1 ;RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=EG %s
2 ;RUN: llc < %s -march=r600 -mcpu=verde -verify-machineinstrs | FileCheck --check-prefix=SI %s
4 ; EG-LABEL: {{^}}or_v2i32:
5 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
6 ; EG: OR_INT {{\*? *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
8 ; SI-LABEL: {{^}}or_v2i32:
9 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
10 ; 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 ; EG-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-LABEL: {{^}}or_v4i32:
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 ; SI: v_or_b32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
33 define void @or_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) {
34 %b_ptr = getelementptr <4 x i32> addrspace(1)* %in, i32 1
35 %a = load <4 x i32> addrspace(1) * %in
36 %b = load <4 x i32> addrspace(1) * %b_ptr
37 %result = or <4 x i32> %a, %b
38 store <4 x i32> %result, <4 x i32> addrspace(1)* %out
42 ; SI-LABEL: {{^}}scalar_or_i32:
44 define void @scalar_or_i32(i32 addrspace(1)* %out, i32 %a, i32 %b) {
46 store i32 %or, i32 addrspace(1)* %out
50 ; SI-LABEL: {{^}}vector_or_i32:
51 ; SI: v_or_b32_e32 v{{[0-9]}}
52 define void @vector_or_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %a, i32 %b) {
53 %loada = load i32 addrspace(1)* %a
54 %or = or i32 %loada, %b
55 store i32 %or, i32 addrspace(1)* %out
59 ; SI-LABEL: {{^}}scalar_or_literal_i32:
60 ; SI: s_or_b32 s{{[0-9]+}}, s{{[0-9]+}}, 0x1869f
61 define void @scalar_or_literal_i32(i32 addrspace(1)* %out, i32 %a) {
62 %or = or i32 %a, 99999
63 store i32 %or, i32 addrspace(1)* %out, align 4
67 ; SI-LABEL: {{^}}vector_or_literal_i32:
68 ; SI: v_or_b32_e32 v{{[0-9]+}}, 0xffff, v{{[0-9]+}}
69 define void @vector_or_literal_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %a, i32 addrspace(1)* %b) {
70 %loada = load i32 addrspace(1)* %a, align 4
71 %or = or i32 %loada, 65535
72 store i32 %or, i32 addrspace(1)* %out, align 4
76 ; SI-LABEL: {{^}}vector_or_inline_immediate_i32:
77 ; SI: v_or_b32_e32 v{{[0-9]+}}, 4, v{{[0-9]+}}
78 define void @vector_or_inline_immediate_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %a, i32 addrspace(1)* %b) {
79 %loada = load i32 addrspace(1)* %a, align 4
80 %or = or i32 %loada, 4
81 store i32 %or, i32 addrspace(1)* %out, align 4
85 ; EG-LABEL: {{^}}scalar_or_i64:
86 ; EG-DAG: OR_INT * T{{[0-9]\.[XYZW]}}, KC0[2].W, KC0[3].Y
87 ; EG-DAG: OR_INT * T{{[0-9]\.[XYZW]}}, KC0[3].X, KC0[3].Z
88 ; SI-LABEL: {{^}}scalar_or_i64:
90 define void @scalar_or_i64(i64 addrspace(1)* %out, i64 %a, i64 %b) {
92 store i64 %or, i64 addrspace(1)* %out
96 ; SI-LABEL: {{^}}vector_or_i64:
97 ; SI: v_or_b32_e32 v{{[0-9]}}
98 ; SI: v_or_b32_e32 v{{[0-9]}}
