AMDGPU: Fix SILoadStoreOptimizer for gfx90a

This was hardcoding the register class to use for the newly created
pointer registers, violating the aligned VGPR requirement.

diff --git a/llvm/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp b/llvm/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
index 957f702..493c1ad 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
+++ b/llvm/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
@@ -1716,7 +1716,7 @@ Register SILoadStoreOptimizer::computeBase(MachineInstr &MI,
   (void)HiHalf;
   LLVM_DEBUG(dbgs() << "    "; HiHalf->dump(););
 
-  Register FullDestReg = MRI->createVirtualRegister(&AMDGPU::VReg_64RegClass);
+  Register FullDestReg = MRI->createVirtualRegister(TRI->getVGPR64Class());
   MachineInstr *FullBase =
     BuildMI(*MBB, MBBI, DL, TII->get(TargetOpcode::REG_SEQUENCE), FullDestReg)
       .addReg(DestSub0)

diff --git a/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm-gfx90a.mir b/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm-gfx90a.mir
new file mode 100644 (file)
index 0000000..64c5bc5
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm-gfx90a.mir
@@ -0,0 +1,221 @@
+# RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -run-pass si-load-store-opt -o - %s | FileCheck -check-prefix=GFX9 %s
+
+# GFX9-LABEL: name: diffoporder_add
+
+# GFX9: %{{[0-9]+}}:vreg_64_align2 = REG_SEQUENCE
+
+# GFX9: S_MOV_B32 6144
+# GFX9-NEXT: V_ADD_CO_U32
+# GFX9-NEXT: V_ADDC_U32
+# GFX9-NEXT: [[PTR0:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE
+# GFX9-NEXT: %{{[0-9]+}}:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[PTR0]], -2048, 0
+# GFX9: %{{[0-9]+}}:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[PTR0]], 0, 0
+
+name: diffoporder_add
+body:             |
+  bb.0.entry:
+    %0:sgpr_64 = COPY $sgpr0_sgpr1
+    %1:sreg_64_xexec = S_LOAD_DWORDX2_IMM %0, 36, 0
+    %3:sgpr_128 = COPY $sgpr96_sgpr97_sgpr98_sgpr99
+    %4:sreg_32_xm0 = COPY $sgpr101
+    %5:sreg_32_xm0 = S_MOV_B32 0
+    $sgpr0_sgpr1_sgpr2_sgpr3 = COPY %3
+    $sgpr4 = COPY %4
+    $vgpr0 = V_MOV_B32_e32 0, implicit $exec
+    %6:vreg_64_align2 = COPY $vgpr0_vgpr1
+    %7:vgpr_32 = V_AND_B32_e32 255, %6.sub0, implicit $exec
+    %8:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+    %9:vreg_64_align2 = REG_SEQUENCE killed %7, %subreg.sub0, %8, %subreg.sub1
+    %10:vgpr_32 = V_LSHLREV_B32_e64 7, %6.sub0, implicit $exec
+    %11:vgpr_32 = V_AND_B32_e32 -32768, killed %10, implicit $exec
+    %12:sgpr_32 = COPY %1.sub1
+    %13:vgpr_32 = COPY %5
+    %14:vgpr_32, %15:sreg_64_xexec = V_ADD_CO_U32_e64 %1.sub0, %11, 0, implicit $exec
+    %16:vgpr_32 = COPY %12
+    %17:vgpr_32, dead %18:sreg_64_xexec = V_ADDC_U32_e64 %16, %13, killed %15, 0, implicit $exec
+    %19:vreg_64_align2 = REG_SEQUENCE %14, %subreg.sub0, %17, %subreg.sub1
+    %20:vreg_64_align2 = V_LSHLREV_B64_e64 3, %9, implicit $exec
+    %21:vgpr_32, %22:sreg_64_xexec = V_ADD_CO_U32_e64 %14, %20.sub0, 0, implicit $exec
+    %23:vgpr_32, dead %24:sreg_64_xexec = V_ADDC_U32_e64 %17, %20.sub1, killed %22, 0, implicit $exec
+    %25:sgpr_32 = S_MOV_B32 4096
+    %26:vgpr_32, %27:sreg_64_xexec = V_ADD_CO_U32_e64 %25, %21, 0, implicit $exec
+    %28:vgpr_32, dead %29:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %27, 0, implicit $exec
+    %30:vreg_64_align2 = REG_SEQUENCE %26, %subreg.sub0, %28, %subreg.sub1
+    %31:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %30, 0, 0, implicit $exec
+    %32:sgpr_32 = S_MOV_B32 6144
+    %33:vgpr_32, %34:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %32, 0, implicit $exec
+    %35:vgpr_32, dead %36:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %34, 0, implicit $exec
+    %37:vreg_64_align2 = REG_SEQUENCE %33, %subreg.sub0, %35, %subreg.sub1
+    %38:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %37, 0, 0, implicit $exec
+...
