1 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
2 ; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
3 ; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck -check-prefix=EG -check-prefix=FUNC %s
6 ; FUNC-LABEL: {{^}}or_v2i32:
7 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
8 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
10 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
11 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
12 define void @or_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %in) {
13 %b_ptr = getelementptr <2 x i32> addrspace(1)* %in, i32 1
14 %a = load <2 x i32> addrspace(1) * %in
15 %b = load <2 x i32> addrspace(1) * %b_ptr
16 %result = or <2 x i32> %a, %b
17 store <2 x i32> %result, <2 x i32> addrspace(1)* %out
21 ; FUNC-LABEL: {{^}}or_v4i32:
22 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
23 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
24 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
25 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
27 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
28 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
29 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
30 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
31 define void @or_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) {
32 %b_ptr = getelementptr <4 x i32> addrspace(1)* %in, i32 1
33 %a = load <4 x i32> addrspace(1) * %in
34 %b = load <4 x i32> addrspace(1) * %b_ptr
35 %result = or <4 x i32> %a, %b
36 store <4 x i32> %result, <4 x i32> addrspace(1)* %out
40 ; FUNC-LABEL: {{^}}scalar_or_i32:
42 define void @scalar_or_i32(i32 addrspace(1)* %out, i32 %a, i32 %b) {
44 store i32 %or, i32 addrspace(1)* %out
48 ; FUNC-LABEL: {{^}}vector_or_i32:
49 ; SI: v_or_b32_e32 v{{[0-9]}}
50 define void @vector_or_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %a, i32 %b) {
51 %loada = load i32 addrspace(1)* %a
52 %or = or i32 %loada, %b
53 store i32 %or, i32 addrspace(1)* %out
57 ; FUNC-LABEL: {{^}}scalar_or_literal_i32:
58 ; SI: s_or_b32 s{{[0-9]+}}, s{{[0-9]+}}, 0x1869f
59 define void @scalar_or_literal_i32(i32 addrspace(1)* %out, i32 %a) {
60 %or = or i32 %a, 99999
61 store i32 %or, i32 addrspace(1)* %out, align 4
65 ; FUNC-LABEL: {{^}}vector_or_literal_i32:
66 ; SI: v_or_b32_e32 v{{[0-9]+}}, 0xffff, v{{[0-9]+}}
67 define void @vector_or_literal_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %a, i32 addrspace(1)* %b) {
68 %loada = load i32 addrspace(1)* %a, align 4
69 %or = or i32 %loada, 65535
70 store i32 %or, i32 addrspace(1)* %out, align 4
74 ; FUNC-LABEL: {{^}}vector_or_inline_immediate_i32:
75 ; SI: v_or_b32_e32 v{{[0-9]+}}, 4, v{{[0-9]+}}
76 define void @vector_or_inline_immediate_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %a, i32 addrspace(1)* %b) {
77 %loada = load i32 addrspace(1)* %a, align 4
78 %or = or i32 %loada, 4
79 store i32 %or, i32 addrspace(1)* %out, align 4
83 ; FUNC-LABEL: {{^}}scalar_or_i64:
84 ; EG-DAG: OR_INT * T{{[0-9]\.[XYZW]}}, KC0[2].W, KC0[3].Y
85 ; EG-DAG: OR_INT * T{{[0-9]\.[XYZW]}}, KC0[3].X, KC0[3].Z
88 define void @scalar_or_i64(i64 addrspace(1)* %out, i64 %a, i64 %b) {
