1 ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=SI < %s | FileCheck -check-prefix=SI %s
2 ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck -check-prefix=SI %s
4 ; FIXME: Broken on evergreen
5 ; FIXME: For some reason the 8 and 16 vectors are being stored as
6 ; individual elements instead of 128-bit stores.
9 ; FIXME: Why is the constant moved into the intermediate register and
10 ; not just directly into the vector component?
12 ; SI-LABEL: {{^}}insertelement_v4f32_0:
13 ; s_load_dwordx4 s{{[}}[[LOW_REG:[0-9]+]]:
15 ; v_mov_b32_e32 [[CONSTREG:v[0-9]+]], 5.000000e+00
16 ; v_mov_b32_e32 v[[LOW_REG]], [[CONSTREG]]
17 ; buffer_store_dwordx4 v{{[}}[[LOW_REG]]:
18 define void @insertelement_v4f32_0(<4 x float> addrspace(1)* %out, <4 x float> %a) nounwind {
19 %vecins = insertelement <4 x float> %a, float 5.000000e+00, i32 0
20 store <4 x float> %vecins, <4 x float> addrspace(1)* %out, align 16
24 ; SI-LABEL: {{^}}insertelement_v4f32_1:
25 define void @insertelement_v4f32_1(<4 x float> addrspace(1)* %out, <4 x float> %a) nounwind {
26 %vecins = insertelement <4 x float> %a, float 5.000000e+00, i32 1
27 store <4 x float> %vecins, <4 x float> addrspace(1)* %out, align 16
31 ; SI-LABEL: {{^}}insertelement_v4f32_2:
32 define void @insertelement_v4f32_2(<4 x float> addrspace(1)* %out, <4 x float> %a) nounwind {
33 %vecins = insertelement <4 x float> %a, float 5.000000e+00, i32 2
34 store <4 x float> %vecins, <4 x float> addrspace(1)* %out, align 16
38 ; SI-LABEL: {{^}}insertelement_v4f32_3:
39 define void @insertelement_v4f32_3(<4 x float> addrspace(1)* %out, <4 x float> %a) nounwind {
40 %vecins = insertelement <4 x float> %a, float 5.000000e+00, i32 3
41 store <4 x float> %vecins, <4 x float> addrspace(1)* %out, align 16
45 ; SI-LABEL: {{^}}insertelement_v4i32_0:
46 define void @insertelement_v4i32_0(<4 x i32> addrspace(1)* %out, <4 x i32> %a) nounwind {
47 %vecins = insertelement <4 x i32> %a, i32 999, i32 0
48 store <4 x i32> %vecins, <4 x i32> addrspace(1)* %out, align 16
52 ; SI-LABEL: {{^}}dynamic_insertelement_v2f32:
53 ; SI: v_mov_b32_e32 [[CONST:v[0-9]+]], 0x40a00000
54 ; SI: v_movreld_b32_e32 v[[LOW_RESULT_REG:[0-9]+]], [[CONST]]
55 ; SI: buffer_store_dwordx2 {{v\[}}[[LOW_RESULT_REG]]:
56 define void @dynamic_insertelement_v2f32(<2 x float> addrspace(1)* %out, <2 x float> %a, i32 %b) nounwind {
57 %vecins = insertelement <2 x float> %a, float 5.000000e+00, i32 %b
58 store <2 x float> %vecins, <2 x float> addrspace(1)* %out, align 8
62 ; SI-LABEL: {{^}}dynamic_insertelement_v4f32:
63 ; SI: v_mov_b32_e32 [[CONST:v[0-9]+]], 0x40a00000
64 ; SI: v_movreld_b32_e32 v[[LOW_RESULT_REG:[0-9]+]], [[CONST]]
65 ; SI: buffer_store_dwordx4 {{v\[}}[[LOW_RESULT_REG]]:
66 define void @dynamic_insertelement_v4f32(<4 x float> addrspace(1)* %out, <4 x float> %a, i32 %b) nounwind {
67 %vecins = insertelement <4 x float> %a, float 5.000000e+00, i32 %b
68 store <4 x float> %vecins, <4 x float> addrspace(1)* %out, align 16
72 ; SI-LABEL: {{^}}dynamic_insertelement_v8f32:
73 ; FIXMESI: buffer_store_dwordx4
74 ; FIXMESI: buffer_store_dwordx4
75 define void @dynamic_insertelement_v8f32(<8 x float> addrspace(1)* %out, <8 x float> %a, i32 %b) nounwind {
76 %vecins = insertelement <8 x float> %a, float 5.000000e+00, i32 %b
77 store <8 x float> %vecins, <8 x float> addrspace(1)* %out, align 32
81 ; SI-LABEL: {{^}}dynamic_insertelement_v16f32:
82 ; FIXMESI: buffer_store_dwordx4
83 ; FIXMESI: buffer_store_dwordx4
84 ; FIXMESI: buffer_store_dwordx4
85 ; FIXMESI: buffer_store_dwordx4
86 define void @dynamic_insertelement_v16f32(<16 x float> addrspace(1)* %out, <16 x float> %a, i32 %b) nounwind {