+---
+
+# GFX9-LABEL: name: LowestInMiddle
+# GFX9: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 11200
+# GFX9: [[BASE_LO:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_5:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 %{{[0-9]+}}, [[S_MOV_B32_1]]
+# GFX9: [[BASE_HI:%[0-9]+]]:vgpr_32, dead %{{[0-9]+}}:sreg_64_xexec = V_ADDC_U32_e64 %{{[0-9]+}}, 0, killed [[V_ADD_CO_U32_e64_5]]
+# GFX9: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[BASE_LO]], %subreg.sub0, [[BASE_HI]], %subreg.sub1
+# GFX9: [[GLOBAL_LOAD_DWORDX2_:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[REG_SEQUENCE2]], -3200, 0
+#
+# GFX9: [[S_MOV_B32_2:%[0-9]+]]:sgpr_32 = S_MOV_B32 6400
+# GFX9: [[BASE1_LO:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_7:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 %{{[0-9]+}}, [[S_MOV_B32_2]]
+# GFX9: [[BASE1_HI:%[0-9]+]]:vgpr_32, dead %{{[0-9]+}}:sreg_64_xexec = V_ADDC_U32_e64 %{{[0-9]+}}, 0, killed [[V_ADD_CO_U32_e64_7]]
+# GFX9: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[BASE1_LO]], %subreg.sub0, [[BASE1_HI]], %subreg.sub1
+# GFX9: [[GLOBAL_LOAD_DWORDX2_1:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[REG_SEQUENCE3]], 0, 0,
+# GFX9: [[GLOBAL_LOAD_DWORDX2_2:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[REG_SEQUENCE2]], 0, 0,
+
+name: LowestInMiddle
+body:             |
+  bb.0.entry:
+    %0:sgpr_64 = COPY $sgpr0_sgpr1
+    %1:sreg_64_xexec = S_LOAD_DWORDX2_IMM %0, 36, 0
+    %3:sgpr_128 = COPY $sgpr96_sgpr97_sgpr98_sgpr99
+    %4:sreg_32_xm0 = COPY $sgpr101
+    %5:sreg_32_xm0 = S_MOV_B32 0
+    $sgpr0_sgpr1_sgpr2_sgpr3 = COPY %3
+    $sgpr4 = COPY %4
+    $vgpr0 = V_MOV_B32_e32 0, implicit $exec
+    %6:vreg_64_align2 = COPY $vgpr0_vgpr1
+    %7:vgpr_32 = V_AND_B32_e32 255, %6.sub0, implicit $exec
+    %8:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+    %9:vreg_64_align2 = REG_SEQUENCE killed %7, %subreg.sub0, %8, %subreg.sub1
+    %10:vgpr_32 = V_LSHLREV_B32_e64 7, %6.sub0, implicit $exec
+    %11:vgpr_32 = V_AND_B32_e32 -32768, killed %10, implicit $exec
+    %12:sgpr_32 = COPY %1.sub1
+    %13:vgpr_32 = COPY %5
+    %14:vgpr_32, %15:sreg_64_xexec = V_ADD_CO_U32_e64 %1.sub0, %11, 0, implicit $exec
+    %16:vgpr_32 = COPY %12
+    %17:vgpr_32, dead %18:sreg_64_xexec = V_ADDC_U32_e64 %16, %13, killed %15, 0, implicit $exec
+    %19:vreg_64_align2 = REG_SEQUENCE %14, %subreg.sub0, %17, %subreg.sub1
+    %20:vreg_64_align2 = V_LSHLREV_B64_e64 3, %9, implicit $exec
+    %21:vgpr_32, %22:sreg_64_xexec = V_ADD_CO_U32_e64 %14, %20.sub0, 0, implicit $exec
+    %23:vgpr_32, dead %24:sreg_64_xexec = V_ADDC_U32_e64 %17, %20.sub1, killed %22, 0, implicit $exec
+    %25:sgpr_32 = S_MOV_B32 8000
+    %26:vgpr_32, %27:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %25, 0, implicit $exec
