AMDGPU: Regenerate baseline mir tests

llvmbot · Feb 27, 2024 · e7900e6 · e7900e6
1 parent b2a4f64
commit e7900e6
Show file tree

Hide file tree

Showing 109 changed files with 4,054 additions and 2,171 deletions.
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.f32-no-rtn.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.f32-no-rtn.ll
@@ -19,6 +19,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_offset_no_rtn(float %val, <4 x i32
   ; GFX908_GFX11-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offset_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
@@ -32,6 +33,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_offset_no_rtn(float %val, <4 x i32
   ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
   ; GFX12-LABEL: name: buffer_atomic_fadd_f32_offset_no_rtn
   ; GFX12: bb.1 (%ir-block.0):
   ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
@@ -64,6 +66,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_offen_no_rtn(float %val, <4 x i32>
   ; GFX908_GFX11-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
@@ -78,6 +81,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_offen_no_rtn(float %val, <4 x i32>
   ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
   ; GFX12-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn
   ; GFX12: bb.1 (%ir-block.0):
   ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
@@ -111,6 +115,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_idxen_no_rtn(float %val, <4 x i32>
   ; GFX908_GFX11-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
@@ -125,6 +130,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_idxen_no_rtn(float %val, <4 x i32>
   ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
   ; GFX12-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn
   ; GFX12: bb.1 (%ir-block.0):
   ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
@@ -160,6 +166,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_bothen_no_rtn(float %val, <4 x i32
   ; GFX908_GFX11-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
@@ -176,6 +183,7 @@ define amdgpu_ps void @buffer_atomic_fadd_f32_bothen_no_rtn(float %val, <4 x i32
   ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
   ; GFX12-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn
   ; GFX12: bb.1 (%ir-block.0):
   ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
@@ -210,6 +218,7 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_offset_no_rtn(float %val, ptr
   ; GFX908_GFX11-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_f32_offset_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
@@ -223,6 +232,20 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_offset_no_rtn(float %val, ptr
   ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
+  ; GFX12-LABEL: name: buffer_ptr_atomic_fadd_f32_offset_no_rtn
+  ; GFX12: bb.1 (%ir-block.0):
+  ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
+  ; GFX12-NEXT: {{  $}}
+  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
+  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
+  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
+  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
+  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
+  ; GFX12-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
+  ; GFX12-NEXT:   BUFFER_ATOMIC_ADD_F32_VBUFFER_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
+  ; GFX12-NEXT:   S_ENDPGM 0
   %ret = call float @llvm.amdgcn.raw.ptr.buffer.atomic.fadd.f32(float %val, ptr addrspace(8) %rsrc, i32 0, i32 %soffset, i32 0)
   ret void
 }
@@ -242,6 +265,7 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_offen_no_rtn(float %val, ptr a
   ; GFX908_GFX11-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_f32_offen_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
@@ -256,6 +280,21 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_offen_no_rtn(float %val, ptr a
   ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
+  ; GFX12-LABEL: name: buffer_ptr_atomic_fadd_f32_offen_no_rtn
+  ; GFX12: bb.1 (%ir-block.0):
+  ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
+  ; GFX12-NEXT: {{  $}}
+  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
+  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
+  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
+  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
+  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
+  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
+  ; GFX12-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
+  ; GFX12-NEXT:   BUFFER_ATOMIC_ADD_F32_VBUFFER_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
+  ; GFX12-NEXT:   S_ENDPGM 0
   %ret = call float @llvm.amdgcn.raw.ptr.buffer.atomic.fadd.f32(float %val, ptr addrspace(8) %rsrc, i32 %voffset, i32 %soffset, i32 0)
   ret void
 }
@@ -275,6 +314,7 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_idxen_no_rtn(float %val, ptr a
   ; GFX908_GFX11-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_f32_idxen_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
@@ -289,6 +329,21 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_idxen_no_rtn(float %val, ptr a
   ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
+  ; GFX12-LABEL: name: buffer_ptr_atomic_fadd_f32_idxen_no_rtn
+  ; GFX12: bb.1 (%ir-block.0):
+  ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
+  ; GFX12-NEXT: {{  $}}
+  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
+  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
+  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
+  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
+  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
+  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
+  ; GFX12-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
+  ; GFX12-NEXT:   BUFFER_ATOMIC_ADD_F32_VBUFFER_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
+  ; GFX12-NEXT:   S_ENDPGM 0
   %ret = call float @llvm.amdgcn.struct.ptr.buffer.atomic.fadd.f32(float %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
   ret void
 }
@@ -310,6 +365,7 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_bothen_no_rtn(float %val, ptr
   ; GFX908_GFX11-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
   ; GFX908_GFX11-NEXT:   BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX908_GFX11-NEXT:   S_ENDPGM 0
+  ;
   ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_f32_bothen_no_rtn
   ; GFX90A_GFX940: bb.1 (%ir-block.0):
   ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
@@ -326,6 +382,23 @@ define amdgpu_ps void @buffer_ptr_atomic_fadd_f32_bothen_no_rtn(float %val, ptr
   ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
   ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
   ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
+  ;
+  ; GFX12-LABEL: name: buffer_ptr_atomic_fadd_f32_bothen_no_rtn
+  ; GFX12: bb.1 (%ir-block.0):
+  ; GFX12-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
+  ; GFX12-NEXT: {{  $}}
+  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
+  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
+  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
+  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
+  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
+  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
+  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
+  ; GFX12-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
+  ; GFX12-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
+  ; GFX12-NEXT:   BUFFER_ATOMIC_ADD_F32_VBUFFER_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
+  ; GFX12-NEXT:   S_ENDPGM 0
   %ret = call float @llvm.amdgcn.struct.ptr.buffer.atomic.fadd.f32(float %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
   ret void
 }