87 %vecins = insertelement <16 x float> %a, float 5.000000e+00, i32 %b
88 store <16 x float> %vecins, <16 x float> addrspace(1)* %out, align 64
92 ; SI-LABEL: {{^}}dynamic_insertelement_v2i32:
93 ; SI: buffer_store_dwordx2
94 define void @dynamic_insertelement_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> %a, i32 %b) nounwind {
95 %vecins = insertelement <2 x i32> %a, i32 5, i32 %b
96 store <2 x i32> %vecins, <2 x i32> addrspace(1)* %out, align 8
100 ; SI-LABEL: {{^}}dynamic_insertelement_v4i32:
101 ; SI: buffer_store_dwordx4
102 define void @dynamic_insertelement_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> %a, i32 %b) nounwind {
103 %vecins = insertelement <4 x i32> %a, i32 5, i32 %b
104 store <4 x i32> %vecins, <4 x i32> addrspace(1)* %out, align 16
108 ; SI-LABEL: {{^}}dynamic_insertelement_v8i32:
109 ; FIXMESI: buffer_store_dwordx4
110 ; FIXMESI: buffer_store_dwordx4
111 define void @dynamic_insertelement_v8i32(<8 x i32> addrspace(1)* %out, <8 x i32> %a, i32 %b) nounwind {
112 %vecins = insertelement <8 x i32> %a, i32 5, i32 %b
113 store <8 x i32> %vecins, <8 x i32> addrspace(1)* %out, align 32
117 ; SI-LABEL: {{^}}dynamic_insertelement_v16i32:
118 ; FIXMESI: buffer_store_dwordx4
119 ; FIXMESI: buffer_store_dwordx4
120 ; FIXMESI: buffer_store_dwordx4
121 ; FIXMESI: buffer_store_dwordx4
122 define void @dynamic_insertelement_v16i32(<16 x i32> addrspace(1)* %out, <16 x i32> %a, i32 %b) nounwind {
123 %vecins = insertelement <16 x i32> %a, i32 5, i32 %b
124 store <16 x i32> %vecins, <16 x i32> addrspace(1)* %out, align 64
129 ; SI-LABEL: {{^}}dynamic_insertelement_v2i16:
130 ; FIXMESI: buffer_store_dwordx2
131 define void @dynamic_insertelement_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> %a, i32 %b) nounwind {
132 %vecins = insertelement <2 x i16> %a, i16 5, i32 %b
133 store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out, align 8
137 ; SI-LABEL: {{^}}dynamic_insertelement_v4i16:
138 ; FIXMESI: buffer_store_dwordx4
139 define void @dynamic_insertelement_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> %a, i32 %b) nounwind {
140 %vecins = insertelement <4 x i16> %a, i16 5, i32 %b
141 store <4 x i16> %vecins, <4 x i16> addrspace(1)* %out, align 16
146 ; SI-LABEL: {{^}}dynamic_insertelement_v2i8:
147 ; FIXMESI: BUFFER_STORE_USHORT
148 define void @dynamic_insertelement_v2i8(<2 x i8> addrspace(1)* %out, <2 x i8> %a, i32 %b) nounwind {
149 %vecins = insertelement <2 x i8> %a, i8 5, i32 %b
150 store <2 x i8> %vecins, <2 x i8> addrspace(1)* %out, align 8
154 ; SI-LABEL: {{^}}dynamic_insertelement_v4i8:
155 ; FIXMESI: buffer_store_dword
156 define void @dynamic_insertelement_v4i8(<4 x i8> addrspace(1)* %out, <4 x i8> %a, i32 %b) nounwind {
157 %vecins = insertelement <4 x i8> %a, i8 5, i32 %b
158 store <4 x i8> %vecins, <4 x i8> addrspace(1)* %out, align 16
162 ; SI-LABEL: {{^}}dynamic_insertelement_v8i8:
163 ; FIXMESI: buffer_store_dwordx2
164 define void @dynamic_insertelement_v8i8(<8 x i8> addrspace(1)* %out, <8 x i8> %a, i32 %b) nounwind {
165 %vecins = insertelement <8 x i8> %a, i8 5, i32 %b
166 store <8 x i8> %vecins, <8 x i8> addrspace(1)* %out, align 16
170 ; SI-LABEL: {{^}}dynamic_insertelement_v16i8:
171 ; FIXMESI: buffer_store_dwordx4
172 define void @dynamic_insertelement_v16i8(<16 x i8> addrspace(1)* %out, <16 x i8> %a, i32 %b) nounwind {
173 %vecins = insertelement <16 x i8> %a, i8 5, i32 %b
174 store <16 x i8> %vecins, <16 x i8> addrspace(1)* %out, align 16
178 ; This test requires handling INSERT_SUBREG in SIFixSGPRCopies. Check that
179 ; the compiler doesn't crash.