+    %28:vgpr_32, dead %29:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %27, 0, implicit $exec
+    %30:vreg_64_align2 = REG_SEQUENCE %26, %subreg.sub0, %28, %subreg.sub1
+    %31:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %30, 0, 0, implicit $exec
+    %32:sgpr_32 = S_MOV_B32 6400
+    %33:vgpr_32, %34:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %32, 0, implicit $exec
+    %35:vgpr_32, dead %36:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %34, 0, implicit $exec
+    %37:vreg_64_align2 = REG_SEQUENCE %33, %subreg.sub0, %35, %subreg.sub1
+    %38:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %37, 0, 0, implicit $exec
+    %39:sgpr_32 = S_MOV_B32 11200
+    %40:vgpr_32, %41:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %39, 0, implicit $exec
+    %42:vgpr_32, dead %43:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %41, 0, implicit $exec
+    %44:vreg_64_align2 = REG_SEQUENCE %40, %subreg.sub0, %42, %subreg.sub1
+    %45:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %44, 0, 0, implicit $exec
+...
+---
+
+# GFX9-LABEL: name: NegativeDistance
+# GFX9: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 10240
+# GFX9: [[V_ADD_CO_U32_e64_4:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_5:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 %{{[0-9]+}}, [[S_MOV_B32_1]]
+# GFX9: [[BASE_HI:%[0-9]+]]:vgpr_32, dead %{{[0-9]+}}:sreg_64_xexec = V_ADDC_U32_e64 %{{[0-9]+}}, 0, killed [[V_ADD_CO_U32_e64_5]]
+# GFX9: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[V_ADD_CO_U32_e64_4]], %subreg.sub0, [[BASE_HI]], %subreg.sub1
+# GFX9: [[GLOBAL_LOAD_DWORDX2_:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[REG_SEQUENCE2]], -4096, 0
+# GFX9: [[GLOBAL_LOAD_DWORDX2_1:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[REG_SEQUENCE2]], -2048, 0
+# GFX9: [[GLOBAL_LOAD_DWORDX2_2:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 [[REG_SEQUENCE2]], 0, 0
+
+name: NegativeDistance
+body:             |
+  bb.0.entry:
+    %0:sgpr_64 = COPY $sgpr0_sgpr1
+    %1:sreg_64_xexec = S_LOAD_DWORDX2_IMM %0, 36, 0
+    %3:sgpr_128 = COPY $sgpr96_sgpr97_sgpr98_sgpr99
+    %4:sreg_32_xm0 = COPY $sgpr10
+    %5:sreg_32_xm0 = S_MOV_B32 0
+    $sgpr0_sgpr1_sgpr2_sgpr3 = COPY %3
+    $sgpr4 = COPY %4
+    $vgpr0 = V_MOV_B32_e32 0, implicit $exec
+    %6:vreg_64_align2 = COPY $vgpr0_vgpr1
+    %7:vgpr_32 = V_AND_B32_e32 255, %6.sub0, implicit $exec
+    %8:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+    %9:vreg_64_align2 = REG_SEQUENCE killed %7, %subreg.sub0, %8, %subreg.sub1
+    %10:vgpr_32 = V_LSHLREV_B32_e64 7, %6.sub0, implicit $exec
+    %11:vgpr_32 = V_AND_B32_e32 -32768, killed %10, implicit $exec
+    %12:sgpr_32 = COPY %1.sub1
+    %13:vgpr_32 = COPY %5
+    %14:vgpr_32, %15:sreg_64_xexec = V_ADD_CO_U32_e64 %1.sub0, %11, 0, implicit $exec
+    %16:vgpr_32 = COPY %12
