[AMDGPU] Mark cluster_workgroup_id_* intrinsics always uniform

Author: rampitec

llvm/lib/Target/AMDGPU/AMDGPUSearchableTables.td

@@ -398,6 +398,14 @@ def : AlwaysUniform<int_amdgcn_icmp>;
 def : AlwaysUniform<int_amdgcn_fcmp>;
 def : AlwaysUniform<int_amdgcn_ballot>;
 def : AlwaysUniform<int_amdgcn_if_break>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_id_x>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_id_y>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_id_z>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_flat_id>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_max_id_x>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_max_id_y>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_max_id_z>;
+def : AlwaysUniform<int_amdgcn_cluster_workgroup_max_flat_id>;
 def : AlwaysUniform<int_amdgcn_workgroup_id_x>;
 def : AlwaysUniform<int_amdgcn_workgroup_id_y>;
 def : AlwaysUniform<int_amdgcn_workgroup_id_z>;

llvm/test/Analysis/UniformityAnalysis/AMDGPU/always_uniform.ll

@@ -119,6 +119,70 @@ define void @s_getreg(ptr addrspace(1) inreg %out) {
   ret void
 }
 
+; CHECK-LABEL: for function 'cluster_workgroup_id_x':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_id_x(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.id.x()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_id_y':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_id_y(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.id.y()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_id_z':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_id_z(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.id.z()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_flat_id':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_flat_id(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.flat.id()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_max_id_x':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_max_id_x(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.max.id.x()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_max_id_y':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_max_id_y(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.max.id.y()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_max_id_z':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_max_id_z(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.max.id.z()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; CHECK-LABEL: for function 'cluster_workgroup_max_flat_id':
+; CHECK: ALL VALUES UNIFORM
+define void @cluster_workgroup_max_flat_id(ptr addrspace(1) inreg %out) {
+  %result = call i32 @llvm.amdgcn.cluster.workgroup.max.flat.id()
+  store i32 %result, ptr addrspace(1) %out, align 4
+  ret void
+}
+
 ; CHECK-LABEL: for function 's_memtime':
 ; CHECK: ALL VALUES UNIFORM
 define void @s_memtime(ptr addrspace(1) inreg %out) {
@@ -144,6 +208,14 @@ declare i64 @llvm.amdgcn.ballot.i32(i1) #1
 declare i32 @llvm.amdgcn.workgroup.id.x() #0
 declare i32 @llvm.amdgcn.workgroup.id.y() #0
 declare i32 @llvm.amdgcn.workgroup.id.z() #0
+declare i32 @llvm.amdgcn.cluster.workgroup.id.x()
+declare i32 @llvm.amdgcn.cluster.workgroup.id.y()
+declare i32 @llvm.amdgcn.cluster.workgroup.id.z()
+declare i32 @llvm.amdgcn.cluster.workgroup.flat.id()
+declare i32 @llvm.amdgcn.cluster.workgroup.max.id.x()
+declare i32 @llvm.amdgcn.cluster.workgroup.max.id.y()
+declare i32 @llvm.amdgcn.cluster.workgroup.max.id.z()
+declare i32 @llvm.amdgcn.cluster.workgroup.max.flat.id()
 
 attributes #0 = { nounwind readnone }
 attributes #1 = { nounwind readnone convergent }