99 define void @vector_or_i64(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 addrspace(1)* %b) {
100 %loada = load i64 addrspace(1)* %a, align 8
101 %loadb = load i64 addrspace(1)* %a, align 8
102 %or = or i64 %loada, %loadb
103 store i64 %or, i64 addrspace(1)* %out
107 ; SI-LABEL: {{^}}scalar_vector_or_i64:
108 ; SI: v_or_b32_e32 v{{[0-9]}}
109 ; SI: v_or_b32_e32 v{{[0-9]}}
110 define void @scalar_vector_or_i64(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 %b) {
111 %loada = load i64 addrspace(1)* %a
112 %or = or i64 %loada, %b
113 store i64 %or, i64 addrspace(1)* %out
117 ; SI-LABEL: {{^}}vector_or_i64_loadimm:
118 ; SI-DAG: s_mov_b32 [[LO_S_IMM:s[0-9]+]], 0xdf77987f
119 ; SI-DAG: s_mov_b32 [[HI_S_IMM:s[0-9]+]], 0x146f
120 ; SI-DAG: buffer_load_dwordx2 v{{\[}}[[LO_VREG:[0-9]+]]:[[HI_VREG:[0-9]+]]{{\]}},
121 ; SI-DAG: v_or_b32_e32 {{v[0-9]+}}, [[LO_S_IMM]], v[[LO_VREG]]
122 ; SI-DAG: v_or_b32_e32 {{v[0-9]+}}, [[HI_S_IMM]], v[[HI_VREG]]
124 define void @vector_or_i64_loadimm(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 addrspace(1)* %b) {
125 %loada = load i64 addrspace(1)* %a, align 8
126 %or = or i64 %loada, 22470723082367
127 store i64 %or, i64 addrspace(1)* %out
131 ; FIXME: The or 0 should really be removed.
132 ; SI-LABEL: {{^}}vector_or_i64_imm:
133 ; SI: buffer_load_dwordx2 v{{\[}}[[LO_VREG:[0-9]+]]:[[HI_VREG:[0-9]+]]{{\]}},
134 ; SI: v_or_b32_e32 {{v[0-9]+}}, 8, v[[LO_VREG]]
135 ; SI: v_or_b32_e32 {{v[0-9]+}}, 0, {{.*}}
137 define void @vector_or_i64_imm(i64 addrspace(1)* %out, i64 addrspace(1)* %a, i64 addrspace(1)* %b) {
138 %loada = load i64 addrspace(1)* %a, align 8
139 %or = or i64 %loada, 8
140 store i64 %or, i64 addrspace(1)* %out
144 ; SI-LABEL: {{^}}trunc_i64_or_to_i32:
145 ; SI: s_load_dword s[[SREG0:[0-9]+]]
146 ; SI: s_load_dword s[[SREG1:[0-9]+]]
147 ; SI: s_or_b32 s[[SRESULT:[0-9]+]], s[[SREG1]], s[[SREG0]]
148 ; SI: v_mov_b32_e32 [[VRESULT:v[0-9]+]], s[[SRESULT]]
149 ; SI: buffer_store_dword [[VRESULT]],
150 define void @trunc_i64_or_to_i32(i32 addrspace(1)* %out, i64 %a, i64 %b) {
152 %trunc = trunc i64 %add to i32
153 store i32 %trunc, i32 addrspace(1)* %out, align 8
157 ; EG-CHECK: {{^}}or_i1:
158 ; EG-CHECK: OR_INT {{\** *}}T{{[0-9]+\.[XYZW], PV\.[XYZW], PS}}
160 ; SI-CHECK: {{^}}or_i1:
161 ; SI-CHECK: s_or_b64 s[{{[0-9]+:[0-9]+}}], s[{{[0-9]+:[0-9]+}}], s[{{[0-9]+:[0-9]+}}]
162 define void @or_i1(float addrspace(1)* %out, float addrspace(1)* %in0, float addrspace(1)* %in1) {
163 %a = load float addrspace(1) * %in0
164 %b = load float addrspace(1) * %in1
165 %acmp = fcmp oge float %a, 0.000000e+00
166 %bcmp = fcmp oge float %b, 0.000000e+00
167 %or = or i1 %acmp, %bcmp
168 %result = select i1 %or, float %a, float %b
169 store float %result, float addrspace(1)* %out