180 ; SI-LABEL: {{^}}insert_split_bb:
181 define void @insert_split_bb(<2 x i32> addrspace(1)* %out, i32 addrspace(1)* %in, i32 %a, i32 %b) {
183 %0 = insertelement <2 x i32> undef, i32 %a, i32 0
184 %1 = icmp eq i32 %a, 0
185 br i1 %1, label %if, label %else
188 %2 = load i32, i32 addrspace(1)* %in
189 %3 = insertelement <2 x i32> %0, i32 %2, i32 1
193 %4 = getelementptr i32, i32 addrspace(1)* %in, i32 1
194 %5 = load i32, i32 addrspace(1)* %4
195 %6 = insertelement <2 x i32> %0, i32 %5, i32 1
199 %7 = phi <2 x i32> [%3, %if], [%6, %else]
200 store <2 x i32> %7, <2 x i32> addrspace(1)* %out
204 ; SI-LABEL: {{^}}dynamic_insertelement_v2f64:
205 ; SI: buffer_store_dwordx2
206 ; SI: buffer_store_dwordx2
207 ; SI: buffer_store_dwordx2
208 ; SI: buffer_store_dwordx2
210 define void @dynamic_insertelement_v2f64(<2 x double> addrspace(1)* %out, <2 x double> %a, i32 %b) nounwind {
211 %vecins = insertelement <2 x double> %a, double 8.0, i32 %b
212 store <2 x double> %vecins, <2 x double> addrspace(1)* %out, align 16
216 ; SI-LABEL: {{^}}dynamic_insertelement_v2i64:
217 ; SI: buffer_store_dwordx2
218 ; SI: buffer_store_dwordx2
220 define void @dynamic_insertelement_v2i64(<2 x i64> addrspace(1)* %out, <2 x i64> %a, i32 %b) nounwind {
221 %vecins = insertelement <2 x i64> %a, i64 5, i32 %b
222 store <2 x i64> %vecins, <2 x i64> addrspace(1)* %out, align 8
226 ; SI-LABEL: {{^}}dynamic_insertelement_v4f64:
227 ; SI: buffer_store_dwordx2
228 ; SI: buffer_store_dwordx2
229 ; SI: buffer_store_dwordx2
230 ; SI: buffer_store_dwordx2
232 define void @dynamic_insertelement_v4f64(<4 x double> addrspace(1)* %out, <4 x double> %a, i32 %b) nounwind {
233 %vecins = insertelement <4 x double> %a, double 8.0, i32 %b
234 store <4 x double> %vecins, <4 x double> addrspace(1)* %out, align 16
238 ; SI-LABEL: {{^}}dynamic_insertelement_v8f64:
239 ; SI: buffer_store_dwordx2
240 ; SI: buffer_store_dwordx2
241 ; SI: buffer_store_dwordx2
242 ; SI: buffer_store_dwordx2
243 ; SI: buffer_store_dwordx2
244 ; SI: buffer_store_dwordx2
245 ; SI: buffer_store_dwordx2
246 ; SI: buffer_store_dwordx2
248 define void @dynamic_insertelement_v8f64(<8 x double> addrspace(1)* %out, <8 x double> %a, i32 %b) nounwind {
249 %vecins = insertelement <8 x double> %a, double 8.0, i32 %b
250 store <8 x double> %vecins, <8 x double> addrspace(1)* %out, align 16