1 ; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs< %s | FileCheck -check-prefixes=GCN,SICIVI,FUNC %s
2 ; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tonga -verify-machineinstrs< %s | FileCheck -check-prefixes=GCN,SICIVI,FUNC %s
3 ; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx900 -verify-machineinstrs< %s | FileCheck -check-prefixes=GCN,GFX9,FUNC %s
4 ; RUN: llc -amdgpu-scalarize-global-loads=false -march=r600 -mcpu=cypress -verify-machineinstrs< %s
6 declare { i32, i1 } @llvm.sadd.with.overflow.i32(i32, i32) nounwind readnone
7 declare { i64, i1 } @llvm.sadd.with.overflow.i64(i64, i64) nounwind readnone
9 ; FUNC-LABEL: {{^}}saddo_i64_zext:
10 define amdgpu_kernel void @saddo_i64_zext(i64 addrspace(1)* %out, i64 %a, i64 %b) nounwind {
11 %sadd = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 %a, i64 %b) nounwind
12 %val = extractvalue { i64, i1 } %sadd, 0
13 %carry = extractvalue { i64, i1 } %sadd, 1
14 %ext = zext i1 %carry to i64
15 %add2 = add i64 %val, %ext
16 store i64 %add2, i64 addrspace(1)* %out, align 8
20 ; FUNC-LABEL: {{^}}s_saddo_i32:
21 define amdgpu_kernel void @s_saddo_i32(i32 addrspace(1)* %out, i1 addrspace(1)* %carryout, i32 %a, i32 %b) nounwind {
22 %sadd = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %a, i32 %b) nounwind
23 %val = extractvalue { i32, i1 } %sadd, 0
24 %carry = extractvalue { i32, i1 } %sadd, 1
25 store i32 %val, i32 addrspace(1)* %out, align 4
26 store i1 %carry, i1 addrspace(1)* %carryout
30 ; FUNC-LABEL: {{^}}v_saddo_i32:
31 define amdgpu_kernel void @v_saddo_i32(i32 addrspace(1)* %out, i1 addrspace(1)* %carryout, i32 addrspace(1)* %aptr, i32 addrspace(1)* %bptr) nounwind {
32 %a = load i32, i32 addrspace(1)* %aptr, align 4
33 %b = load i32, i32 addrspace(1)* %bptr, align 4
34 %sadd = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %a, i32 %b) nounwind
35 %val = extractvalue { i32, i1 } %sadd, 0
36 %carry = extractvalue { i32, i1 } %sadd, 1
37 store i32 %val, i32 addrspace(1)* %out, align 4
38 store i1 %carry, i1 addrspace(1)* %carryout
42 ; FUNC-LABEL: {{^}}s_saddo_i64:
43 define amdgpu_kernel void @s_saddo_i64(i64 addrspace(1)* %out, i1 addrspace(1)* %carryout, i64 %a, i64 %b) nounwind {
44 %sadd = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 %a, i64 %b) nounwind
45 %val = extractvalue { i64, i1 } %sadd, 0
46 %carry = extractvalue { i64, i1 } %sadd, 1
47 store i64 %val, i64 addrspace(1)* %out, align 8
48 store i1 %carry, i1 addrspace(1)* %carryout
52 ; FUNC-LABEL: {{^}}v_saddo_i64:
53 ; SICIVI: v_add_{{[iu]}}32_e32 v{{[0-9]+}}, vcc
54 ; SICIVI: v_addc_u32_e32 v{{[0-9]+}}, vcc
56 ; GFX9: v_add_co_u32_e32 v{{[0-9]+}}, vcc
57 ; GFX9: v_addc_co_u32_e32 v{{[0-9]+}}, vcc
58 define amdgpu_kernel void @v_saddo_i64(i64 addrspace(1)* %out, i1 addrspace(1)* %carryout, i64 addrspace(1)* %aptr, i64 addrspace(1)* %bptr) nounwind {
59 %a = load i64, i64 addrspace(1)* %aptr, align 4
60 %b = load i64, i64 addrspace(1)* %bptr, align 4
61 %sadd = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 %a, i64 %b) nounwind
62 %val = extractvalue { i64, i1 } %sadd, 0
63 %carry = extractvalue { i64, i1 } %sadd, 1
64 store i64 %val, i64 addrspace(1)* %out, align 8
65 store i1 %carry, i1 addrspace(1)* %carryout