90 store i64 %or, i64 addrspace(1)* %out
94 ; FUNC-LABEL: {{^}}vector_or_i64:
95 ; SI: v_or_b32_e32 v{{[0-9]}}
96 ; SI: v_or_b32_e32 v{{[0-9]}}
97 define void @vector_or_i64(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 addrspace(1)* %b) {
98 %loada = load i64 addrspace(1)* %a, align 8
99 %loadb = load i64 addrspace(1)* %a, align 8
100 %or = or i64 %loada, %loadb
101 store i64 %or, i64 addrspace(1)* %out
105 ; FUNC-LABEL: {{^}}scalar_vector_or_i64:
106 ; SI: v_or_b32_e32 v{{[0-9]}}
107 ; SI: v_or_b32_e32 v{{[0-9]}}
108 define void @scalar_vector_or_i64(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 %b) {
109 %loada = load i64 addrspace(1)* %a
110 %or = or i64 %loada, %b
111 store i64 %or, i64 addrspace(1)* %out
115 ; FUNC-LABEL: {{^}}vector_or_i64_loadimm:
116 ; SI-DAG: s_mov_b32 [[LO_S_IMM:s[0-9]+]], 0xdf77987f
117 ; SI-DAG: s_movk_i32 [[HI_S_IMM:s[0-9]+]], 0x146f
118 ; SI-DAG: buffer_load_dwordx2 v{{\[}}[[LO_VREG:[0-9]+]]:[[HI_VREG:[0-9]+]]{{\]}},
119 ; SI-DAG: v_or_b32_e32 {{v[0-9]+}}, [[LO_S_IMM]], v[[LO_VREG]]
120 ; SI-DAG: v_or_b32_e32 {{v[0-9]+}}, [[HI_S_IMM]], v[[HI_VREG]]
122 define void @vector_or_i64_loadimm(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 addrspace(1)* %b) {
123 %loada = load i64 addrspace(1)* %a, align 8
124 %or = or i64 %loada, 22470723082367
125 store i64 %or, i64 addrspace(1)* %out
129 ; FIXME: The or 0 should really be removed.
130 ; FUNC-LABEL: {{^}}vector_or_i64_imm:
131 ; SI: buffer_load_dwordx2 v{{\[}}[[LO_VREG:[0-9]+]]:[[HI_VREG:[0-9]+]]{{\]}},
132 ; SI: v_or_b32_e32 {{v[0-9]+}}, 8, v[[LO_VREG]]
133 ; SI: v_or_b32_e32 {{v[0-9]+}}, 0, {{.*}}
135 define void @vector_or_i64_imm(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 addrspace(1)* %b) {
136 %loada = load i64 addrspace(1)* %a, align 8
137 %or = or i64 %loada, 8
138 store i64 %or, i64 addrspace(1)* %out
142 ; FUNC-LABEL: {{^}}trunc_i64_or_to_i32:
143 ; SI: s_load_dword s[[SREG0:[0-9]+]]
144 ; SI: s_load_dword s[[SREG1:[0-9]+]]
145 ; SI: s_or_b32 s[[SRESULT:[0-9]+]], s[[SREG1]], s[[SREG0]]
146 ; SI: v_mov_b32_e32 [[VRESULT:v[0-9]+]], s[[SRESULT]]
147 ; SI: buffer_store_dword [[VRESULT]],
148 define void @trunc_i64_or_to_i32(i32 addrspace(1)* %out, i64 %a, i64 %b) {
150 %trunc = trunc i64 %add to i32
151 store i32 %trunc, i32 addrspace(1)* %out, align 8
155 ; FUNC-LABEL: {{^}}or_i1:
156 ; EG: OR_INT {{\** *}}T{{[0-9]+\.[XYZW], PV\.[XYZW], PS}}
158 ; SI: s_or_b64 s[{{[0-9]+:[0-9]+}}], s[{{[0-9]+:[0-9]+}}], s[{{[0-9]+:[0-9]+}}]
159 define void @or_i1(float addrspace(1)* %out, float addrspace(1)* %in0, float addrspace(1)* %in1) {
160 %a = load float addrspace(1)* %in0
161 %b = load float addrspace(1)* %in1
162 %acmp = fcmp oge float %a, 0.000000e+00
163 %bcmp = fcmp oge float %b, 0.000000e+00
164 %or = or i1 %acmp, %bcmp
165 %result = select i1 %or, float %a, float %b
166 store float %result, float addrspace(1)* %out
170 ; FUNC-LABEL: {{^}}s_or_i1:
171 ; SI: s_or_b64 s[{{[0-9]+:[0-9]+}}], vcc, s[{{[0-9]+:[0-9]+}}]
172 define void @s_or_i1(i1 addrspace(1)* %out, i32 %a, i32 %b, i32 %c, i32 %d) {
173 %cmp0 = icmp eq i32 %a, %b
174 %cmp1 = icmp eq i32 %c, %d
175 %or = or i1 %cmp0, %cmp1
176 store i1 %or, i1 addrspace(1)* %out