diff options
Diffstat (limited to 'test/CodeGen/R600/llvm.AMDGPU.div_fmas.ll')
| -rw-r--r-- | test/CodeGen/R600/llvm.AMDGPU.div_fmas.ll | 48 |
1 files changed, 24 insertions, 24 deletions
diff --git a/test/CodeGen/R600/llvm.AMDGPU.div_fmas.ll b/test/CodeGen/R600/llvm.AMDGPU.div_fmas.ll index 239fd53..bcb7f87 100644 --- a/test/CodeGen/R600/llvm.AMDGPU.div_fmas.ll +++ b/test/CodeGen/R600/llvm.AMDGPU.div_fmas.ll @@ -78,7 +78,7 @@ define void @test_div_fmas_f64(double addrspace(1)* %out, double %a, double %b, } ; GCN-LABEL: {{^}}test_div_fmas_f32_cond_to_vcc: -; SI: v_cmp_eq_i32_e64 vcc, s{{[0-9]+}}, 0 +; SI: v_cmp_eq_i32_e64 vcc, 0, s{{[0-9]+}} ; SI: v_div_fmas_f32 {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}} define void @test_div_fmas_f32_cond_to_vcc(float addrspace(1)* %out, float %a, float %b, float %c, i32 %i) nounwind { %cmp = icmp eq i32 %i, 0 @@ -110,21 +110,21 @@ define void @test_div_fmas_f32_imm_true_cond_to_vcc(float addrspace(1)* %out, fl ; SI-DAG: buffer_load_dword [[B:v[0-9]+]], {{v\[[0-9]+:[0-9]+\]}}, {{s\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:4{{$}} ; SI-DAG: buffer_load_dword [[C:v[0-9]+]], {{v\[[0-9]+:[0-9]+\]}}, {{s\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}} -; SI-DAG: v_cmp_eq_i32_e64 [[CMP0:s\[[0-9]+:[0-9]+\]]], v{{[0-9]+}}, 0 -; SI-DAG: v_cmp_ne_i32_e64 [[CMP1:s\[[0-9]+:[0-9]+\]]], s{{[0-9]+}}, 0 +; SI-DAG: v_cmp_eq_i32_e32 [[CMP0:vcc]], 0, v{{[0-9]+}} +; SI-DAG: v_cmp_ne_i32_e64 [[CMP1:s\[[0-9]+:[0-9]+\]]], 0, s{{[0-9]+}} ; SI: s_and_b64 vcc, [[CMP0]], [[CMP1]] ; SI: v_div_fmas_f32 {{v[0-9]+}}, [[A]], [[B]], [[C]] ; SI: s_endpgm define void @test_div_fmas_f32_logical_cond_to_vcc(float addrspace(1)* %out, float addrspace(1)* %in, i32 %d) nounwind { %tid = call i32 @llvm.r600.read.tidig.x() nounwind readnone - %gep.a = getelementptr float addrspace(1)* %in, i32 %tid - %gep.b = getelementptr float addrspace(1)* %gep.a, i32 1 - %gep.c = getelementptr float addrspace(1)* %gep.a, i32 2 - %gep.out = getelementptr float addrspace(1)* %out, i32 2 + %gep.a = getelementptr float, float addrspace(1)* %in, i32 %tid + %gep.b = getelementptr float, float addrspace(1)* %gep.a, i32 1 + %gep.c = getelementptr float, float addrspace(1)* %gep.a, i32 2 + %gep.out = getelementptr float, float addrspace(1)* %out, i32 2 - %a = load float addrspace(1)* %gep.a - %b = load float addrspace(1)* %gep.b - %c = load float addrspace(1)* %gep.c + %a = load float, float addrspace(1)* %gep.a + %b = load float, float addrspace(1)* %gep.b + %c = load float, float addrspace(1)* %gep.c %cmp0 = icmp eq i32 %tid, 0 %cmp1 = icmp ne i32 %d, 0 @@ -136,17 +136,17 @@ define void @test_div_fmas_f32_logical_cond_to_vcc(float addrspace(1)* %out, flo } ; GCN-LABEL: {{^}}test_div_fmas_f32_i1_phi_vcc: -; SI: v_cmp_eq_i32_e64 [[CMPTID:s\[[0-9]+:[0-9]+\]]], v{{[0-9]+}}, 0 -; SI: s_and_saveexec_b64 [[CMPTID]], [[CMPTID]] -; SI: s_xor_b64 [[CMPTID]], exec, [[CMPTID]] +; SI: v_cmp_eq_i32_e32 vcc, 0, v{{[0-9]+}} +; SI: s_and_saveexec_b64 [[SAVE:s\[[0-9]+:[0-9]+\]]], vcc +; SI: s_xor_b64 [[SAVE]], exec, [[SAVE]] ; SI: buffer_load_dword [[LOAD:v[0-9]+]] -; SI: v_cmp_ne_i32_e64 [[CMPLOAD:s\[[0-9]+:[0-9]+\]]], [[LOAD]], 0 -; SI: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, [[CMPLOAD]] +; SI: v_cmp_ne_i32_e32 vcc, 0, [[LOAD]] +; SI: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc ; SI: BB9_2: -; SI: s_or_b64 exec, exec, [[CMPTID]] +; SI: s_or_b64 exec, exec, [[SAVE]] ; SI: v_cmp_ne_i32_e32 vcc, 0, v0 ; SI: v_div_fmas_f32 {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}} ; SI: buffer_store_dword @@ -154,20 +154,20 @@ define void @test_div_fmas_f32_logical_cond_to_vcc(float addrspace(1)* %out, flo define void @test_div_fmas_f32_i1_phi_vcc(float addrspace(1)* %out, float addrspace(1)* %in, i32 addrspace(1)* %dummy) nounwind { entry: %tid = call i32 @llvm.r600.read.tidig.x() nounwind readnone - %gep.out = getelementptr float addrspace(1)* %out, i32 2 - %gep.a = getelementptr float addrspace(1)* %in, i32 %tid - %gep.b = getelementptr float addrspace(1)* %gep.a, i32 1 - %gep.c = getelementptr float addrspace(1)* %gep.a, i32 2 + %gep.out = getelementptr float, float addrspace(1)* %out, i32 2 + %gep.a = getelementptr float, float addrspace(1)* %in, i32 %tid + %gep.b = getelementptr float, float addrspace(1)* %gep.a, i32 1 + %gep.c = getelementptr float, float addrspace(1)* %gep.a, i32 2 - %a = load float addrspace(1)* %gep.a - %b = load float addrspace(1)* %gep.b - %c = load float addrspace(1)* %gep.c + %a = load float, float addrspace(1)* %gep.a + %b = load float, float addrspace(1)* %gep.b + %c = load float, float addrspace(1)* %gep.c %cmp0 = icmp eq i32 %tid, 0 br i1 %cmp0, label %bb, label %exit bb: - %val = load i32 addrspace(1)* %dummy + %val = load i32, i32 addrspace(1)* %dummy %cmp1 = icmp ne i32 %val, 0 br label %exit |