1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve.fp -o - %s | FileCheck %s
3 ; RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve -o - %s | FileCheck %s
5 define arm_aapcs_vfpcc <4 x i32> @vector_add_by_value(<4 x i32> %lhs, <4 x i32>%rhs) {
6 ; CHECK-LABEL: vector_add_by_value:
9 ; CHECK-NEXT: vadd.i32 q0, q0, q1
12 %result = tail call <4 x i32> asm "vadd.i32 $0,$1,$2", "=t,t,t"(<4 x i32> %lhs, <4 x i32> %rhs)
16 define void @vector_add_by_reference(ptr %resultp, ptr %lhsp, ptr %rhsp) {
17 ; CHECK-LABEL: vector_add_by_reference:
19 ; CHECK-NEXT: vldrw.u32 q0, [r1]
20 ; CHECK-NEXT: vldrw.u32 q1, [r2]
22 ; CHECK-NEXT: vadd.i32 q0, q0, q1
24 ; CHECK-NEXT: vstrw.32 q0, [r0]
26 %lhs = load <4 x i32>, ptr %lhsp, align 16
27 %rhs = load <4 x i32>, ptr %rhsp, align 16
28 %result = tail call <4 x i32> asm "vadd.i32 $0,$1,$2", "=t,t,t"(<4 x i32> %lhs, <4 x i32> %rhs)
29 store <4 x i32> %result, ptr %resultp, align 16
33 define void @vector_f64_copy(ptr %from, ptr %to) {
34 ; CHECK-LABEL: vector_f64_copy:
36 ; CHECK-NEXT: vldrw.u32 q0, [r0]
37 ; CHECK-NEXT: vstrw.32 q0, [r1]
39 %v = load <2 x double>, ptr %from, align 16
40 store <2 x double> %v, ptr %to, align 16
44 define arm_aapcs_vfpcc <16 x i8> @stack_slot_handling(<16 x i8> %a) #0 {
45 ; CHECK-LABEL: stack_slot_handling:
46 ; CHECK: @ %bb.0: @ %entry
47 ; CHECK-NEXT: sub sp, #16
48 ; CHECK-NEXT: mov r0, sp
49 ; CHECK-NEXT: vstrw.32 q0, [r0]
50 ; CHECK-NEXT: vldrw.u32 q0, [r0]
51 ; CHECK-NEXT: add sp, #16
54 %a.addr = alloca <16 x i8>, align 8
55 store <16 x i8> %a, ptr %a.addr, align 8
56 %0 = load <16 x i8>, ptr %a.addr, align 8
60 attributes #0 = { noinline optnone }