, align 16 ; <ptr> [#uses=2]
11 %0 = getelementptr inbounds %foo, ptr %quat_addr, i32 0, i32 0 ; <ptr> [#uses=1]
12 store <4 x float> %quat.0, ptr %0
13 %1 = call arm_aapcs_vfpcc <4 x float> @quux(ptr %quat_addr) nounwind ; <<4 x float>> [#uses=3]
14 %2 = fmul <4 x float> %1, %1 ; <<4 x float>> [#uses=2]
15 %3 = shufflevector <4 x float> %2, <4 x float> undef, <2 x i32> <i32 0, i32 1> ; <<2 x float>> [#uses=1]
16 %4 = shufflevector <4 x float> %2, <4 x float> undef, <2 x i32> <i32 2, i32 3> ; <<2 x float>> [#uses=1]
19 %5 = call <2 x float> @llvm.arm.neon.vpadd.v2f32(<2 x float> %3, <2 x float> %4) nounwind ; <<2 x float>> [#uses=2]
20 %6 = call <2 x float> @llvm.arm.neon.vpadd.v2f32(<2 x float> %5, <2 x float> %5) nounwind ; <<2 x float>> [#uses=2]
21 %7 = shufflevector <2 x float> %6, <2 x float> %6, <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; <<4 x float>> [#uses=2]
23 %8 = call <4 x float> @llvm.arm.neon.vrsqrte.v4f32(<4 x float> %7) nounwind ; <<4 x float>> [#uses=3]
24 %9 = fmul <4 x float> %8, %8 ; <<4 x float>> [#uses=1]
25 %10 = call <4 x float> @llvm.arm.neon.vrsqrts.v4f32(<4 x float> %9, <4 x float> %7) nounwind ; <<4 x float>> [#uses=1]
26 %11 = fmul <4 x float> %10, %8 ; <<4 x float>> [#uses=1]
27 %12 = fmul <4 x float> %11, %1 ; <<4 x float>> [#uses=1]
28 %13 = call arm_aapcs_vfpcc ptr @baz(ptr %agg.result, <4 x float> %12) nounwind ; <ptr> [#uses=0]
32 declare arm_aapcs_vfpcc ptr @baz(ptr, <4 x float>) nounwind
34 declare arm_aapcs_vfpcc <4 x float> @quux(ptr nocapture) nounwind readonly
36 declare <2 x float> @llvm.arm.neon.vpadd.v2f32(<2 x float>, <2 x float>) nounwind readnone
38 declare <4 x float> @llvm.arm.neon.vrsqrte.v4f32(<4 x float>) nounwind readnone
40 declare <4 x float> @llvm.arm.neon.vrsqrts.v4f32(<4 x float>, <4 x float>) nounwind readnone