+    %17:vgpr_32, dead %18:sreg_64_xexec = V_ADDC_U32_e64 %16, %13, killed %15, 0, implicit $exec
+    %19:vreg_64_align2 = REG_SEQUENCE %14, %subreg.sub0, %17, %subreg.sub1
+    %20:vreg_64_align2 = V_LSHLREV_B64_e64 3, %9, implicit $exec
+    %21:vgpr_32, %22:sreg_64_xexec = V_ADD_CO_U32_e64 %14, %20.sub0, 0, implicit $exec
+    %23:vgpr_32, dead %24:sreg_64_xexec = V_ADDC_U32_e64 %17, %20.sub1, killed %22, 0, implicit $exec
+    %25:sgpr_32 = S_MOV_B32 6144
+    %26:vgpr_32, %27:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %25, 0, implicit $exec
+    %28:vgpr_32, dead %29:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %27, 0, implicit $exec
+    %30:vreg_64_align2 = REG_SEQUENCE %26, %subreg.sub0, %28, %subreg.sub1
+    %31:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %30, 0, 0, implicit $exec
+    %32:sgpr_32 = S_MOV_B32 8192
+    %33:vgpr_32, %34:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %32, 0, implicit $exec
+    %35:vgpr_32, dead %36:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %34, 0, implicit $exec
+    %37:vreg_64_align2 = REG_SEQUENCE %33, %subreg.sub0, %35, %subreg.sub1
+    %38:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %37, 0, 0, implicit $exec
+    %39:sgpr_32 = S_MOV_B32 10240
+    %40:vgpr_32, %41:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %39, 0, implicit $exec
+    %42:vgpr_32, dead %43:sreg_64_xexec = V_ADDC_U32_e64 %23, 0, killed %41, 0, implicit $exec
+    %44:vreg_64_align2 = REG_SEQUENCE %40, %subreg.sub0, %42, %subreg.sub1
+    %45:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %44, 0, 0, implicit $exec
+...
+---
+
+# Tests for a successful compilation.
+name: assert_hit
+body:             |
+    bb.0.entry:
+    %0:sgpr_64 = COPY $sgpr0_sgpr1
+    %1:sreg_64_xexec = S_LOAD_DWORDX2_IMM %0, 36, 0
+    %3:sgpr_128 = COPY $sgpr96_sgpr97_sgpr98_sgpr99
+    %4:sreg_32_xm0 = COPY $sgpr101
+    %5:sreg_32_xm0 = S_MOV_B32 0
+    $sgpr0_sgpr1_sgpr2_sgpr3 = COPY %3
+    $sgpr4 = COPY %4
+    $vgpr0 = V_MOV_B32_e32 0, implicit $exec
+    %6:vreg_64_align2 = COPY $vgpr0_vgpr1
+    %7:vgpr_32 = V_AND_B32_e32 255, %6.sub0, implicit $exec
+    %8:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+    %9:vreg_64_align2 = REG_SEQUENCE killed %7, %subreg.sub0, %8, %subreg.sub1
+    %10:vgpr_32 = V_LSHLREV_B32_e64 7, %6.sub0, implicit $exec
+    %11:vgpr_32 = V_AND_B32_e32 -32768, killed %10, implicit $exec
+    %12:sgpr_32 = COPY %1.sub1
+    %13:vgpr_32 = COPY %5
+    %14:vgpr_32, %15:sreg_64_xexec = V_ADD_CO_U32_e64 %1.sub0, %11, 0, implicit $exec
+    %16:vgpr_32 = COPY %12
+    %17:vgpr_32, dead %18:sreg_64_xexec = V_ADDC_U32_e64 %16, %13, killed %15, 0, implicit $exec
+    %19:vreg_64_align2 = REG_SEQUENCE %14, %subreg.sub0, %17, %subreg.sub1
+    %20:vreg_64_align2 = V_LSHLREV_B64_e64 3, %9, implicit $exec
