1 ; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,FUNC %s
2 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,FUNC %s
3 ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,VIGFX9,FUNC %s
4 ; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,VIGFX9,FUNC %s
6 ; FUNC-LABEL: {{^}}ds_ordered_swap:
7 ; GCN: s_mov_b32 m0, s0
9 ; GCN-NEXT: ds_ordered_count v{{[0-9]+}}, v0 offset:4868 gds
10 ; GCN-NEXT: s_waitcnt expcnt(0) lgkmcnt(0)
11 define amdgpu_cs float @ds_ordered_swap(i32 addrspace(2)* inreg %gds, i32 %value) {
12 %val = call i32@llvm.amdgcn.ds.ordered.swap(i32 addrspace(2)* %gds, i32 %value, i32 0, i32 0, i1 false, i32 1, i1 true, i1 true)
13 %r = bitcast i32 %val to float
17 ; FUNC-LABEL: {{^}}ds_ordered_swap_conditional:
18 ; GCN: v_cmp_ne_u32_e32 vcc, 0, v0
19 ; GCN: s_and_saveexec_b64 s[[SAVED:\[[0-9]+:[0-9]+\]]], vcc
20 ; // We have to use s_cbranch, because ds_ordered_count has side effects with EXEC=0
21 ; GCN: s_cbranch_execz [[BB:BB._.]]
22 ; GCN: s_mov_b32 m0, s0
23 ; VIGFX9-NEXT: s_nop 0
24 ; GCN-NEXT: ds_ordered_count v{{[0-9]+}}, v0 offset:4868 gds
26 ; // Wait for expcnt(0) before modifying EXEC
27 ; GCN-NEXT: s_waitcnt expcnt(0)
28 ; GCN-NEXT: s_or_b64 exec, exec, s[[SAVED]]
29 ; GCN-NEXT: s_waitcnt lgkmcnt(0)
30 define amdgpu_cs float @ds_ordered_swap_conditional(i32 addrspace(2)* inreg %gds, i32 %value) {
32 %c = icmp ne i32 %value, 0
33 br i1 %c, label %if-true, label %endif
36 %val = call i32@llvm.amdgcn.ds.ordered.swap(i32 addrspace(2)* %gds, i32 %value, i32 0, i32 0, i1 false, i32 1, i1 true, i1 true)
40 %v = phi i32 [ %val, %if-true ], [ undef, %entry ]
41 %r = bitcast i32 %v to float
45 declare i32 @llvm.amdgcn.ds.ordered.swap(i32 addrspace(2)* nocapture, i32, i32, i32, i1, i32, i1, i1)