+    %21:vgpr_32, %22:sreg_64_xexec = V_ADD_CO_U32_e64 %14, %20.sub0, 0, implicit $exec
+    %23:vgpr_32, dead %24:sreg_64_xexec = V_ADDC_U32_e64 %17, %20.sub1, killed %22, 0, implicit $exec
+
+    %25:sgpr_32 = S_MOV_B32 6144
+    %26:vgpr_32, %27:sreg_64_xexec = V_ADD_CO_U32_e64 %21, %25, 0, implicit $exec
+    %28:vgpr_32, dead %29:sreg_64_xexec = V_ADDC_U32_e64 %23, 4294967295, killed %27, 0, implicit $exec
+    %30:vreg_64_align2 = REG_SEQUENCE %26, %subreg.sub0, %28, %subreg.sub1
+    %31:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %30, 0, 0, implicit $exec
+...
+---
+
+# GFX9-LABEL: name: diffoporder_add_store
+# GFX9: GLOBAL_STORE_DWORD %{{[0-9]+}}, %0.sub0, 1000, 0,
+# GFX9: GLOBAL_STORE_DWORD %{{[0-9]+}}, %0.sub1, 0, 0,
+
+name: diffoporder_add_store
+body:             |
+  bb.0.entry:
+
+    %0:vreg_64_align2 = COPY $vgpr0_vgpr1
+
+    %1:sgpr_32 = S_MOV_B32 4000
+    %2:vgpr_32, %3:sreg_64_xexec = V_ADD_CO_U32_e64 %0.sub0, %1, 0, implicit $exec
+    %4:vgpr_32, dead %5:sreg_64_xexec = V_ADDC_U32_e64 %0.sub1, 0, %3, 0, implicit $exec
+    %6:vreg_64_align2 = REG_SEQUENCE %2, %subreg.sub0, %4, %subreg.sub1
+    GLOBAL_STORE_DWORD %6, %0.sub0, 0, 0, implicit $exec
+
+    %8:sgpr_32 = S_MOV_B32 3000
+    %9:vgpr_32, %10:sreg_64_xexec = V_ADD_CO_U32_e64 %0.sub0, %8, 0, implicit $exec
+    %11:vgpr_32, dead %12:sreg_64_xexec = V_ADDC_U32_e64 %0.sub1, 0, %10, 0, implicit $exec
+    %13:vreg_64_align2 = REG_SEQUENCE %9, %subreg.sub0, %11, %subreg.sub1
+    GLOBAL_STORE_DWORD %13, %0.sub1, 0, 0, implicit $exec
+...

diff --git a/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm.ll b/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm.ll
index 9e5c56d..e2fbc0b 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm.ll
+++ b/llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm.ll
@@ -1,6 +1,7 @@
 ; RUN: llc -mtriple=amdgcn -mcpu=gfx803 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX8 %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX9 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX9,GFX900 %s
 ; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX10 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX9,GFX90A %s
 
 declare i64 @_Z13get_global_idj(i32)
 
@@ -86,17 +87,17 @@ define hidden amdgpu_kernel void @clmem_read(i8 addrspace(1)*  %buffer) {
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ;
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX900:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ;
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
@@ -109,6 +110,19 @@ define hidden amdgpu_kernel void @clmem_read(i8 addrspace(1)*  %buffer) {
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX90A:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+
 entry:
   %call = tail call i64 @_Z13get_global_idj(i32 0)
   %conv = and i64 %call, 255